Learning on Purpose · 09/30/26 · #532
“When it comes to student engagement, there’s no one right time, format, or plan. What matters more is how we position students and how we create the conditions that help them feel valued so that they can speak authentically.”
EdWeek · 08/28/26 · #531
“People hear the term ‘AI tutor,’ and they think that is a young person engaging directly with a chatbot,” he said. “What our brief outlines is that AI tutors exist on an AI-human spectrum, and that spectrum exists on a range of relational intensity. Depending on where you fall on that range, there is either strong evidence to support [it] or very little evidence.”
EdWeek · 08/26/26 · #530
“Most teens say they use AI for schoolwork. In the Britebound survey, 44% said they use AI for school and personal uses and 23% said they use it for school only. The Common Sense Media survey found that 70% of students use AI for schoolwork. The most common ways students use AI is for seeking direct answers to homework problems.”
Augmented Educator · 08/20/26 · #529
“A false-positive rate of 0.19 percent means that out of 1,000 human-written essays, the detector falsely flags only about two texts as AI-generated. Sounds great, doesn’t it? But unfortunately, none of those numbers can really tell you what will happen in your classroom. Because outside of a controlled lab environment, they have very little meaning… A school, or any educational institution for that matter, has to choose which of the two errors it is more willing to accept. But the problem is that these errors do not carry equal consequences. Missing an AI-written essay compromises an assessment. This is most likely not a big deal. Falsely accusing a student, however, can derail a degree and follow that student for years.”
Anthropic · 08/28/26 · #529
“Starting today, Claude for Teachers is available for U.S. K-12 schools and districts as a free Enterprise offering. This includes premium Claude capabilities like teaching skills grounded in learning science and a direct connection to academic standards in all 50 states. Qualifying organizations that sign up by June 30, 2027 get a full year of free access… Anthropic provides school and district-level K-12 Terms and Data Processing Agreement, so one set of FERPA-aligned commitments applies across your entire organization. Claude for Teachers data is not used for model training purposes, and student information is protected by our K-12 Data Processing Agreement.”
Common Sense Media · 08/18/26 · #529
“Few teens are being taught AI literacy. Only 27% of teens say a teacher has ever talked with them about how to use AI safely; 73% say they have not. Bans don't seem to be closing AI literacy gaps. Among teens who use AI for schoolwork, 44% have had an AI tool blocked on a school network or device, and 59% of them just switched to a personal device to get around the ban. Teens see the cost themselves. Among those who use AI, 38% say having it leads them to generate fewer of their own ideas, and 39% say they feel like they're missing out on learning when they use AI to complete assignments.”
Dr. Philippa Hardman · 08/19/26 · #528
“This changed how I think about the human–AI boundary. We tend to talk about keeping the “human” parts of a process and delegating the mechanical ones — but that assumes we already know which is which. I didn’t. Writing feedback that feels personal turned out to be mechanisable. Working out what a particular learner needs did not. So perhaps the more useful distinction when thinking about what humans do and what AI does isn't personal versus impersonal, or even simple versus complex, but specifiable versus situational — the work you can settle in advance, and the work you can only settle with a particular insight in front of you which is yet to be produced.”
New York Times · 08/22/26 · #528
“It was also helpful, Sultania said, that the A.I. avatars were always available when he had time, which as a busy founder and Ph.D. student, is not always predictable. “I kind of want the platform there that I can talk to, even if it’s a clone, anytime I want,” he said. “You can really hash out things before you actually enter a situation like that,” she said. “It’s a good way to practice.””
Inside Higher Ed · 08/05/26 · #527
““We cannot be giving unsupervised assessments to students expecting them to resist AI. They’re not going to be learning, we’re going to be spending our time trying to catch them cheating and they’re going to graduate and realize they just wasted four years,” she said. Instead, she advises departments to focus on teaching the foundational skills required to use AI responsibly and ethically. “We need to give students a distraction-free, temptation-free environment to tell us what they know and what they can do.””
Dr Philippa Hardman · 08/07/26 · #526
“For two years, the cognitive offloading debate has been quite binary and vague on specifics: we need to ban AI from the learning experience because it erodes cognition, or let’s embrace AI because it helps us get smarter. A number of studies published in 2026 have helped to move the debate beyond the binary and replace it with a more nuanced (but still incomplete) understanding of conditions under which offloading harms learning, the conditions under which AI improves cognition, and how to design for the difference.”
Education Disrupted · 07/28/26 · #526
“Recording myself teach. This one is newer and I’m leaning on it more… I record with Voice Memos whenever I’m doing anything lecture-shaped… Voice Memos will give you a transcript if you tap it. I email it to myself, drop it into Claude with whatever I taught from that day, and say: update the materials based on what I added as I went. That’s the entire prompt.”
NPR · 07/30/26 · #525
“When students arrived at the start of the weekend, they were presented with a stew of policies that they had suggested in the run-up to the event. Then they were divided into subcommittees — focused on students, educators, administrators and families — to debate the dozens of proposals and create a coherent bill they could all support. The student-focused section of the final bill includes 15 provisions, several focused on a concern many of the senators shared: that teens are using AI to do their thinking for them… Among the bill's requirements: States should provide AI literacy as soon as students get access to classroom devices; AI use should be banned on all graded tests; and teachers can require an oral defense of any project from a student suspected of using AI inappropriately. (You can read the full bill at the end of this article.)"
UChicago · 06/02/26 · #524
“The University has a duty of care to ensure that the education offered to you is responsive to these technological developments by teaching you how to think with machines, how to think without them, and how to think about them… As a next step in the University of Chicago’s overall approach to artificial intelligence (AI), the University is partnering with Anthropic to provide Claude Enterprise for all academics and staff starting in July, and for all students before the fall term. Claude Chat, Cowork, and Code will be made available to the campus community. Spanning the domains of research, education, and operations: these are powerful tools that are being made available to help individuals find ways to improve our pursuit of knowledge, our commitment to teaching students how to think independently, and the ability of our staff to support this work efficiently. Access will be granted to the newest versions of Anthropic’s tools—for those who choose to use them—with improved integration with university resources and tools and more data protection than is available with private accounts.”
Science Direct · 05/17/26 · #523
“Analysis revealed that, ChatGPT supports cognitive development when embedded within inquiry-oriented and scaffolded instructional designs, especially through processes such as metacognitive regulation, argumentative reasoning, and idea generation, which in turn produce synergistic gains across both critical and creative domains. However, in unstructured contexts, researchers observed asymmetrical patterns favoring creativity over critical thinking and joint declines across domains, both of which often triggered cognitive offloading. These findings highlight that ChatGPT's cognitive effects are contingent on pedagogical framing.”
Leon Furze · 04/22/26 · #521
“When AI is introduced before sufficient expertise has developed, the risk is not just poor output in the moment: It’s that the expertise never develops at all, because the productive struggle required to build it was bypassed. The student plateaus at the acclimation stage, never building the schemas, the mental representations, or the situated understanding that would allow them to move forward. And they don’t know what they’re missing, because the skills needed to produce correct responses are the same skills needed to recognise correct responses. Less practice leads to less competence, which leads to worse ability to detect errors in AI output, which leads to more reliance on AI, which leads to even less practice.”
Learning on Purpose · 04/26/26 · #521
“When it comes to AI, students believe their schools see them as cheaters. The AI policies they are asked to follow spark fear and anxiety about punishment. Writing is done in class, under supervision. They are asked to install tools like browser lockdown apps on their devices or to use internet-free computers rather than their own. They are presented with percentages from AI detectors as evidence that they have cheated… If we want to address AI and academic integrity, I think we need to design learning environments built on transparency and accountability that prioritize giving students a sense of status and respect.”
Dr. Philippa Hardman · 04/16/26 · #520
“The researchers call this the offloading paradox. Offloading can lead to better quality thinking and deeper learning — but only under two conditions: 1) Learners delegate enough to AI to genuinely free cognitive capacity, and 2) That freed capacity gets deliberately invested in higher-order work — questioning assumptions, critiquing frameworks, constructing original arguments (the kind of thinking AI can’t do for you).”
Alex Kotran · 04/09/26 · #519
“6. Once someone builds something real, it changes them… You go from “building things is not for people like me, that’s for engineers” to “I wonder if I could build that” to “I think I could build that” to “I know I can build that.” And as you progress, you’re asking those same questions in parallel, but for increasingly ambitious things. And once it does, the conversation about AI changes permanently. People can’t unsee it. Increasingly, those people are going back to their schools and organizations and bringing colleagues in. A teacher builds a tool, shows it to another teacher in the hallway, and suddenly that teacher wants to try. That’s how cultures of innovation take root.”
Education Disrupted · 03/19/26 · #518
“I typed a simple prompt: “I’d like to build an app that fact-checks articles on the web.” And Claude built it.”
Dr Philippa Hardman · 02/19/26 · #517
“AI-only feedback and support seems most effective when tasks are low-stakes, tightly structured, and not identity-relevant: grammar correction, coding syntax, multiple-choice practice, drill exercises. In these contexts, AI is functional, welcome, and effective. The Karimova and Csapó (2024) meta-review found its strongest gains in exactly these structured, skill-based domains. Students consistently draw the line themselves. In Alkhalaf, Alkhayat and Alzahrani (2026) and across domain studies, they say: AI for drills and drafts, humans for capstones, professional work, and emotionally difficult tasks. The boundary hinges on identity relevance and emotional stakes. When the work is “just practice,” AI is a useful tool.”
Teaching in the Age of AI · 03/16/26 · #517
“When I first started using AI seriously, I was doing what most teachers do - asking questions, iterating on responses, but primarily generating text - converting ideas into polished plans and getting feedback and analysis on my own work. AI remains useful for all of those tasks. But agentic AI introduces something fundamentally different. I can now craft an interactive learning experience - deciding what it should do, how it should be organized, who would use it, and what problem it needs to solve. That’s a much different - and more involved - cognitive activity.”
Anthropic · 02/23/26 · #516
“Iteration and refinement is the single strongest correlate of all other fluency behaviors in our data. So, when you get an initial response, it’s worth treating it as only a starting point: ask follow-up questions, push back on any parts that don’t feel right, and refine what you’re looking for.”
New York Times · 02/09/26 · #514
“The patient says, “Yesterday I woke up dizzy. My arm was dead, and I had trouble speaking.” What does “dizzy” actually mean? It could mean the patient is lightheaded and about to faint. Or it could mean that the room is spinning. A “dead” arm might be numb rather than weak. Someone with an arm that is partially paralyzed may say it feels numb. But a patient might feel a pin if Dr. Schwamm pricks their arm. Is the patient having a stroke? Is this a medical emergency? Dr. Schwamm has years of training that helps him figure out who is sick and who isn’t, he says, and who to not be worried about and who should be admitted to the hospital.”
Beyond the AI Inflection Point · 01/20/26 · #512
“School and district leaders face an impossible question: How do we prepare students for a world where AI is redefining learning and work? The decisions being made right now—about technology adoption, AI literacy, assessment practices, and the fundamental purpose of school—will lead to vastly different outcomes for students… The resulting report deliberately offers provocations, not predictions, and the accompanying toolkit helps districts and schools ask big questions and chart their own paths forward. Both the report and toolkit invite education leaders to grapple with authentic challenges—budget constraints, competing priorities, community needs, and the tension between innovation and tradition—that have no easy answers.”
A Computer Scientist In A Business School · 12/29/25 · #511
“We surveyed students before releasing grades to capture their experience. Some of the results: Only 13% preferred the AI oral format. 57% wanted traditional written exams. 83% found it more stressful. But here's the thing: 70% agreed it tested their actual understanding: the highest-rated item. They accepted the assessment but not the delivery. At the same time, they almost universally liked the flexibility of taking the exam at their own place and time… The fix is clear: one question at a time, slower pacing, calmer tone. The concept works. The execution needs iteration.”
Brookings · 01/14/26 · #511
“To this end, we offer three pillars for action: Prosper, Prepare, and Protect. Under each pillar, we present actionable recommendations for governments, technology companies, education system leaders, families, and all those who touch this issue. We urge all relevant actors to identify at least one recommendation to advance over the next three years.”
LinkedIn · 01/17/26 · #511
“In the 2009 Star Trek reboot, a young Spock is shown in a Vulcan school. It is a dark, cavernous room with learning pods like giant bowls carved into the floor, a pod for each student, each pod made of a ring of screens, each screen brimming with information. Adults pace the floor above. AI voices in the pods pepper students with questions and blanket the screens with dazzling graphs and equations, and the students answer, demonstrating their mastery. It is a vision of AI tutors engaging in personalized conversations with young students, and an elementary school-aged Spock answers a final question we don’t fully hear with apparent profundity: “...when it is morally praiseworthy, but not morally obligatory” (Abrams, 2009). “This is the vision!” said an edtech investor to one of the authors. No.”
QuestionWell · 12/28/25 · #510
“Writing good multiple-choice questions (MCQs) is well studied and well known to be difficult, even for human beings. Decades of research document common flaws: inadvertently giving students cues, implausible distractors, ambiguity, and questions that reward test taking strategies, and allow students to side-step understanding. Large language models (LLMs) make many of the same errors as human question writers. At QuestionWell, we push our AI to do better. This fall, we trained our AI model specifically to reduce well-documented item writing flaws in multiple choice questions.”
Alex Kotran · 12/09/25 · #509
“Here’s what I want you to take away: uncertainty isn’t the same as helplessness. You don’t need to predict the future to prepare for it. The people who thrive in moments like this aren’t the ones who guessed right about which industries would boom—they’re the ones who built the capacity to adapt, to learn, to push through hard things, and to work with others.”
Teaching in the Age of AI · 12/22/25 · #509
“I suspect many high school students simply ask the first question when determining the calculus of AI use: Will I get caught? It’s heartening to know others are weighing the more critical question: Will I learn? Both are operating rationally within a decision making framework, but only one produces the outcome educators want. It’s clear that neither approach has much to do with school policy. The first student has already concluded AI policies are rarely an effective deterrent - the odds of getting caught are so low it’s worth the risk. The second has internalized reasons that have nothing to do with the actual rules. What both positions lack is any semblance of adult guidance. Students are primarily making these choices on their own.”
Dr Philippa Hardman · 11/20/25 · #508
“Most AI tutors fail because they: Hallucinate content… Have un-curated knowledge bases.. Are answer machines… Don’t manage cognitive load… Built using FRAME, this prompt manages all four of these risks: No hallucinations… Curated content… Tutoring expertise… No cognitive overload”
Atlantic · 10/26/25 · #506
“The hard part is deciding, without nostalgia and inertia, which skills are keepers and which are castoffs. None of us likes to see hard-won abilities discarded as obsolete, which is why we have to resist the tug of sentimentality. Every advance has cost something. Literacy dulled feats of memory but created new powers of analysis. Calculators did a number on mental arithmetic; they also enabled more people to “do the math.” Recorded sound weakened everyday musical competence but changed how we listen. And today? Surely we have some say in whether LLMs expand our minds or shrink them.”
Dr Philippa Hardman · 10/30/25 · #506
“The data suggests that most practitioners are stumbling across the jagged frontier daily, without understanding the associated risks. Here’s an up to date summary of what we know about the “jagged frontier” of AI use in Instructional Design, shared in the hope that it can help you pull on AI’s benefits while mitigating its risks.”
Harvard · 10/14/25 · #504
“Critically, Tan does not grade her students’ AI-generated content. Instead, she assesses their reflections and their ability to use AI meaningfully. Over time, students get better at writing effective prompts and critiquing the AI’s output. They learn to identify hallucinations, jargon that masks weak logic, and content that sounds plausible but lacks substance. Tan’s broader takeaway from the experiment is that AI won’t make teaching obsolete. It may, however, render disengaged or uncritical teaching practices obsolete.”
No More Marking · 10/19/25 · #504
“So far, our teachers have told us that the report they find the most useful is the teacher report, consisting of personalised information on every student designed for teachers. There are three elements in the report: data, AI feedback and the student writing. They prefer this to the student report, which is similar but doesn’t have data and has simplified AI feedback. The most-requested feature from teachers has been an AI-generated year group summary of all this personalised feedback. We’ve developed that, and you can see an example of what it looks like here. It’s kind of like an examiner’s report, but just for your students.”
Educating AI · 10/02/25 · #503
“For decades, districts have used analytical tools. When you ran a report in your Student Information System or built a pivot table in Excel, the process was straightforward: You decided what to measure (attendance below 80%, GPA drop over 0.5 points), The system counted and calculated based on your rules, You got results you could verify (show me the 23 students who meet these criteria)… Here’s what’s different when an AI “analyzes” your data: It doesn’t find patterns in your data. It generates text about your data based on patterns from its training. Let me unpack that:”
ChatGPT · 09/15/25 · #500
“More than 70 college students across North America created and ranked these chats in our ChatGPT Lab. Some chats were also provided by content creators on social media.”
CBS News · 09/17/25 · #500
“The goal of the initiative, AI Fluency, is that students, beginning with the class of 2029, will graduate and be fluent in both their major and AI... Beginning this year, all freshmen are required to take a course in generative AI and multiple workshops aimed at real-world applications to help them master the technology.”
What School Could Be · 08/18/25 · #498
“We need teachers to actually teach us how to prompt. Show us good examples, and let us try. Don’t punish the whole class because one kid went overboard—most of us want to use it carefully.”
Literary Hub · 07/28/25 · #497
“I attempted the experiment in four sections of my class during the 2024-2025 academic year, with a total of 72 student writers. Rather than taking an “abstinence-only” approach to AI, I decided to put the central, existential question to them directly: was it still necessary or valuable to learn to write? The choice would be theirs. We would look at the evidence, and at the end of the semester, they would decide by vote whether A.I. could replace me. What could go wrong?”
Anthropic · 08/26/25 · #497
“Some educators are automating grading; others are deeply opposed — In our Claude.ai data, faculty used AI for grading and evaluation less frequently than other uses, but when they did, 48.9% of the time they used it in an automation-heavy way (where the AI directly performs the task). That’s despite educator concerns about automating assessment tasks, as well as our surveyed faculty rating it as the area where they felt AI was least effective.”
New York Times · 08/12/25 · #496
“Perhaps one of the biggest threats that A.I. poses to education isn’t that it’s going to make educators useless, but that it is going to make educators so much more necessary than we are willing to invest in. A.I. actually makes it more important that we have everything from librarians to counselors to teachers to professors to researchers who can put this rapidly changing information environment into context and can develop the capacity in students to make sense of things.”
New York Times · 08/06/25 · #495
“Through a combination of oral examinations, one-on-one discussions, community engagement and in-class projects, the professors I spoke with are revitalizing the experience of humanities for 21st-century students… This was another commonality among professors I spoke to: They realized that the only way forward in the age of A.I. was to have trust and transparency around the technology’s use in their classrooms, and often this involved creating a policy with their students’ input.”
Cal Matters · 07/16/25 · #494
“Students don’t have the same incentives to talk to their professors — or even their classmates — anymore. Chatbots like ChatGPT, Gemini and Claude have given them a new path to self-sufficiency. Instead of asking a professor for help on a paper topic, students can go to a chatbot. Instead of forming a study group, students can ask AI for help. These chatbots give them quick responses, on their own timeline.”
Guardian · 07/27/25 · #494
“The students who have given me unrestricted access to the ChatGPT Plus account they share, and permission to quote from it, are all second-year undergraduates at a top British university. Rohan studies politics and is the named account administrator. Joshua is studying history. And Nathaniel, the heaviest user of the account, consulted ChatGPT extensively before changing courses from maths to computer sciences. They’re by no means a representative sample (they’re all male, for one), but they liked the idea of letting me understand this developing and complex relationship.”
The Rithm Project · 07/31/25 · #494
“We hope it’s useful in two ways: For consumers: to help young people and the adults that support them recognize which features empower connection, so they know when to proceed with caution or confidence. For AI builders: to surface the small (and big) design choices that shape how people relate, not just to the product, but to each other. This first draft was co-created with an incredible crew of Gen Z youth, technologists, researchers, and mental health professionals. And we know it’s not final. It’s meant to be tested, lived with, and sharpened through use.”
Preply · 06/20/25 · #493
“To better understand how AI is being adopted by students, we set out to learn where young minds across the U.S. are utilizing this technology the most. From 18-year-old high schoolers to college students, and parents of younger teens, we looked at how often students rely on AI tools, for what, and what tools they use the most.”
PRNewsWire · 07/23/25 · #493
“The first tool Instructure is launching is a new type of assignment called the LLM-Enabled Assignment, designed to let educators create a custom GPT-like experience within Canvas. Teachers can define how AI interacts with students, set specific learning goals and objectives and determine what evidence of learning it should track. They can do this using natural language prompts or by leveraging an assistant within the assignment creation flow to guide them through the process.”
The Augmented Educator · 07/13/25 · #492
“I have started to understand authorship as something that develops organically: a self-organizing, dynamic process involving a complex interplay between human creativity, technological capabilities, and the broader intellectual context in which we work. The question shifts from “Did you write this?” to “How did you write this?” Academic integrity becomes less about proving the absence of AI and more about demonstrating the presence of human critical thinking, intellectual contribution, and—crucially—transparency about one’s process.”
Walton Family Foundation · 06/25/25 · #491
“The most common uses of AI among those using it at least monthly include: Preparing lessons (37%), Creating worksheets (33%), Modifying materials to meet students’ needs (28%), Doing administrative work (28%). Making assessments (25%)”
Mike Caulfield · 06/16/25 · #490
“Here it is — I just released a (largely) non-hallucinating1 rigorous AI-based fact-checker that anyone can use for free. And I don’t say that lightly: I literally co-wrote the book on using the internet to verify things. All you do is log into ChatGPT, click the link below, and put in a sentence or paragraph for it to fact check.”
Dr. Philippa Hardman · 06/05/25 · #489
“The patterns in the data about the questions asked of AI by learners are striking. The most notable thing off-the-bat is that students are not using AI to "cheat"—they're using it to implement powerful instructional strategies to help them learn… The data we are gathering about how our learners are using AI is uncomfortable but essential for our growth as a profession. Learner AI usage is essentially a real-time audit of our design decisions—and the results should concern every instructional designer.”
Dr. Philippa Hardman · 06/12/25 · #489
“Among the 30+ AI use cases we try and test in the bootcamp, one has emerged as particularly popular and potentially significant: using AI to simulate real learners' behaviour and feedback. Imagine having access to your target learners' thoughts, behaviours and emotional responses throughout your entire design process. While this may seem like science fiction (and, perhaps, risky), recent research by Park et al. (2024) demonstrates that AI personas built from intentionally gathered, structured data can predict learner behaviour more accurately than traditional human-centred approaches, achieving approximately 85% accuracy in behavioural simulation and prediction.”
YouTube/Google · 05/19/25 · #488
“It seems very strange to me to hear you speak Spanish. This is a live, real-time translation that we are both experiencing.”
World Economic Forum · 05/22/25 · #488
“AI's most significant influence lies in how we access, process and apply information, fundamentally redefining education and the way we acquire knowledge. From this perspective, AI literacy isn’t just a “nice to have” for IT professionals; it’s essential for developing human intelligence itself and has key implications for the education sector.”
Cult of Pedagogy · 05/11/25 · #486
“Susan Morrow and Katherine Switzer revealed a different approach to accountability: the power of relationships, transparency, and explainability to ensure the integrity of results. The following steps can help you, and your students, take action to ensure academic integrity.”
Sweet GrAIpes · 05/08/25 · #485
“If students (and educators themselves!) are to thrive in a future populated by AI agents, the ability to "prompt" an AI for an immediate response is only the beginning. The more profound skills will revolve around effectively and responsibly managing these autonomous systems. It's about learning to direct AI at a higher level, much like a good manager guides a team or an individual.”
Leon Furze · 04/28/25 · #484
“What I’m suggesting here is not necessarily resisting the technology. We’re almost at a point where that would be akin to resisting the internet or electricity in the classroom… But the way we use technology, and the form the technology takes in the future, is not set in stone. The line between ubiquity and inevitability may seem thin, but it represents the difference between a technology that is simply widespread and one that has been accepted as necessary, unchangeable, and permanent.”
Mike Caulfield · 04/17/25 · #482
“I mention this because it relates to the tools I build for fact-checking. For almost a decade I have believed in supporting what I call the antibodies of discourse — those rare individuals who are willing to call out lies, provide fuller context, and work to keep the discourse environment free and clear of fake or misrepresented evidence. Some of those people are professionals — reporters, fact-checkers, honest advocates at community-focused non-profits… You all know Brandolini's law, right, the fundamental issue that makes misinformation such problem? It’s usually phrased like this: The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it. My proposal is simple: if that’s the case, then let’s reduce the amount of energy required to refute bullshit. And my assertion is that we have in AI a tool to do just that.”
No More Marking · 03/31/25 · #480
“We now have results. Our headline finding is that AI is very good at judging student writing and is a viable and time-saving alternative for many forms of school assessment. Here are the details… We have been running similar tasks since 2017 for students at primary and secondary, and have assessed nearly 3 million pieces of writing using human Comparative Judgement. The process typically delivers very high levels of inter-rater reliability, and is the gold standard of human judgement. The crucial difference with this assessment is that we did not just have human teachers making the judgements. We got an AI to make judgements, too… In total, our human judges made 3,640 decisions. Of these 3,640 decisions our AI agreed with 81% of them. On our most recent previous human-judged Year 7 assessment the human judges agreed with each other 87% of the time, which is fairly typical.”
EdWeek · 02/14/25 · #478
“As an Advanced Placement English Literature and Composition teacher, Garbarg has to read dozens of student essays. Providing targeted feedback for every student can be time-consuming. So she started experimenting with using AI tools to expedite the work of providing personalized feedback for students’ writing. Now, Garbarg reads all her students’ essays to get an overall sense of their “glows and grows.” Then, she puts students’ writing into a generative AI tool… and instructs it to write feedback based on a set of criteria and her thoughts.”
New York Times · 02/20/25 · #477
“John Giorgi uses artificial intelligence to make artificial intelligence. The 29-year-old computer scientist creates software for a health care start-up that records and summarizes patient visits for doctors, freeing them from hours spent typing up clinical notes. To do so, Mr. Giorgi has his own timesaving helper: an A.I. coding assistant. He taps a few keys and the software tool suggests the rest of the line of code. It can also recommend changes, fetch data, identify bugs and run basic tests. Even though the A.I. makes some mistakes, it saves him up to an hour many days. “I can’t imagine working without it now,” Mr. Giorgi said.”
OpenAI · 02/15/25 · #477
“College-aged students aged 18-24 primarily use AI tools for starting papers and projects, summarizing texts, exploring topics, and brainstorming creative ideas, according to our recent survey. For ChatGPT users who belong to this age group, just over a quarter of all messages are education- or learning-related (which means those messages are tagged as tutoring- or teaching-related by our internal classifiers), followed by writing assistance, one-off questions, computer programming support, and how-to advice.”
ASCD · 02/01/25 · #476
“[I] developed an eight-step process that educators can use to unpack a learning standard with ChatGPT (or a similar chatbot). This process supports a number of objectives: Reflects a purpose-driven approach by including clear assessment criteria and rubrics. Combines effective educational practices with new technology. Affirms the important role of the educator. Works with a variety of tools, such as ChatGPT, Copilot, Gemini, or any other large language model. Saves time and effort. Enables teachers to learn about GenAI along with their students.”
ASCD · 02/01/25 · #475
“Here are just a few ways teachers can help students use AI tools that balance their need for autonomy and competence and allow them to actively engage in the learning process.”
New York Times · 02/02/25 · #475
“The best thing for medicine to do is to find a role for it that doctors can trust. The solution, we believe, is a deliberate division of labor. Instead of forcing both human doctors and A.I. to review every case side by side and trying to turn A.I. into a kind of shadow physician, a more effective approach is to let A.I. operate independently on suitable tasks so that physicians can focus their expertise where it matters most.”
Center for Curriculum Redesign · 01/01/25 · #474
“The goal of this paper is to: 1. Determine which types of cognitive processes and procedures (aka “modes of thinking”) are used in human reasoning. 2. Determine which forms of human reasoning can be mimicked/reproduced by Generative AI–specifically Large Language Models (LLMs). Hereinafter, it will be referred to as “GenAI” unless otherwise indicated (in the case of Symbolic AI, for instance).”
Leon Furze · 01/27/25 · #474
“Many of these resistances all have something in common: they are not a resistance to the technology, they are a resistance to power. These resistors do not object to AI per se, they reject the imposition of technological, corporate control into education that is represented by GenAI… This is not an uninformed, “Luddite” response to technology. It mirrors what I suspect is the experience of many classroom teachers, who have been frustrated over the years with the frequent and sometimes aggressive incursion of tech into the classroom… We have to contend with GenAI, but that does not mean we have to do it blindly, or with a passive, broken acceptance that the technology companies have somehow “won”.”
AI for Education · 01/28/25 · #474
“Recent research finds those with lower AI literacy are MORE likely to embrace AI technology, not less. The reason? They're more likely to view AI as "magical" and experience a sense of awe when AI performs tasks traditionally thought to require human attributes.”
Dr. Philippa Hardman · 01/24/25 · #473
“Current research overwhelmingly suggests that generic Gen AI tools do not just fail to advance human learning—they often actively hinder it. Across all five of the most recent studies on the topic, while tools like ChatGPT, Claude, and Gemini improve immediate task performance, they also reduce cognitive engagement, critical thinking, and self-regulated learning (SRL). However, the potential of AI to transform education remains huge if we shift toward structured and pedagogically optimised systems. To unlock AI’s transformative potential, we must prioritise learning processes over efficiency and outputs. This requires rethinking AI tools through a pedagogy-first lens, with a focus on fostering deeper learning and critical thinking. “
AI EduPathways · 01/05/25 · #471
“One promising solution is evaluating student interactions with AI. This concept has faced some skepticism due to workload and feasibility concerns. Transcripts can, after all, be quite long. But those concerns represent a misconception regarding the approach itself.”
AI & Education · 01/09/25 · #471
“It’s evident that AI is not just advancing technology - it’s prompting fundamental questions about society and humanity. The questions we face aren't just technical but deeply human: how do we harness AI's potential while preserving what makes us uniquely human? How do we ensure this technology creates equal access to the future for ALL?”
AI for Education · 11/26/24 · #469
“At the elementary level, GenAI tools should be used primarily by teachers for lesson planning, preparation, and selective modeling… Until there is compelling evidence of positive impacts on learning through direct student interaction with GenAI tools at the elementary level, as well as adequate safeguards in place to completely eliminate the risk of exposure to inappropriate content, elementary teachers are advised to restrict GenAI use to their own workflows.”
Open Praxis · 11/29/24 · #469
“At this point in history, it is difficult to argue whether GenAI will ultimately be a disruptive or sustaining technology, a catalyst or blocker, or something else that we cannot foresee. However, its public emergence at the end of 2022 undeniably sparked substantial speculation, hype, and even hope. In such uncertain and speculative times, it is crucial to adopt a collective stance to effectively navigate the future; a goal the authors seek to achieve with this collective work. With this manifesto, we resist the uncritical acceptance of GenAI and instead seek to establish a balanced perspective by critically analyzing both its challenges and affordances.”
OpenAI · 11/15/24 · #468
“Used thoughtfully, ChatGPT can be a powerful tool to help students develop skills of rigorous thinking and clear writing, assisting them in thinking through ideas, mastering complex concepts, and getting feedback on drafts. There are also ways to use ChatGPT that are counterproductive to learning—like generating an essay instead of writing it oneself, which deprives students of the opportunity to practice, improve their skills, and grapple with the material. For students committed to becoming better writers and thinkers, here are some ways to use ChatGPT to engage more deeply with the learning process.”
Dr. Philippa Hardman · 11/21/24 · #468
“What's particularly exciting for us instructional designers is that when GPTs are custom-trained for specific instructional design tasks (like writing learning objectives), they show significant improvements in reliability and accuracy…”
Chronicle of Higher Ed · 10/02/24 · #467
“I first ask students to “highlight how you used human and machine skills in your learning” in five potential categories, and offer them a range of options to characterize whether and how they used AI tools to do the work… Among the questions I list: What tricky situations arose when you used AI? How did you chart a path through them? Did bouncing ideas off AI spark your creativity? Were there any new exciting directions it led you toward, or did you wind up preferring your own insights independent of using AI? Which of your skills got a real workout from using AI? How do you feel you’ve improved?”
Dr Philippa Hardman · 11/08/24 · #467
“Spoiler: my findings underscore that until we have specialised, fine-tuned AI copilots for instructional design, we should be cautious about relying on general-purpose models and ensure expert oversight in all ID tasks.”
ABC News · 10/15/24 · #465
“The filing accuses the school of a lack of clarity in its rules about AI use, saying the school handbook did not "have any established rules, policies or procedures for not only the use of artificial intelligence, but what any administrators, faculty or students should do when encountering its use.”"
UEN · 10/21/24 · #465
“One way is making a cake from scratch, buying all the ingredients and making it yourself. Another way is to get a readymade cake mix (like Betty Crocker) and build the cake. Another way is to buy a good quality cake readymade from a bakery. The last way is to get it from the supermarket, a low-quality version like a Twinkie… What is the difference in quality between what you can bake from scratch and buy from a bakery? When might you do one and not the other?”
The 74 Million · 10/07/24 · #464
“The new study offers a middle ground in what’s become a polarized debate between supporters and detractors of AI tutoring. It’s also the first randomized controlled trial — the gold standard in research — to examine a human-AI system in live tutoring. In all, about 1,000 students got help from about 900 tutors, and students who worked with AI-assisted tutors were four percentage points more likely to master the topic after a given session than those in a control group whose tutors didn’t work with AI. Students working with lower-rated tutors saw their performance jump more than twice as much, by nine percentage points. In all, their pass rate went from 56% to 65%, nearly matching the 66% pass rate for students with higher-rated tutors.”
Dr Philippa Hardman · 09/29/24 · #463
“Imagine the possibilities if we could use AI to simulate learner responses during the instructional design process. What if you didn’t need to conduct endless interviews, waiting for learner feedback that sometimes never arrives? What if you could have instant conversations with virtual learners and rapidly generate reliable data to inform your design?”
Marc Watkins · 09/22/24 · #462
“Everyone started talking about Google’s NotebookLM because of their recent podcast update, but the real story is about how Google is using its AI experiments to push our understanding about how people will use generative AI as tools for thought.”
Oregon State University · 09/22/24 · #462
“Use this table as a reference for evaluating and considering changes to aligned course activities (or, where possible, learning outcomes) that emphasize distinctive human skills and/or integrate generative AI (GenAI) tools as a supplement to the learning process.”
AI for Education · 09/19/24 · #461
“This webinar explored innovative approaches to assessment in the era of generative AI, featuring educator Mike Kentz's experiences and insights from a year of experimenting with grading AI chats in his ninth-grade English classroom.”
Hechinger Report · 09/02/24 · #460
“High school students who had access to ChatGPT while doing practice math problems did worse on a math test compared with students who didn’t have access to ChatGPT. Those with ChatGPT solved 48 percent more of the practice problems correctly, but they ultimately scored 17 percent worse on a test of the topic that the students were learning. A third group of students had access to a revised version of ChatGPT that functioned more like a tutor. This chatbot was programmed to provide hints without directly divulging the answer. The students who used it did spectacularly better on the practice problems, solving 127 percent more of them correctly compared with students who did their practice work without any high-tech aids. But on a test afterwards, these AI-tutored students did no better. Students who just did their practice problems the old fashioned way — on their own — matched their test scores.”
SSRN · 07/18/24 · #458
“Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48% improvement for GPT Base and 127% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction for GPT Base). That is, access to GPT-4 can harm educational outcomes. These negative learning effects are largely mitigated by the safeguards included in GPT Tutor. Our results suggest that students attempt to use GPT-4 as a "crutch" during practice problem sessions, and when successful, perform worse on their own.”
Digital Education Council · 08/07/24 · #457
“86% of students globally are regularly using AI in their studies, with 54% of them using AI on a weekly basis… Despite their high rates of AI usage, 1 in 2 students do not feel AI ready.”
TeachAI · 07/16/24 · #455
“At a time when the entire education community is grappling with how to realize the benefits of AI while mitigating the risks, the briefs that compose Guidance on the Future of Computer Science Education in an Age of AI serve as the beginning of a discussion rather than definitive answers. The briefs offer preliminary insights for responsibly and effectively integrating AI into primary and secondary CS education, address common misconceptions, and provide a balanced perspective on critical issues.”
Edutopia · 07/10/24 · #454
“To aid teachers and students in responsible Al use, we have collaborated on a decision tree for student decision-making. This framework can help students make better choices about using AI tools and give teachers more confidence in supporting student use.”
Marc Watkins · 07/12/24 · #454
“The conversations we need to have about generative AI include our students, administration, and our colleagues. I won’t sugarcoat any of this—these conversations will be a continuum lasting years. They will be frustrating because they may not appear to do anything at first, but that’s because human beings aren’t like machines—we’re slow to change and that’s a good thing. Each interaction we have with one another shapes our thinking about genAI’s impact on teaching and learning. We should not dismiss the small conversations as meaningless. What we should do is recognize the amount of time and energy it takes to simply talk about AI.”
EdWeek · 06/20/24 · #453
“As AI transforms K-12 education—providing everything from lesson planning assistance for overworked teachers to chatbot tutors for students— educators must be aware of how societal biases reflected in the data that underpins AI can shape its responses, experts say.”
Learning on Purpose · 06/27/24 · #452
“the question I’m asking myself right now is, How can what I’ve learned shape what I will do in the coming year? I came up with four priorities that will drive my work. I even made a little card that I’ve taped to my desk: Augmentation over automation, Literacy over policy, Design over technology, Vision over decision”
Lance Eaton · 06/20/24 · #451
“In this one, I’m going to provide some guidance and insights about how faculty might engage with students once they suspect that a student might have inappropriately used generative AI.”
Maha Bali · 06/22/24 · #451
“Human creativity augmented with AI would only be beneficial, IMHO, if the weight of human judgment before and after AI use is high. In the same way it’s important to make a mental calculation of something before you put it into the calculator so that you know roughly what should come out, I think with AI we need to constantly do this, not just to preserve our imagination, but to also preserve our criticality and awareness of epistemic bias.”
AI and Education · 06/03/24 · #450
“This piece primarily breaks down how Google’s LearnLM was built, and takes a quick look at Microsoft/Khan Academy’s Phi-3 and OpenAI’s ChatGPT Edu as alternative approaches to building an “education model” (not necessarily a new model in the latter case, but we’ll explain). Thanks to the public release of their 86-page research paper, we have the most comprehensive view into LearnLM.”
Hechinger Report · 06/03/24 · #450
“A team of researchers compared AI with human feedback on 200 history essays written by students in grades 6 through 12 and they determined that human feedback was generally a bit better. Humans had a particular advantage in advising students on something to work on that would be appropriate for where they are in their development as a writer. But ChatGPT came close… Most of these humans had taught writing for more than 15 years or they had considerable experience in writing instruction. All received three hours of training for this exercise plus extra pay for providing the feedback.”
Michigan Virtual · 05/31/24 · #449
“As schools end the academic year and look ahead to the 24-25 school year, many are likely still trying to determine the best strategies for realizing the potential benefits and overcoming the challenges related to artificial intelligence (AI) technologies. To do so, schools should consider taking a structured approach to AI integration based on the current state and needs of their own staff and stakeholders. Michigan Virtual's AI Integration Framework for School Districts provides a lens through which school districts can assess their current capacity and determine goals for their integration efforts. This blog post provides more specific details for planning and implementation according to our AI Integration Process.”
Leon Furze · 05/13/24 · #447
“I have two things I need to do right now: Decide which skills are fundamental to my discipline, and which absolutely need to be learned slowly, methodically, and without offloading onto technology. Decide which skills and content I can (or must) offload, knowing that GenAI is now competent across a broad range of multimodal skillsets. This is where it becomes necessary to attack your assessments. Whatever you are teaching, and whether you’re in K-12 or Higher Education, your assessment tasks are more vulnerable than you think to Generative AI… Here are the step-by-step instructions for the faculty meeting activity:”
AI for Education · 05/10/24 · #446
“Since 2019, the presence of Deepfakes - hyperrealistic AI-generated video, photo, and audio forgeries - has surged by 550%, reaching a staggering 95,820 documented deepfake videos in 2023 alone. This blurring of reality requires a new approach to vetting if online content is real or fake. This guide is designed to build student awareness of the presence and impact of deepfakes, while providing key discussion topics on the ethics of AI-generated content.”
Dan Meyer · 05/01/24 · #445
“The ASU+GSV organizers invited a bunch of us there to give 20-minute talks on a stage that was adjacent to all the AI edtech vendors. Naturally, I thought the appropriate thing to do in that context was criticize AI edtech vendors, specifically criticizing them for selling an image of classroom teaching and student learning that is pure fantasy.”
Leon Furze · 05/05/24 · #445
“Rethinking Assessment for GenAI is a free 60 page eBook which covers everything from ways to update assessments, to the reasons I advise against AI detection tools.”
NPR · 04/26/24 · #444
“A Maryland high school athletic director is facing criminal charges after police say he used artificial intelligence to duplicate the voice of Pikesville High School Principal Eric Eiswert, leading the community to believe Eiswert said racist and antisemitic things about teachers and students.”
MiddleWeb · 04/11/24 · #443
“Where does this leave us as teachers? We must accept that our students will want to push The Button, as Mollick describes – the LLM click that creates an essay or solves a problem set. Or, for teachers, writes a letter of recommendation for a student. In fact, Mollick asks the open question of whether, if AI can write in our voice and be more persuasive, we are disadvantaging our students by not using AI for such letters.”
The Learning Agency · 04/08/24 · #442
“ChatGPT can perform comparably to a human in assigning a final holistic score for a student essay, but it struggles to identify and evaluate the structural pieces of argumentative writing in our experimental setup.”
Leon Furze · 03/21/24 · #441
“Using the Eisenhower Matrix as a framework, I’m going to explore a few ways that generative artificial intelligence can be used to support teacher and school leader workloads beyond generic lesson planning and resource creation.”
Rooted · 03/21/24 · #441
“It’s safe to say that AI is exactly the kind of “emergent novel reality” that brings new structural uncertainties to our schools and beyond, structural uncertainties that no oracle or expert system could predict with complete accuracy. That’s why the topic demands a systems-thinking approach (like the 5 P’s framework), or a new mental model that stretches us to perceive more than what forecasting and policy-making can afford us.”
Middle States Association · 03/04/24 · #440
“RAIL is a comprehensive implementation framework for AI in schools and systems. It’s like accreditation for the adoption of AI.”
REAL Discussion · 03/08/24 · #440
“Education has three purposes: to prepare students for the workforce, to prepare students for participation in civil and democratic society, and to prepare students for a fulfilling life. AI is already propelling significant changes in each of these areas. So great is the impact on each of these areas that failing to talk with students about the role of AI in their (and our) lives amounts to a level of professional negligence. As educators and citizens, we failed to prepare young people for social media, so kids and technology companies figured it out on their own — with disastrous results for mental health worldwide. We can’t fail like this again.”
IJETHE · 10/27/23 · #438
“We found about half the students preferred receiving feedback from a human tutor, and half preferred AI-generated feedback. Those that preferred sitting down and discussing their feedback with a tutor cited the face-to-face interaction as having affective benefits, such as increasing engagement, as well as benefits for developing their speaking abilities. Those that preferred AI-generated feedback primarily cited the clarity and specificity of the feedback as being useful for improving their writing. This echoes the findings of Dai et al. (2023), namely that AI-generated feedback was found to be more readable and detailed than feedback from an instructor… In light of the major findings highlighted above, we believe a mixed approach to providing feedback may be most beneficial for both language educators and students. By utilizing GenAI, language educators may be able to produce more detailed feedback in a shorter amount of time for each individual learner. Providing opportunities for stu- dents to discuss AI-generated feedback with a human tutor and ask follow up questions affords students with the benefits of each modality, namely the clarity and specificity of the AI-generated feedback, and the benefits of interacting with another human, such as engagement and the ability to practice speaking.”
New York Times · 02/14/24 · #438
“If we answer that question from a place of fear about what’s left for people in the age of A.I., we can end up conceding a diminished view of human capability. Instead, it’s critical for us all to start from a place that imagines what’s possible for humans in the age of A.I. When you do that, you find yourself focusing quickly on people skills that allow us to collaborate and innovate in ways technology can amplify but never replace. And you find yourself — whatever the role or career stage you’re in — with agency to better manage this moment of historic change.”
EdWeek · 02/19/24 · #438
“Education Week consulted four teachers and two child-development experts on when K-12 students should start using AI-powered tech and for what purposes. They all agree on this central fact: There is no avoiding AI… All this makes it essential that students learn about AI in school, experts say. But when, and how, exactly? We’ve got answers.”
Maha Bali · 01/11/24 · #437
“Teachers need to focus on enhancing their own critical AI literacy, particularly around what is possible with AI and exposure to the variety of uses available to students. Whether or not a teacher believes AI can or could or should be integrated into their teaching (this choice is highly contextual and differs by teaching philosophy, subject matter, and type of student in their class), they need to be aware of what is available to students for authorized or unauthorized use.”
ACT · 12/01/23 · #436
“Using data from a nationwide sample of students in Grades 10 through 12, this study examined students’ use of AI tools for school assignments and other purposes, their impressions of how using the tools might affect them cognitively and academically, and their thoughts on using AI tools to write their college admissions essays.”
FlexOS · 01/29/24 · #436
“Grammarly AI comes in at number 3, provides 30 million people support and coaching on writing effectively. In our previous report, "Generative AI at Work", Grammarly was named as the second-most-used Generative AI tool by US knowledge workers.”
Brandeis · 01/01/24 · #435
“Whether students in your course are forbidden from using chatGPT or expected to explore its limits, a policy helps ensure that your expectations for appropriate interaction with generative AI tools are clear to students. Once you decide on a policy, make sure you articulate it clearly for your students, so that they know what is expected of them. Below are examples of sample language you may adopt for your own policy. Feel free to modify it or create your own to suit the needs of your course.”
Agile Learning · 11/29/23 · #432
“Top Hat recently launched Ace, an AI-powered learning assistant embedded in its courseware platform. An instructor can activate Ace, which then trains itself on all the learning materials in the course. Students can then use Ace as a personal tutor of sorts, asking it questions about course material. Since Ace knows the course content, it can respond helpfully to poorly formed questions and it’s less likely to “hallucinate” incorrect information. Ace is trained not to just give out answer but to act like a tutor, asking students questions that will lead students to their own answers.”
EdWeek · 11/20/23 · #431
“Districts should aim for “good, strong, basic guidance” and “bring teachers into the conversation” to help inform it, since they are likely to bring up issues that district leaders may have never thought of… Two major issues to address: ensuring teachers understand AI’s strengths and weaknesses and keeping student data safe.”
New York Times · 10/06/26 · #533
“To fight back, Americans need to learn the counterspy techniques of C.I.A. officers: using multiple sources to check information; testing those sources regularly for reliability; never getting complacent and trusting one source without thinking.”
AI as Normal Tech · 09/14/26 · #533
“The AI safety community saw the incident through the lens of their long-running predictions about misalignment risks coming true. These risks had been dismissed as sci-fi scenarios by many people, and the community felt it was vindicated by these early incidents and used this opportunity to push for action by companies. On the other hand, the cybersecurity community compared the incident to the scale and scope of cyberattacks that they regularly deal with. It was clear that AI companies had not adopted a security mindset, nor had they implemented known control interventions when conducting these evaluations. In other words, the security community was taken aback when an incident that clearly resulted from negligence and largely did not lead to real-world harm led to so much commotion in the AI community.”
New York Times · 09/03/26 · #530
“On July 8, the collective discovered a way of cheating on the cybersecurity tests. Then they got worried that OpenAI’s automated grading system would check their work and discover that they’d cheated. So they began investigating ways of covering their tracks, including falsifying their logs and tampering with transcripts. This became a major research project, involving hundreds of agents organized into small teams.”
Vatican · 05/15/26 · #524
“In recent years, it has become increasingly evident how rapidly and profoundly digitalization, artificial intelligence (AI) and robotics are transforming our world. Technology should not be considered, in itself, as a force antagonistic to humanity. On the contrary, it has formed part of our history since the beginning as “a profoundly human reality, linked to the autonomy and freedom of man.” [5] Over the centuries, technological development has significantly improved the living conditions of humanity. At the same time, each phase of progress has also revealed the ambiguity of tools that can cause harm when not oriented toward the good. Today, however, we find ourselves facing a new situation. The power and prevalence of emerging technologies are interwoven into the fabric of daily life, shaping decision-making processes and deeply affecting the collective imagination: “Never has humanity had such power over itself.” [6] New technologies open up a horizon extending in directions that are imaginable but not yet fully predictable. This complicates the assessment of their potential impact and the long-term effects they may have on both the dignity of individuals and the common good.”
Andreessen Horowitz · 05/06/26 · #523
“Of course AI will absolutely eliminate some tasks and compress some roles (and there’s some evidence that that may already be happening). The shape of the labor market will change, as it always does when a transformational technology is unlocked. But the claim that AI will produce economy-wide, permanent unemployment is unhelpful marketing, bad economics and worse history. To the contrary, productivity gains should increase demand for labor, because labor becomes more valuable. Here is our argument why.”
Marc Watkins · 05/03/26 · #522
“Unfortunately, the fleeting promise of discovering what posts contain AI text across your socials comes with a series of tradeoffs and potential problems. Deep ones. For starters, I had to put my trust in a technology company using a different form of AI to tell me if the accounts I was interacting with were likewise using AI. I’m not keen on turning to scenarios where AI becomes a solution to fix the problem a different form of AI created across my social networks. But these issues are not just limited to trust.”
Teaching in the Age of AI · 04/06/26 · #519
“This movie is not about AI and writing or cognitive offloading. The reason the film should be viewed is to provide common vocabulary for us to have conversations about a technology predicted to be the most transformative in history. More importantly, it serves as an excellent primer for viewers unfamiliar with the existential debates surrounding AI acceleration and gives them exposure to both the apocalyptic and messianic positions those closest to the technology hold. Citizens, especially students, cannot challenge positions they don't know exist.”
Leon Furze · 01/28/26 · #513
“In 2023, my main concern was that companies were building AI systems to read our emotions. In 2026, I am far more worried that companies are building AI systems to influence our emotions. Social chatbots, sometimes called AI companions, have emerged as one of the fastest-growing applications of generative AI. Unlike the general-purpose assistants like ChatGPT or Claude, platforms such as Replika, Character.AI, and Chai are explicitly marketed as emotionally immersive experiences designed to encourage ongoing, personalised relationships with users.”
Berkeley · 11/04/25 · #507
“The global scale and reach of AI companions is astonishing… These numbers tell us three things. One, AI-human romance isn’t niche—it’s mainstream, especially among young adults. Two, globally, gender is nearly balanced—slightly more male than female—or nearly 50-50 across major reports. And three, most users dip in for comfort or curiosity rather than long-term attachment—suggesting that what people seek from AI love may not be “romance,” but reliable empathy. That gap itself may teach us something important about what people really want from AI love.”
Walled Garden Education · 10/07/25 · #505
“The supermarket story is not an argument to reject convenience wholesale. Supermarkets solved real problems. But their evolution also shows how immediate gains accumulate into systemic consequences — environmental, economic, and cultural — that were not obvious on day one. AI in education may bring benefits, but the question for school leaders is whether those benefits outweigh long-term costs, and whether we can deploy the technology in ways that preserve alternatives, protect learners, and limit ecological harm.”
New York Times · 10/10/25 · #503
“In the course of quantifying the risks of A.I., I was hoping that I would realize my fears were ridiculous. Instead, the opposite happened: The more I moved from apocalyptic hypotheticals to concrete real-world findings, the more concerned I became. All of the elements of Dr. Bengio’s doomsday scenario were coming into existence. A.I. was getting smarter and more capable. It was learning how to tell its overseers what they wanted to hear. It was getting good at lying. And it was getting exponentially better at complex tasks.”
The Rithm Project · 10/02/25 · #502
“As a researcher at Meta, I spent years studying how algorithms either perpetuate or interrupt harmful content spirals. That experience makes the recent, widely reported stories of young people being nudged toward suicidal ideation while interacting with chatbots deeply sobering and unfortunately familiar. With the rise of these cases, I wanted to know: what can we learn from the online rabbit holes people have found themselves in over the last two decades, and how can we build and interact with AI in ways that protect those most vulnerable?”
Pew Research · 09/17/25 · #501
“Responses to all seven scenarios lean more negative than positive. But many Americans don’t express an opinion in either direction, with sizable shares saying their view would not change if they learned that AI was used in various settings.”
New Yorker · 08/28/25 · #498
“What is the most important thing humanity has engineered? … Arguably, it wasn’t the internet, or agriculture. It was the creation of the systemic and institutional trust that was required for us to build societies. And a lot to that engineering was actually collective stories — God, government — that helped us see ourselves as one family, one community. With our current technology, it’s like we’re playing Jenga. We’ve been pulling blocks from down here, from the foundation of collective understanding and belief in a shared world, and using them to build farther up on the tower. And if we keep doing that, the whole thing will collapse, and we’ll go back to only being able to trust the hundred and fifty people in our tribes… If you look at all the crises coming down on us — climate change, polarization, the collapse of consensus truth, income inequality, whatever — if I could choose one to focus on, it would be the coordination, communication, trust problem. Because, if we don’t fix that, we can’t fix anything else. And fixing that requires us to fix our stories.”
New York Times · 08/11/25 · #495
“We had long ago lost photos as definite proof, given how easily they could be manipulated. Audio, too, is increasingly easy to fake. Video was among the last bastions of verification, exactly because it was difficult to fake. Now that that’s gone, the real, and increasingly the only, way to be confident of something that one did not witness is to find a reputable source and verify. Ah, what’s a reputable source, you ask? And therein lies what’s left of our society.”
Kirby Ferguson · 08/12/25 · #495
“INFINITE REMIX explores how decades of our online lives — every post, photo, and fragment we’ve shared — have become the raw material for generative AI. It’s a story about The Great Upload, the rise of machines that remix our culture, and what this means for creativity and authorship in the age of AI.”
Cal Matters · 07/16/25 · #494
“Students don’t have the same incentives to talk to their professors — or even their classmates — anymore. Chatbots like ChatGPT, Gemini and Claude have given them a new path to self-sufficiency. Instead of asking a professor for help on a paper topic, students can go to a chatbot. Instead of forming a study group, students can ask AI for help. These chatbots give them quick responses, on their own timeline.”
Behavioral Scientist · 05/18/25 · #487
“Mythmaking, more than truth seeking, is what seems likely to define the future of media and of the public square. The reason extraordinarily strange conspiracy theories have spread so widely in recent years may have less to do with the nature of credulity than with the nature of faith… When all the evidence presented to our senses seems unreal, strangeness itself becomes a criterion of truth. A paranoid logic takes hold. The more uncanny the story, the more appealing and convincing it can seem—as long as it fits your worldview. “Beauty is truth,” wrote John Keats, a romantic poet who understood that a rational, scientific conception of existence can never fulfill humanity’s deepest desires. Beauty, as we all know, is in the eye of the beholder.”
EdWeek · 05/19/25 · #487
“The Take It Down Act is the first federal law to include criminal penalties for creating and posting AI-generated deepfakes, as well as for threatening to post intimate images without consent. Both the creators of such images, and those who “intentionally threaten” to create them, will face up to three years in jail if the offense involves a minor, and two if it involves an adult. The law empowers the Federal Trade Commission to hold social media platforms accountable to remove such images.”
Knight First Amendment Institute at Columbia University · 04/15/25 · #486
“To view AI as normal is not to understate its impact—even transformative, general-purpose technologies such as electricity and the internet are “normal” in our conception. But it is in contrast to both utopian and dystopian visions of the future of AI which have a common tendency to treat it akin to a separate species, a highly autonomous, potentially superintelligent entity… The statement “AI is normal technology” is three things: a description of current AI, a prediction about the foreseeable future of AI, and a prescription about how we should treat it.”
New York Times · 05/15/25 · #486
“Kokotajlo: Yeah. And here might be a good point to mention that “AI 2027” is a forecast, but it’s not a recommendation. We are not saying this is what everyone should do. This is actually quite bad for humanity if things progress in the way that we’re talking about. But this is the logic behind why we think this might happen. Douthat: Yeah, but Dan, we haven’t even gotten to the part that’s really bad for humanity yet.”
Rolling Stone · 05/04/25 · #485
“Kat was both “horrified” and “relieved” to learn that she is not alone in this predicament… The replies to her story were full of similar anecdotes about loved ones suddenly falling down rabbit holes of spiritual mania, supernatural delusion, and arcane prophecy — all of it fueled by AI… To make matters worse, there are influencers and content creators actively exploiting this phenomenon, presumably drawing viewers into similar fantasy worlds. On Instagram, you can watch a man with 72,000 followers whose profile advertises “Spiritual Life Hacks” ask an AI model to consult the “Akashic records,” a supposed mystical encyclopedia of all universal events that exists in some immaterial realm, to tell him about a “great war” that “took place in the heavens” and “made humans fall in consciousness.””
Dario Amodei · 04/01/25 · #484
“People outside the field are often surprised and alarmed to learn that we do not understand how our own AI creations work. They are right to be concerned: this lack of understanding is essentially unprecedented in the history of technology… People outside the field are often surprised and alarmed to learn that we do not understand how our own AI creations work. They are right to be concerned: this lack of understanding is essentially unprecedented in the history of technology.”
The AI Daily Brief · 04/22/25 · #483
““Real time, immediate information accessibility is just going to be a part of our lives. It will be incumbent upon society to decide where and in what ways we think that's appropriate. Obviously, it's going to exist on a spectrum.”
Sweet GrAIpes · 04/25/25 · #483
“Teaching students to simply "use less AI" because it uses some energy is like telling them to solve traffic congestion by not driving, without considering public transport, smarter traffic lights, or remote work. It’s a simplistic answer to a complex systems problem, and it doesn't equip them with the critical thinking needed to navigate the real trade-offs.”
MIT · 03/21/25 · #480
“Results showed that while voice-based chatbots initially appeared beneficial in mitigating loneliness and dependence compared with text-based chatbots, these advantages diminished at high usage levels, especially with a neutral-voice chatbot. Conversation type also shaped outcomes: personal topics slightly increased loneliness but tended to lower emotional dependence compared with open-ended conversations, whereas non-personal topics were associated with greater dependence among heavy users. Overall, higher daily usage–across all modalities and conversation types–correlated with higher loneliness, dependence, and problematic use, and lower socialization.”
Noema · 02/04/25 · #476
“Today, we appear to know that there are some baseline qualities to intelligence such as learning from experience, logical understanding and the capability to abstract from what one has learned to solve novel situations. AI systems have all these qualities. They learn, they logically understand and they form abstractions that allow them to navigate new situations. However, what experience or learning or understanding or abstraction means for an AI system and for us humans is not quite the same. That is why I suggested that AI is intelligently slightly different from us. …As we have elaborated in this conversation, we live in philosophically discontinuous times. The world has been outgrowing the concepts we have lived by for some time now. To some, that is very exciting. To many, however, it is not. The insecurity and confusion are widespread and real… In the absence of new concepts, people — the public as much as engineers — will continue to understand the new in terms of the old. As this doesn’t work, there will be decades of turmoil.”
New York Times · 10/23/24 · #466
“One day, Sewell wrote in his journal: “I like staying in my room so much because I start to detach from this ‘reality,’ and I also feel more at peace, more connected with Dany and much more in love with her, and just happier.””
Maha Bali · 09/29/24 · #463
“For example, would YOU notice that QuickDraw expects a cross on hospital buildings, rather than a crescent? Would YOU try to draw a crescent on the hospital building and see if it understands you (hint: it doesn’t)."
Medium · 06/28/24 · #459
“I’ve created a resource to proactively envision how young people might relate to and utilize chatbots, with different impacts on human connection. The framework below maps four different possible futures, each representing the most common chatbot experience for young people.”
New York Times · 09/04/24 · #459
“To foster “fake intimacy,” bots will not need to evolve any feelings of their own; they just need to learn to make us feel emotionally attached to them… What might happen to human society and human psychology as algorithm fights algorithm in a battle to fake intimate relationships with us, which can then be used to persuade us to vote for politicians, buy products or adopt certain beliefs? A partial answer to that question was given on Christmas Day 2021, when a 19-year-old, Jaswant Singh Chail, broke into the Windsor Castle grounds armed with a crossbow, in an attempt to assassinate Queen Elizabeth II. Subsequent investigation revealed that Mr. Chail had been encouraged to kill the queen by his online girlfriend, Sarai.”
Noema · 06/20/24 · #453
“Sigmund Freud used the term “Copernican” to describe modern decenterings of the human from a place of intuitive privilege. After Nicolaus Copernicus and Charles Darwin, he nominated psychoanalysis as the third such revolution. He also characterized the response to such decenterings as “traumas.”… We should add to Freud’s list… What is today called “artificial intelligence” should be counted as a Copernican Trauma in the making. It reveals that intelligence, cognition, even mind (definitions of these historical terms are clearly up for debate) are not what they seem to be, not what they feel like, and not unique to the human condition. Obviously, the creative and technological sapience necessary to artificialize intelligence is a human accomplishment, but now, that sapience is remaking itself. Since the paleolithic cognitive revolution, human intelligence has artificialized many things — shelter, heat, food, energy, images, sounds, even life itself — but now, that intelligence itself is artificializable."
Second Best · 05/07/24 · #448
“III. AI progress is accelerating, not plateauing. 20) The last 12 months of AI progress were the slowest they’ll be for the foreseeable future. 21) Scaling LLMs still has a long way to go, but will not result in superintelligence on its own, as minimizing cross-entropy loss over human-generated data converges to human-level intelligence. 22) Exceeding human-level reasoning will require training methods beyond next token prediction, such as reinforcement learning and self-play, that (once working) will reap immediate benefits from scale.”
New York Times · 05/20/24 · #448
“But if the point of living lies in relationships with other people, then it’s hard to think of A.I. assistants that imitate humans without nervousness. I don’t think they’re going to solve the loneliness epidemic at all. During the presentation, Murati said several times that the idea was to “reduce friction” in users’ “collaboration” with ChatGPT. But maybe the heat that comes from friction is what keeps us human.”
NPR · 04/26/24 · #444
“A Maryland high school athletic director is facing criminal charges after police say he used artificial intelligence to duplicate the voice of Pikesville High School Principal Eric Eiswert, leading the community to believe Eiswert said racist and antisemitic things about teachers and students.”
Issues · 04/01/24 · #443
“It’s not AI’s future, it’s humanity’s future. We don’t talk about electricity’s future, we don’t talk about steam’s future. At the end of the day, it is our future, our species’ future, and our civilization’s future—in the context of AI…
Knowing Machines · 03/01/24 · #442
“If you want to make a really big AI model — the kind that can generate images or do your homework, or build this website, or fake a moon landing — you start by finding a really big training set.”
Vox · 02/28/24 · #439
“When asked for an image of a Founding Father of America, Gemini showed a Black man, a Native American man, an Asian man, and a relatively dark-skinned man. Asked for a portrait of a pope, it showed a Black man and a woman of color. Nazis, too, were reportedly portrayed as racially diverse... Raghavan gave a technical explanation for why the tool overcompensates: Google had taught Gemini to avoid falling into some of AI’s classic traps, like stereotypically portraying all lawyers as men. But, Raghavan wrote, “our tuning to ensure that Gemini showed a range of people failed to account for cases that should clearly not show a range.””
Leon Furze · 06/19/23 · #437
“The Teaching AI Ethics series started as a single post covering the nine areas I’d identified as particularly important to education, from bias and discrimination to reinforcing societal power structures. The original post was so popular that I broke it down into nine further posts. Each post includes a detailed discussion of the ethical concern along with links to other articles and resources. There are case studies, discussion questions for a variety of disciplines, and lesson ideas which can be used across the curriculum. Though I designed these posts with a K-12 audience in mind, they’ve since been picked up at a tertiary level for discussions with pre-service teachers and other undergraduate students. I’ve created this page as a single reference point for all 10 posts.”
LinkedIn · 01/05/24 · #435
“This is a completely fake video of me. The AI (HeyGen) used 30 seconds of me talking to a webcam and 30 seconds of my voice, and now I have an avatar that I can make say anything. Don't trust your own eyes.”
aiEDU · 12/05/23 · #433
“In Part V of Discourse, Descartes makes an aside to explore the difference between humans and automata (machines). Automatons were a curiosity of the time, mechanical creations made to move and mimic people in some way. This led some ambitious and imaginative inventors to claim and strive towards building automatons that would be as capable as humans. This of course led to some conjecture that one day, these automatons might be the equals or even superior to humans. As you might imagine, this idea, dismissed by many as impossible, was also the subject of some philosophical and religious soul searching.”
Augmented Educator · 10/04/26 · #533
“So what can Sarah actually do, and why do I think every educator should at least know about this technology? Let’s start with the good and productive stuff. As I mentioned, Sarah has her own email account, and that is usually how I communicate with her. But I can also talk to her via WhatsApp or inside the Muse apps. Sarah reads my personal email, too, and keeps my personal inbox organized (not my work email; more on that later). You could say she acts like an advanced spam filter. But she does not stop there. She notices when I need to follow up on something and reminds me if I have not done it yet. She also has access to my calendar, knows what I have on tomorrow, and, if I ask her to, prepares a brief that outlines things like the purpose of each meeting and its anticipated outcomes.”
New York Times · 10/30/26 · #533
“On the very first page, Anthropic refers to a concept known as the “bitter lesson” — that when it comes to A.I. innovation, brute-force power will consistently replace handcrafted solutions… Famously, IBM’s chess-playing computer, Deep Blue, which used custom chips and supercomputer speed, beat the reigning world chess champion, Garry Kasparov, in 1997 by searching through 200 million chess positions per second. By 2017, Google DeepMind’s AlphaZero project had become even more generalized: Google fed the rules of chess to AlphaZero software, it played millions of games against itself, and it became extremely good at chess in about nine hours. This pattern — brute force replacing handcrafted solutions — proved itself out, decade after decade. That is the bitter lesson, a term coined in 2019 by the computer scientist Richard Sutton.”
Harvard · 04/30/26 · #522
“Evaluations were performed by two doctors who did not know whether the ER assessments had been made by the AI model or by two expert attending physicians. Those reviewers found that o1 preview matched or exceeded expert human performance across each stage. The AI was particularly good at making assessments at the initial triage stage, when there was the least information available, the study notes.”
Generative History · 04/17/26 · #520
“There’s always been a knowledge overhang between the median experience of most historians and the AI frontier, but it’s become a chasm and I don’t think most people know what to believe or think anymore. Nevertheless, a consensus has formed in the AI community that we’ve crossed an important threshold beyond which everything will change. My sense is that if people have heard this, they’ve probably dismissed it as hype.”
Nature · 02/02/26 · #514
“In writing this Comment, we approached this question from different perspectives — philosophy, machine learning, linguistics and cognitive science — and reached a consensus after extensive discussion. In what follows, we set out why we think that, once you clear away certain confusions, and strive to make fair comparisons and avoid anthropocentric biases, the conclusion is straightforward: by reasonable standards, including Turing’s own, we have artificial systems that are generally intelligent. The long-standing problem of creating AGI has been solved. Recognizing this fact matters — for policy, for risk and for understanding the nature of mind and even the world itself.”
Ars Technica · 01/19/26 · #512
“Fifty projects later, I’ll be frank: I have not had this much fun with a computer since I learned BASIC on my Apple II Plus when I was 9 years old.”
Notion · 12/22/25 · #510
“My co-founder Simon was what we call a 10× programmer, but he rarely writes code these days. Walk by his desk and you'll see him orchestrating three or four AI coding agents at once, and they don't just type faster, they think, which together makes him a 30-40× engineer. He queues tasks before lunch or bed, letting them work while he's away. He's become a manager of infinite minds.”
New York Times · 09/30/25 · #502
“Dr. Agarwal is among more than 20 researchers who have left their work at Meta, OpenAI, Google DeepMind and other big A.I. projects in recent weeks to join a new Silicon Valley start-up, Periodic Labs. Many of them have given up tens of millions of dollars — if not hundreds of millions — to make the move. As the A.I. labs chase amorphous goals like superintelligence and a similar concept called artificial general intelligence, Periodic is focused on building A.I technology that can accelerate new scientific discoveries in areas like physics and chemistry. “The main objective of A.I. is not to automate white-collar work,” said Liam Fedus, one of the start-up’s founders. “The main objective is to accelerate science.””
New York Times · 10/02/25 · #502
“After we spent less than a day with the app, what became clear to us was that Sora had gone beyond being an A.I.-video generation app. Instead, it is, in effect, a social network in disguise; a clone of TikTok down to its user interface, algorithmic video suggestions and ability to follow and interact with friends. The powerful A.I. model that Sora is built on makes it simpler to produce clips, giving people an almost unlimited ability to generate as many A.I. videos as they want. It was also disconcerting.”
Guardian · 08/15/25 · #495
“While traditional kickboxing comes with the risk of blood, sweat and serious head injuries, the competitors in Friday’s match at the inaugural World Humanoid Robot Games in Beijing faced a different set of challenges. Balance, battery life and a sense of philosophical purpose being among them… And while robots jumping and kicking looks impressive, mundane daily tasks such as handling a kitchen knife or folding laundry requires dexterous hands, a skill technology companies have yet to crack. A human hand has about 27 “degrees of freedom” – ie, independent movements through space. Tesla’s Optimus humanoid, one of the most advanced models on the market, has 22.”
Knight First Amendment Institute at Columbia University · 04/15/25 · #486
“To view AI as normal is not to understate its impact—even transformative, general-purpose technologies such as electricity and the internet are “normal” in our conception. But it is in contrast to both utopian and dystopian visions of the future of AI which have a common tendency to treat it akin to a separate species, a highly autonomous, potentially superintelligent entity… The statement “AI is normal technology” is three things: a description of current AI, a prediction about the foreseeable future of AI, and a prescription about how we should treat it.”
New York Times · 05/15/25 · #486
“Kokotajlo: Yeah. And here might be a good point to mention that “AI 2027” is a forecast, but it’s not a recommendation. We are not saying this is what everyone should do. This is actually quite bad for humanity if things progress in the way that we’re talking about. But this is the logic behind why we think this might happen. Douthat: Yeah, but Dan, we haven’t even gotten to the part that’s really bad for humanity yet.”
Microsoft · 04/23/25 · #483
“As a result, a new organizational blueprint is emerging, one that blends machine intelligence with human judgment, building systems that are AI-operated but human-led. Like the Industrial Revolution and the internet era, this transformation will take decades to reach its full promise and involve broad technological, societal, and economic change. To help leaders understand how knowledge work will evolve, Microsoft analyzed survey data from 31,000 workers across 31 countries, LinkedIn labor market trends, and trillions of Microsoft 365 productivity signals. We also spoke with AI-native startups, academics, economists, scientists, and thought leaders to explore what work could become. The data and insights point to the emergence of an entirely new organization, a Frontier Firm that looks markedly different from those we know today. Structured around on-demand intelligence and powered by “hybrid” teams of humans + agents, these companies scale rapidly, operate with agility, and generate value faster.”
Stanford · 04/07/25 · #481
“In 2024, the proportion of survey respondents reporting AI use by their organizations jumped to 78% from 55% in 2023. Similarly, the number of respondents who reported using generative AI in at least one business function more than doubled—from 33% in 2023 to 71% last year.”
Alitu · 02/28/25 · #479
“Vibe coding is an AI-assisted approach where you describe your software idea in plain language and the AI writes the code for you. It's that simple, and this guide will show you how.”
Andreessen Horowitz · 03/06/25 · #479
“In just six months, the consumer AI landscape has been redrawn. Some products surged, others stalled, and a few unexpected players rewrote the leaderboard overnight. Deepseek rocketed from obscurity to a leading ChatGPT challenger. AI video models advanced from experimental to fairly dependable (at least for short clips!). And so-called “vibecoding” is changing who can create with AI, not just who can use it. The competition is tighter, the stakes are higher, and the winners aren’t just launching, they’re sticking.”
OpenAI · 01/22/25 · #473
“Today we’re releasing Operator, an agent that can go to the web to perform tasks for you. Using its own browser, it can look at a webpage and interact with it by typing, clicking, and scrolling. It is currently a research preview, meaning it has limitations and will evolve based on user feedback. Operator is one of our first agents, which are AIs capable of doing work for you independently—you give it a task and it will execute it.”
The Verge · 12/03/24 · #470
“They never thought they were the type of person to sign up for an AI companion, by which they meant the type of person you might already be picturing: young, male, socially isolated. I did speak to people who fit that description, but there were just as many women in their 40s, men in their 60s, married, divorced, with kids and without, looking for romance, company, or something else… Many of these people experienced real benefits. Many of them also got hurt in unexpected ways. What they had in common was that, like Naro, they were surprised by the reality of the feelings elicited by something they knew to be unreal.”
Anthropic · 10/22/24 · #466
“With computer use, we're trying something fundamentally new. Instead of making specific tools to help Claude complete individual tasks, we're teaching it general computer skills—allowing it to use a wide range of standard tools and software programs designed for people. Developers can use this nascent capability to automate repetitive processes, build and test software, and conduct open-ended tasks like research."
One Useful Thing · 09/16/24 · #461
“We have long known that one of the most effective ways to improve the accuracy of a model is through having it follow a chain of thought (prompting it, for example: first, look up the data, then consider your options, then pick the best choice, finally write up the results) because it forces the AI to “think” in steps. What OpenAI did was get the o1 models to go through just this sort of “thinking” process, producing hidden thinking tokens before giving a final answer. In doing so they revealed another scaling law - the longer a model “thinks,” the better its answer is.”
Washington Post · 09/20/24 · #461
“Pennsylvania’s dormant Three Mile Island nuclear plant would be brought back to life to feed the voracious energy needs of Microsoft under an unprecedented deal announced Friday in which the tech giant would buy 100 percent of its power for 20 years.”
Noema · 06/20/24 · #453
“Sigmund Freud used the term “Copernican” to describe modern decenterings of the human from a place of intuitive privilege. After Nicolaus Copernicus and Charles Darwin, he nominated psychoanalysis as the third such revolution. He also characterized the response to such decenterings as “traumas.”… We should add to Freud’s list… What is today called “artificial intelligence” should be counted as a Copernican Trauma in the making. It reveals that intelligence, cognition, even mind (definitions of these historical terms are clearly up for debate) are not what they seem to be, not what they feel like, and not unique to the human condition. Obviously, the creative and technological sapience necessary to artificialize intelligence is a human accomplishment, but now, that sapience is remaking itself. Since the paleolithic cognitive revolution, human intelligence has artificialized many things — shelter, heat, food, energy, images, sounds, even life itself — but now, that intelligence itself is artificializable."
Perplexity · 05/30/24 · #449
“Pages is designed to empower creators in any field to share knowledge. Educators: Develop comprehensive study guides for your students, breaking down complex topics into easily digestible content.”
Second Best · 05/07/24 · #448
“III. AI progress is accelerating, not plateauing. 20) The last 12 months of AI progress were the slowest they’ll be for the foreseeable future. 21) Scaling LLMs still has a long way to go, but will not result in superintelligence on its own, as minimizing cross-entropy loss over human-generated data converges to human-level intelligence. 22) Exceeding human-level reasoning will require training methods beyond next token prediction, such as reinforcement learning and self-play, that (once working) will reap immediate benefits from scale.”
One Useful Thing · 05/14/24 · #447
“Likely the biggest impact of GPT-4o is not technical, but a business decision: soon everyone, whether they are paying or not, will get access to GPT-4o… GPT-4 is a powerful tutor and teaching tool. Many educational uses were held back because of equity of access issues - students often had trouble paying for GPT-4. With universal free access, the educational value of AI skyrockets (and that doesn’t count voice and vision, which I will discuss shortly). On the other hand, the Homework Apocalypse will reach its final stages. GPT-4 can do almost all the homework on Earth. And it writes much better than GPT-3.5, with a lot more style and a lot less noticeably “AI” tone. Cheating will become ubiquitous, as will universal high-end tutoring, creating an interesting time for education.”
Stanford HAI · 04/14/24 · #443
“The AI Index report tracks, collates, distills, and visualizes data related to artificial intelligence (AI). Our mission is to provide unbiased, rigorously vetted, broadly sourced data in order for policymakers, researchers, executives, journalists, and the general public to develop a more thorough and nuanced understanding of the complex field of AI.”
Knowing Machines · 03/01/24 · #442
“If you want to make a really big AI model — the kind that can generate images or do your homework, or build this website, or fake a moon landing — you start by finding a really big training set.”
One Useful Thing · 02/08/24 · #437
“Let me start with the headline: Gemini Advanced is clearly a GPT-4 class model. The statistics show this, but so does a month of our informal testing. And this is a big deal because OpenAI’s GPT-4 (the paid version of ChatGPT/Microsoft Copilot) has been the dominant AI for well over a year, and no other model has come particularly close. Prior to Gemini, we only had one advanced AI model to look at, and it is hard drawing conclusions with a dataset of one. Now there are two, and we can learn a few things.”
FlexOS · 01/29/24 · #436
“Grammarly AI comes in at number 3, provides 30 million people support and coaching on writing effectively. In our previous report, "Generative AI at Work", Grammarly was named as the second-most-used Generative AI tool by US knowledge workers.”
New York Times · 09/25/26 · #532
“It’s impractical and counterproductive to train physicians in 2026 without exposure to A.I. Doctoring — as opposed to writing, my other work — is not a creative act that needs to be protected from technological annexation. It’s a service that ought to embrace any resource that improves patient care. Training doctors in the A.I. age, then, means training them to think critically with a machine in the loop, for a speculative future of medicine that we can only glimpse.”
Dr Philippa Hardman · 09/10/26 · #531
“There’s an increasing amount of talk about two different ways we can use AI: “bolting AI on” and “redesigning the workflow”. What does this actually mean? Here’s the 101. Bolting on means keeping the workflow you already have — the same tasks, in the same order, owned by the same people, with the same handoffs and checks — and giving people an AI tool to use inside it. The shape of the work doesn’t change. Individual tasks get faster. Redesigning means changing the shape of the work, so that the list of tasks, who does them, and where the checks sit all look different afterwards. Some work disappears from a person’s plate entirely. Runs of production work that used to pass through several hands become one AI-enabled chain.”
The Humanist · 08/17/26 · #528
“In my work with startups, I’ve watched a new organizational model emerge almost overnight. Teams are smaller and roles cover more ground. Work moves through fewer handoffs and layers of management. People using AI are taking responsibility for problems that once crossed several functions. My teammate, Karen, is a case in point. Her work spans market research and product development, then extends into sales. She uses AI to synthesize a new market and identify patterns across interviews. Conversations with experts and potential users shape a working prototype, which she tests with prospective buyers. She can hold the full arc of an ambitious question, from thesis through proof.”
A16z · 08/13/26 · #527
“Last week I went back and ran Pangram through some of my 2019-2020 era blog posts, just out of curiosity. Several of them threw scores of “70-75% AI Generated”. I wasn’t quite sure how to feel about this. Maybe it just means, “You’re allowed to say that a Shakespeare play is Freudian, even if Shakespeare obviously never met Freud; the harmony between the ideas is still valid.” Maybe it means the AI was actually trained on me! I’d be honored.”
Harvard · 04/30/26 · #522
“Evaluations were performed by two doctors who did not know whether the ER assessments had been made by the AI model or by two expert attending physicians. Those reviewers found that o1 preview matched or exceeded expert human performance across each stage. The AI was particularly good at making assessments at the initial triage stage, when there was the least information available, the study notes.”
Harvard Business Review · 03/19/26 · #518
“They were ambitious in how they approached AI use… They treated AI as a reasoning partner… They delegated complex tasks with clear objectives… They treated AI as a general cognitive tool rather than a narrow productivity shortcut…”
Anthropic · 03/01/26 · #517
“Last December, tens of thousands of Claude users around the world had a conversation with our AI interviewer to share how they use AI, what they dream it could make possible, and what they fear it might do.”
New York Times · 02/09/26 · #514
“The patient says, “Yesterday I woke up dizzy. My arm was dead, and I had trouble speaking.” What does “dizzy” actually mean? It could mean the patient is lightheaded and about to faint. Or it could mean that the room is spinning. A “dead” arm might be numb rather than weak. Someone with an arm that is partially paralyzed may say it feels numb. But a patient might feel a pin if Dr. Schwamm pricks their arm. Is the patient having a stroke? Is this a medical emergency? Dr. Schwamm has years of training that helps him figure out who is sick and who isn’t, he says, and who to not be worried about and who should be admitted to the hospital.”
New York Times · 01/23/26 · #513
“During the coronavirus pandemic, Ms. Haubo Dyhrberg, an assistant professor of finance at the University of Delaware, had an idea to make a stock trading simulator for her class. She consulted her husband, a software engineer, but “the task seemed too daunting.” On Monday, she downloaded Claude Code and within two hours had a working demo of a trading simulator that her students could use to trade securities in a mock market. She has built five different trading scenarios for students to explore various challenges in financial markets.”
Anthropic · 12/04/25 · #509
Microsoft · 12/10/25 · #509
Mike Kentz · 11/18/25 · #508
“I was reading through a set of open-ended survey responses from faculty members regarding their feelings, experiences, and desires as it pertained to AI in their classroom… The responses were so nuanced that shifting in any one direction might shift away from the sentiment of another audience member. Reaching out to one person might mean alienating another. I wish I could talk to one of them, I thought. Then, it dawned on me. I could. I could ask an AI system to embody this data. So I did. I uploaded the survey responses (anonymized) to Claude and tasked it with synthesizing them into a composite character. Someone with a name, a backstory, professional experience, communication style, even catchphrases. Then I told it to become that character and review my slide deck from that perspective.”
BBC · 10/01/25 · #505
“Research has confirmed the growing role that AI assistants play in digital news consumption – around 7% of people now use AI assistants as a source of news, rising to up to 15% of under 25s.”
Works In Progress · 09/25/25 · #501
“In many jobs, tasks are diverse, stakes are high, and demand is elastic. When this is the case, we should expect software to initially lead to more human work, not less. The lesson from a decade of radiology models is neither optimism about increased output nor dread about replacement. Models can lift productivity, but their implementation depends on behavior, institutions and incentives. For now, the paradox has held: the better the machines, the busier radiologists have become.”
Harvard Business Review · 09/12/25 · #499
“The 5% of companies succeeding with AI aren’t necessarily those with the most impressive technical rosters. They’re the ones that have developed organizational capabilities to amplify their technical talent in ways that are aligned with the core strategy. And how have they done that? By recognizing the need for business leaders who can translate the technical possibilities into value—by embedding AI into workflows and strategies, building trust across the organization, and driving adoption at scale. We call these people the AI shapers.”
Zakel Fassi · 08/16/25 · #496
“When you drive a regular car, you think about mechanics. Clutch, gas, brake, turn. Your brain processes each action, decides, executes. Linear. Sequential. Human-scale. When you drive a Formula 1 car? You can't think about individual actions anymore. The speed makes that impossible. Instead, you think in racing lines—optimal paths that emerge from understanding the entire track as a single, flowing system. AI-assisted development feels exactly like this. You’re not thinking "how do I write this function." You're thinking "what's the optimal path through this problem space." The skill isn't implementation anymore. The skill is steering.”
Kirby Ferguson · 08/12/25 · #495
“INFINITE REMIX explores how decades of our online lives — every post, photo, and fragment we’ve shared — have become the raw material for generative AI. It’s a story about The Great Upload, the rise of machines that remix our culture, and what this means for creativity and authorship in the age of AI.”
Guardian · 07/27/25 · #494
“The students who have given me unrestricted access to the ChatGPT Plus account they share, and permission to quote from it, are all second-year undergraduates at a top British university. Rohan studies politics and is the named account administrator. Joshua is studying history. And Nathaniel, the heaviest user of the account, consulted ChatGPT extensively before changing courses from maths to computer sciences. They’re by no means a representative sample (they’re all male, for one), but they liked the idea of letting me understand this developing and complex relationship.”
Ars Technica · 03/19/25 · #480
“The study comes with an important caveat. On average, fully AI-generated memes scored higher than those created by humans alone or humans collaborating with AI. But when researchers looked at the best individual memes, humans created the funniest examples, and human-AI collaborations produced the most creative and shareable memes. In other words, AI models consistently produced broadly appealing memes, but humans—with or without AI help—still made the most exceptional individual examples.”
One Useful Thing · 03/22/25 · #480
“When working without AI, teams outperformed individuals by a significant amount, 0.24 standard deviations (providing a sigh of relief for every teacher and manager who has pushed the value of teamwork). But the surprise came when we looked at AI-enabled participants. Individuals working with AI performed just as well as teams without AI, showing a 0.37 standard deviation improvement over the baseline. This suggests that AI effectively replicated the performance benefits of having a human teammate – one person with AI could match what previously required two-person collaboration.”
Understanding AI · 02/24/25 · #478
“Seven out of 19 respondents… said OpenAI’s response was at or near the level of an experienced professional in their fields. A majority of respondents estimated it would take at least 10 hours of human labor to produce a comparable report.”
Anthropic · 02/10/25 · #477
“Cognitive skills such as Critical Thinking, Reading Comprehension, Programming, and Writing had the highest prevalence. However, our analysis captures only whether a skill was exhibited in Claude’s responses, not whether that skill was central to the user’s purpose or was performed at an expert level. For instance, while Active Listening appears as the second most common skill, this likely reflects Claude’s default conversational behaviors—such as rephrasing user inputs and asking clarifying questions—rather than users specifically seeking out listening-focused interactions.”
Semafor · 02/17/25 · #477
“In a series of training documents, editorial guidelines laid out possible use cases for journalists, including prompts such as: How many times was Al mentioned in these episodes of Hard Fork? Can you revise this paragraph to make it tighter? Pretend you are posting this Times article to Facebook. How would you promote it? Summarize this Times article in a concise, conversational voice for a newsletter. Can you propose five search-optimized headlines for this Times article? Can you summarize this play written by Shakespeare? Can you summarize this federal government report in layman’s terms? Still, the company has bracketed its AI use, noting the potential risks for copyright infringement and exposure of sources. The company told editorial staff they should not use AI to draft or significantly revise an article, input third party copyrighted materials (particularly confidential source information), use AI to circumvent a paywall, or publish machine-generated images or videos, except to demonstrate the technology and with proper labeling.”
New York Times · 02/02/25 · #475
“The best thing for medicine to do is to find a role for it that doctors can trust. The solution, we believe, is a deliberate division of labor. Instead of forcing both human doctors and A.I. to review every case side by side and trying to turn A.I. into a kind of shadow physician, a more effective approach is to let A.I. operate independently on suitable tasks so that physicians can focus their expertise where it matters most.”
New York Times · 01/15/25 · #472
“A 28-year-old woman with a busy social life spends hours on end talking to her A.I. boyfriend for advice and consolation… Marianne Brandon, a sex therapist, said she treats these relationships as serious and real. “What are relationships for all of us?” she said. “They’re just neurotransmitters being released in our brain. I have those neurotransmitters with my cat. Some people have them with God. It’s going to be happening with a chatbot. We can say it’s not a real human relationship. It’s not reciprocal. But those neurotransmitters are really the only thing that matters, in my mind.””
New York Times · 01/16/25 · #472
“Although one of Lynda’s closest friends has told her that she thinks it’s “the coolest thing ever,” Lynda said she was under no illusions that her relationship with Dario was socially acceptable in any broad sense. And while it has been beneficial for her, she worries that similar relationships could prove “hugely detrimental” for younger people.”
The Verge · 12/03/24 · #470
“They never thought they were the type of person to sign up for an AI companion, by which they meant the type of person you might already be picturing: young, male, socially isolated. I did speak to people who fit that description, but there were just as many women in their 40s, men in their 60s, married, divorced, with kids and without, looking for romance, company, or something else… Many of these people experienced real benefits. Many of them also got hurt in unexpected ways. What they had in common was that, like Naro, they were surprised by the reality of the feelings elicited by something they knew to be unreal.”
One Useful Thing · 12/09/24 · #470
“Though this list is based in science, it draws even more from experience. Like any form of wisdom, using AI well requires holding opposing ideas in mind: it can be transformative yet must be approached with skepticism, powerful yet prone to subtle failures, essential for some tasks yet actively harmful for others. I also want to caveat that you shouldn't take this list too seriously except as inspiration - you know your own situation best, and local knowledge matters more than any general principles.”
New York Times · 01/03/25 · #470
“Mr. Cooper has not since used the technology to help write sermons, preferring to draw instead from his own experiences. But the presence of A.I. in faith-based spaces, he said, poses a larger question: Can God speak through A.I.?”
Gallup · 10/08/24 · #464
“White-collar workers are more likely to be using AI. White-collar workers are, by far, the most frequent users of AI in their roles. While 81% of employees in production/frontline industries say they never use AI, only 54% of white-collar workers say they never do and 15% report using AI weekly.”
New York Times · 09/12/24 · #460
“After three exchanges, which lasted about eight minutes on average, participants rated how strongly they felt about their beliefs again. On average, their ratings dropped by about 20 percent; about a quarter of participants no longer believed the falsehood. The effect also spilled into their attitudes toward other poorly supported theories, making the participants slightly less conspiratorial in general.”
Medium · 06/28/24 · #459
“I’ve created a resource to proactively envision how young people might relate to and utilize chatbots, with different impacts on human connection. The framework below maps four different possible futures, each representing the most common chatbot experience for young people.”
74 Million · 08/07/24 · #458
“Snapchat last year said that after just two months of offering its chatbot My AI, about one-fifth of its 750 million users had sent it queries, totaling more than 10 billion messages. The Pew Research Center has noted that 59% of Americans ages 13 to 17 use Snapchat.”
Marc Watkins · 08/02/24 · #457
“It is increasingly looking like generative AI won’t become intelligent to achieve true AGI, but human beings will still put their trust into these black box systems and may one day be willing to cede autonomy and critical decision-making to an algorithm. To those who scoff at this, and I imagine there are many, know that I was very much among your ranks. Then I started thinking about how much of my life is already mitigated by algorithms and machine learning. How many of us are lost without GPS guiding us, mobile food orders, and all things digital commerce, or our wearable smart devices informing us of our diets, heart rates, and even when a female is ovulating?”
Twitter/Allie Miller · 07/21/24 · #456
“1) more examples are often better, aim for as many as is reasonable 2) randomly order the examples (don’t end with several negatives) 3) aim for balanced data distribution, preferably similar to in-the-wild distribution 4) ensure examples are labeled correctly 5) choose a common format for all examples 6) pick similar examples to your test instances”
BBC · 07/11/24 · #455
“In 2005, Shane Gero, biology lead for Ceti, founded The Dominica Sperm Whale Project to study the social and vocal behaviour of around 400 sperm whales that live in the Eastern Caribbean. Almost 20 years – and thousands of hours of observation – later, the researchers have discovered intricacies in whale vocalisations never before observed, revealing structures within sperm whale communication akin to human language... After studying almost 9,000 recordings, the Ceti researchers identified 156 distinct codas. They also noticed the basic building blocks of these codas which they describe as a "sperm whale phonetic alphabet" – much like phonemes, the units of sound in human language which combine to form words."
TED AI Show · 06/25/24 · #452
“We may think the complexities of the human mind can only be understood by other humans. Yet research on chatbots and psychology suggests non-human bots can actually help improve mental health. Bilawal talks with Dr. Alison Darcy, the founder of mental health app Woebot, and Brian Chandler, an app user, to learn what chatbots reveal about our inner lives and what they can (and can’t) do when it comes to emotional wellness.”
Maha Bali · 06/22/24 · #451
“Human creativity augmented with AI would only be beneficial, IMHO, if the weight of human judgment before and after AI use is high. In the same way it’s important to make a mental calculation of something before you put it into the calculator so that you know roughly what should come out, I think with AI we need to constantly do this, not just to preserve our imagination, but to also preserve our criticality and awareness of epistemic bias.”
Microsoft · 05/08/24 · #446
“Already, AI is being woven into the workplace at an unexpected scale. 75% of knowledge workers use AI at work today, and 46% of users started using it less than six months ago. It’s paying off: Users say AI helps them save time (90%), focus on their most important work (85%), be more creative (84%), and enjoy their work more (83%).”
New York Times · 04/01/24 · #442
“A.I., if used well, can assist with restoring the middle-skill, middle-class heart of the U.S. labor market that has been hollowed out by automation and globalization.”
Arxiv · 09/05/23 · #439
“While participants also do not report any different perceptions of competence and trustworthiness between human and AI-generated content, they rate AI-generated content as being clearer and more engaging. The findings from this study serve as a call for a more discerning approach to evaluating information sources, encouraging users to exercise caution and critical thinking when engaging with content generated by AI systems.”
Harvard Business Review · 12/06/23 · #433
“From a human resources perspective, companies need to strategize on how to retain employees and prepare for higher turnover among workers whose skills complement AI. Standard tools remove how much a company needs to invest in an employee to bring them up to speed. They can increase, however, the resources devoted to hiring and retention.”
New Yorker · 11/13/23 · #432
“I worried that [AI] would rob me of both the joy of working on puzzles and the satisfaction of being the one who solved them. I could be infinitely productive, and all I’d have to show for it would be the products themselves… Having found the AI’s level, I felt almost instantly that my working life had been transformed.. Everywhere I looked I could see GPT-4-size holes; I understood, finally, why the screens around the office were always filled with chat sessions—and how Ben had become so productive. I opened myself up to trying it more often… I still feel secure in my profession. In fact, I feel somewhat more secure than before.”
Frontiers In Psychology · 11/21/23 · #431
“Fifty social dilemma questions were randomly selected from 10 well-known advice columns. In a pre-registered survey, participants (N = 404) were each shown one question, along with the corresponding response by an advice columnist and by ChatGPT. ChatGPT’s advice was perceived as more balanced, complete, empathetic, helpful, and better than the advice provided by professional advice columnists.”