“A growing field of computer science known as interpretability embodies the conceit that in order to narrow or even bridge the expanding knowledge gap between A.I. models and humans, we need to treat A.I. more like a natural phenomenon than a human invention.”
“The bad news is that if this tool falls into the hands of bad actors, they could hack pretty much every major software system in the world, including all those made by the companies in the consortium.”
“This time, Anthropic is making a different, more urgent claim. The company’s executives say Claude Mythos Preview is already capable of carrying out autonomous security research, including scanning for and exploiting so-called zero-day vulnerabilities in critical software programs, flaws that are unknown even to the software’s developer. These efforts can often be triggered by amateurs […]
“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 […]
“The sole intent of this piece is modeling a scenario that’s been relatively underexplored. Our friend Alap Shah posed the question, and together we brainstormed the answer. We wrote this part, and he’s written two others you can find here. Hopefully, reading this leaves you more prepared for potential left tail risks as AI makes […]
“Amanda Askell knew from the age of 14 that she wanted to teach philosophy. What she didn’t know then was that her only pupil would be an artificial-intelligence chatbot named Claude.”
“Because humans update their views partly based on social evidence—looking to peers to see what is “normal”—fabricated swarms can make fringe views look like majority opinions. If swarms flood the web with duplicative, crawler-targeted content, they can execute “LLM grooming,” poisoning the training data that future AI models (and citizens) rely on. Even so-called “thinking” […]
“Our central aim is for Claude to be a good, wise, and virtuous agent, exhibiting skill, judgment, nuance, and sensitivity in handling real-world decision-making, including in the context of moral uncertainty and disagreement. In this section, we discuss the high standards of honesty we want Claude to hold, and the nuanced reasoning we want Claude […]
“If you told someone about Figma or Google Adwords in the 1970s they’d have expected marketing jobs to plummet since we could do many different jobs inside of a single role in the future; well, the opposite has happened. Back of the envelope math (from AI of course) suggests that there were a few hundred […]
“Another example is that many teachers have declared they will never use AI for assessment because it’s central to their relationship with students. That’s a valid value. But it collapses all assessment into one category. Formative feedback—comments on a draft, a quick check on whether a student grasps a concept—has different goals than summative evaluation. […]
“Participants showed more openness to opposing views when the arguments were attributed to AI, perceiving it as more objective. The findings raise concerns about potential misinformation and increased polarization, with researchers urging caution in utilizing AI in political discourse.”
“Because data centers handle many types of workloads, it’s difficult to distinguish the exact share of their total electricity demand that comes from AI alone. But a typical AI-focused hyperscaler annually consumes as much electricity as 100,000 households. The larger ones currently under construction are expected to use 20 times as much, the IEA predicts.”
“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 […]
“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 […]
“What we found was incredibly fascinating, and reveals a mix of lawyers blaming IT issues, personal and family emergencies, their own poor judgment and carelessness, and demands from their firms and the industry to be more productive and take on more casework. But most often, they simply blame their assistants.”
“Now generative A.I. will compete with you for your power, water and land.”
“The most alarming cost may be interpersonal. Low effort, unhelpful AI generated work is having a significant impact on collaboration at work. Approximately half of the people we surveyed viewed colleagues who sent workslop as less creative, capable, and reliable than they did before receiving the output. Forty-two percent saw them as less trustworthy, and […]
“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.”
“Technical reports often compare LLMs’ outputs with “human” performance on various tests. Here, we ask, “Which humans?” Much of the existing literature largely ignores the fact that humans are a cultural species with substantial psychological diversity around the globe that is not fully captured by the textual data on which current LLMs have been trained. […]
“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 […]
“Why Adam took his life — or what might have prevented him — is impossible to know with certainty. He was spending many hours talking about suicide with a chatbot. He was taking medication. He was reading dark literature. He was more isolated doing online schooling. He had all the pressures that accompany being a […]
“When A.I. chatbots are purposely trained as digital therapists, they show more promise. One example is Therabot, designed by Dartmouth College researchers. In a randomized controlled trial completed earlier this year, adult participants who used Therabot reported significant reductions in depression, anxiety and weight concerns. They also expressed a strong sense of connection to the […]
“In total, the median prompt—one that falls in the middle of the range of energy demand—consumes 0.24 watt-hours of electricity, the equivalent of running a standard microwave for about one second. The company also provided average estimates for the water consumption and carbon emissions associated with a text prompt to Gemini.”
“The arrival of Seemingly Conscious AI is inevitable and unwelcome. Instead, we need a vision for AI that can fulfill its potential as a helpful companion without falling prey to its illusions… those actually working on the science of consciousness tell me they are inundated with queries from people asking ‘is my AI conscious?’ What […]
“We recently gave Claude Opus 4 and 4.1 the ability to end conversations in our consumer chat interfaces. This ability is intended for use in rare, extreme cases of persistently harmful or abusive user interactions. This feature was developed primarily as part of our exploratory work on potential AI welfare, though it has broader relevance […]
“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 […]
“This is distinct from the post‑truth era we’re still free‑falling through. Post‑truth meant arguing over facts inside a loosely shared stage—often delusional, but still shared. Most people agreed COVID‑19 existed (come on, folks!), but the debates were about masks, data points, and motives. In a post‑reality era, reality may be AI-generated, but it’s stamped as […]
“Content generated by artificial intelligence has become a factor in elections around the world. Most of it is bad, misleading voters and discrediting the democratic process.”
“We had our research conference a week ago and almost every HR leader or head of recruiting told me that they are rebuilding their entry level development programs for young employees. Why? Driven by tightening the budgets and entry hire slowdowns, their talent pipeline was weakened.”
“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 […]
“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 […]
“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 […]
“Children shouldn’t speak with companion chatbots because such interactions risk self harm and could exacerbate mental health problems and addiction. That’s according to a risk assessment by children’s advocacy group Common Sense Media conducted with input from a lab at the Stanford University School of Medicine.”
“Social AI companions have unacceptable risks for teen users and should not be used by anyone under the age of 18. These social AI companions are designed to create emotional attachment and dependency. This is particularly concerning for developing adolescent brains that may struggle to maintain healthy boundaries between AI and human relationships.”
“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 […]
“Human welfare is at the heart of our work at Anthropic: our mission is to make sure that increasingly capable and sophisticated AI systems remain beneficial to humanity. But as we build those AI systems, and as they begin to approximate or surpass many human qualities, another question arises. Should we also be concerned about […]
““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.”
“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 […]
“Who are we? Daniel Kokotajlo (TIME100, NYT piece) is a former OpenAI researcher whose previous AI predictions have held up well. Eli Lifland co-founded AI Digest, did AI robustness research, and ranks #1 on the RAND Forecasting Initiative all-time leaderboard. Thomas Larsen founded the Center for AI Policy and did AI safety research at the […]
“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 […]
“We observe that very high usage correlates with increased self-reported indicators of dependence. From our RCT, we find that the impact of voice-based interactions on emotional well-being to be highly nuanced, and influenced by factors such as the user’s initial emotional state and total usage duration”
“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 […]
“So these AI companies, out of sheer necessity, are essentially transforming into energy companies.”
“Another lawyer was caught using AI and not checking the output for accuracy, while a previously-reported case just got hit with sanctions.”
“Early therapy chatbots, such as Woebot and Wysa, were trained to interact based on rules and scripts developed by mental health professionals, often walking users through the structured tasks of cognitive behavioral therapy, or C.B.T. Then came generative A.I., the technology used by apps like ChatGPT, Replika and Character.AI. These chatbots are different because their […]
“But a metaphor of hallucination reinforces the misconception that AI is conscious; it implies that AI experiences reality and sometimes becomes delirious… Ultimately, we chose a more fitting term: AI mirage. Just as a desert mirage is an artifact of physical conditions, an AI mirage is an artifact of how systems process training data and […]
“A misconception about voice acting is that it takes only a voice. But our bodies and souls are involved to get the proper believability. When I first watched Dan Castellaneta, who plays Homer, and Harry Shearer, who plays Mr. Burns and many other characters, doing vocal recordings, I was almost embarrassed by how silly they […]
“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 […]
“Jevons paradox may remain somewhat obscure, but over the last few decades, it has been cited as a reason why more energy-efficient cars, appliances and light bulbs may fail to reduce our consumption of fossil fuels. It has been cited as a reason why building more highway lanes fails to solve traffic congestion. And it’s […]
“Imagine that you’re talking with someone over Zoom and, as the conversation goes on, you begin to feel that they really get you — to the extent that a strong bond of trust quickly forms. Now imagine that this sense of trust is the result of an AI that, unknown to either of you, is […]
“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.””
“For many in Silicon Valley and Washington, D.C., the biggest fear is that China or Russia rolls out fully autonomous weapons first, forcing the U.S.’s hand. At a UN debate on AI arms last year, a Russian diplomat was notably coy. “We understand that for many delegations the priority is human control,” he said. “For […]
“Three years after the 30-year-old South Korean woman received a barrage of online fake images that depicted her nude, she is still being treated for trauma. She struggles to talk with men. Using a mobile phone brings back the nightmare. “It completely trampled me, even though it wasn’t a direct physical attack on my body,” […]
“We scoped the work in three categories of representational harms: erasure, subordination, and stereotypes.”
“If you asked me how to scale clean energy, I would prescribe a magical source of urgent energy demand. Someone willing to pay a premium to build solar+batteries, geothermal, and nuclear, in order to bring them down the cost curve and make them cheaper for everyone. That is exactly what AI data centres are. For […]
“Two Harvard students have created an eerie demo of how smart glasses can use facial recognition tech to instantly dox people’s identities, phone numbers, and addresses. The most unsettling part is the demo uses current, widely available technology like the Ray-Ban Meta smart glasses and public databases.”
“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).”
“Since 2020, CDT has conducted annual or semi-annual surveys with students, teachers, and/or parents. The surveys measure and track changes in perceptions, experiences, training, engagement, and concerns about student data privacy, student activity monitoring, content filtering and blocking software, generative AI, NCII, and deepfakes in schools.”
“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 […]
“As Generative AI becomes more capable, it’s getting harder and harder to spot AI-generated images. Powerful image generation models like Flux (available through Grok on X) and Midjourney can create very believable images in many styles. Can you tell real from fake in the ten pairs of images below?”
“We decided to hold a contest between ChatGPT and me, to see who could write — or “write” — a better beach read. I thought going head-to-head with the machine would give us real answers about what A.I. is and isn’t currently capable of and, of course, how big a threat it is to human […]
“After the Zoom interview ended, he received an email from Otter, a popular transcription program that uses AI to convert audio to text, sharing a both a transcript and the original audio. The recording had continued after Hill left the Zoom interview, capturing the researchers’ thoughts, Hill said.”
“Attempts at lying and falsification have existed since the beginning of history. In fact, history may depend on them. Part of our job is to sort through error, and try to find truth, if possible.”
“If generative AI is truly good at one thing, it is tearing down the facade of a given practice and revealing how shoddy the foundation was to begin with. ChatGPT made many in education confront writing practices that produced rote outcomes. AI detection has done the same with how little work we put into teaching […]
“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 […]
“If you debate with an AI, they are 87% more likely to persuade you to their assigned viewpoint than if you debate with an average human. GPT-4 helps people reappraise a difficult emotional situation better than 85% of humans, beating human advice-givers on the effectiveness, novelty, and empathy of their reappraisal. GPT-4 generates startup ideas […]
“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.”
“I said “delve” was overused by ChatGPT compared to the internet at large. But there’s one part of the internet where “delve” is a much more common word: the African web. In Nigeria, “delve” is much more frequently used in business English than it is in England or the US. So the workers training their […]
“The basic structure of Bjarnason’s essay is to compare the methodology of mentalists (aka mind-readers, entertainers and con artists who pretend to read minds) with what happens in LLMs. The essence is distilled in a pair of summary tables.”
“In this post, I’ll discuss some of my personal objections to AI detection tools, and explore a new piece of research that once again proves AI detection tools don’t work.”
“Using artificial intelligence, middle and high school students have fabricated explicit images of female classmates and shared the doctored pictures.”
“Rather than just being worried about one giant AI apocalypse, we need to worry about the many small catastrophes that AI can bring. Unimaginative or stressed leaders may decide to use these new tools for surveillance and for layoffs. Educators may decide to use AI in ways that leave some students behind. And those are […]
“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…
“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 […]
“To cool these data centers, huge quantities of water are used – about 16 oz. for every 5-50 AI prompts… Generating an image can be especially energy-intensive, requiring as much power as fully charging your smartphone.”
“Hong Kong police said at a press conference Friday that the employee at the unnamed firm’s Hong Kong branch initially suspected phishing when he received an email last month purporting to be from the company’s UK-based chief financial officer, CNN reported. However, after attending a video conference and seeing convincing deepfakes of the CFO and […]
“In short, it’s capitalism versus humanity.”
“In the hands of anonymous internet users, A.I. tools can create waves of harassing and racist material.”
“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.”
“Those findings come from a survey of 2700 AI researchers who have recently published work at six of the top AI conferences – the largest such survey to date.”
“The 10-minute post was one of more than 4,500 videos in an unusually large network of YouTube channels spreading pro-China and anti-U.S. narratives.. Some of the videos used artificially generated avatars or voice-overs, making the campaign the first influence operation known to the institute to pair A.I. voices with video essays.”
“Here are three dispatches highlighting the various ways that candidates — and crucially, third parties — seem ready to use AI as America chooses its next president.”
“To gauge the magnitude of biases in generative AI, Bloomberg used Stable Diffusion to generate thousands of images related to job titles and crime. We prompted the text-to-image model to create representations of workers for 14 jobs — 300 images each for seven jobs that are typically considered “high-paying” in the US and seven that […]
“If we follow the tale, this all might be ok in the end. Eucrates learned his lesson — when asked to perform the magic, he says, “if it once becomes a water carrier… we shall be obliged to let the house be flooded with the water that is poured in!” His lesson is much like […]
“I am going to make several arguments. One, that there are two different kinds of trust—interpersonal trust and social trust—and that we regularly confuse them. Two, that the confusion will increase with artificial intelligence. We will make a fundamental category error. We will think of AIs as friends when they’re really just services. Three, that […]
“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 […]
“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 […]
“The problems start with organizational policy. Many companies have banned ChatGPT use, often because of legal concerns that remain somewhat vague, based on uncertainty over the technology and regulatory worries… But these bans are having a big effect… they are causing employees to bring their phones into work and access AI from personal devices. While […]