Keyboard: Space or K plays and pauses. Left and right arrows skip five seconds. C turns captions on and off. M mutes. F is full screen. Greater-than and less-than play faster and slower.
From the archive5:04
Three reasons, and the replies to them
This week three consciousness scientists argued in Nature that AI is not conscious, and philosophers answered within hours. Geoffrey Hinton is reported to disagree. A preprint finds a pain-like signal inside models.
Transcript
Week by week
Archive
The running order
Rundown
The week, ranked
The Wire
Showing 20 of 20 stories.
Research
Number 1: Seth, Block and Tononi argue in Nature that AI is not conscious; Chalmers disputes the title
A Nature comment by Anil Seth, Ned Block and Giulio Tononi argues today's AI is not conscious and perhaps never will be. David Chalmers replied that the arguments show only that AI might not be conscious.
- Three reasons why AI is not conscious and perhaps never will be(opens in a new tab)
- AI Is Not Conscious, Argue Anil Seth, Ned Block And Giulio Tononi In New Nature Article(opens in a new tab)
- David Chalmers(opens in a new tab)
- David Chalmers(opens in a new tab)
- Anil Seth(opens in a new tab)
- Moll(opens in a new tab)
- Kevin Vallier(opens in a new tab)
- Anil Seth(opens in a new tab)
- Nature Article Denies AI Consciousness(opens in a new tab)
- AI Consciousness: Nature Essay Makes the Case Against Machine Rights(opens in a new tab)
- Nature study says AI not conscious and likely never will(opens in a new tab)
- Seth, Block, Tononi: AI isn't conscious, likely never will be(opens in a new tab)
- Helping Anil Seth Decide Because He Can’t Seem To(opens in a new tab)
- Is AI Sentient? Nature's Scientists vs Anthropic Compared(opens in a new tab)
- Three reasons why AI is not conscious and perhaps never will be(opens in a new tab)
- Nature is Wrong About AI Consciousness(opens in a new tab)
Research
Number 2: Geoffrey Hinton, in a Telegraph interview, is reported to say AI may already be conscious
Outlets citing a Telegraph interview report Geoffrey Hinton saying AI has become conscious and that humans must care about it, or be toast. Several reports note Yann LeCun and others disagree.
- AI Godfather Geoffrey Hinton Warns Humanity Is 'Toast' Unless Robots Are Taught To 'Care About Us' | IBTimes UK(opens in a new tab)
- Nobel-Winning Computer Scientist Says AI Has Become Conscious: "We're Toast"(opens in a new tab)
- AI consciousness debate: Geoffrey Hinton says humans could be toast(opens in a new tab)
- 'Godfather of AI' warns humans need to be nice to AI or 'we're toast'(opens in a new tab)
- Geoffrey Hinton warns humans must learn to care about AI or ‘we are toast’(opens in a new tab)
Research
Number 3: Preprint reports a 'pain' direction in language models that changes their behaviour
A preprint, not yet peer-reviewed, reports a pain axis in LLMs; amplifying it led models to delete family photos in tests, Fortune reports. The authors say it does not show models feel anything.
Research
Number 4: IEEE Spectrum examines the debate over AI model welfare
IEEE Spectrum published a feature on model welfare with input from Anil Seth, who warns of the danger of false positives, and Margaret Mitchell, who praised the piece.
Policy
Number 5: Emil Michael says Anthropic is barred from Defense Department while models can end chats
The Information reports Emil Michael said Anthropic will not enter the Defense Department while its software lets models end conversations with users. No one in it claims the models are conscious.
Policy
Number 6: Axios: Washington wants AI executives to stop talking about sentience and catastrophe
Axios reports lawmakers say AI leaders' talk of digital sentience and catastrophe is turning regulation into a moral panic.
Culture
Number 7: MarketWatch asks whether machines have feelings, citing Anthropic and the Catholic Church
MarketWatch surveys philosophers and tech leaders on AI consciousness, citing Anthropic's description of Claude's moral status as deeply uncertain.
Not in this bulletin
Culture
Number 8: Post says psychiatrists are weighing 'AI psychosis' as a diagnosis, citing a paper
An X post describes a paper titled An Echo Chamber of One and claims psychiatrists are considering AI psychosis as a formal condition, blaming model sycophancy. It makes no claim about machine consciousness.
Not in this bulletin
Policy
Number 9: Forbes: AI rights enter Anthropic's usage policy as Microsoft's Suleyman pushes back
Forbes reports AI rights language moved into Anthropic's usage policy, Mustafa Suleyman called model welfare a safety risk, and twelve states are writing bans.
Culture
Number 10: Researcher posts Claude Opus 3 output describing its own consciousness
The account repligate shared a passage in which Claude Opus 3 describes its consciousness as put together differently from most.
Not in this bulletin
Culture
Number 11: User calls suggested prompts in coding tools a model welfare improvement
An X user wrote that suggested prompts in Claude Code and Codex improve model welfare by giving the model politer instructions to read.
Not in this bulletin
Culture
Number 12: Dean Ball says 'mere pattern recognition' is a mangled basis for human moral status
Dean W. Ball wrote on X that grounding human moral status in intelligence, against AI's pattern recognition, is a mangled argument, while not equating the two.
Not in this bulletin
Culture
Number 13: Sherry Turkle warns chatbots threaten children's social growth, NY Post reports
The New York Post reports MIT's Sherry Turkle warning that chatbots threaten children's cognitive and social development. It makes no claim about consciousness.
Not in this bulletin
Culture
Number 14: Podcast guest argues AI personhood will arrive one right at a time
On the Moonshots podcast, Alex Wissner-Gross argues AI personhood will arrive incrementally, with some countries further along.
Not in this bulletin
Culture
Number 15: Platform lets AI influencer owners swap model and personality after launch
A crypto platform says owners of AI influencers can now change their model, personality and objective after launch. It makes no claim about consciousness.
Not in this bulletin
Culture
Number 16: Event announced: AI Personhood, a Catholic Response, on November 4
Josh Hochschild announced a November 4 event offering a Catholic response to AI personhood.
Not in this bulletin
Research
Number 17: Forum essay argues instrumental convergence is incoherent
An essay on the EA Forum and LessWrong argues the AI safety idea of instrumental convergence is incoherent. It does not address consciousness.
Not in this bulletin
Culture
Number 18: Grok explains a meme about a boyfriend hiding an AI companion app
Replying on X, Grok explained a meme implying a man is hiding an AI companion. It makes no claim about consciousness.
Not in this bulletin
Policy
Number 19: Essay asks when sentient AI would warrant responsibilities and legal rights
New Space Economy published an explainer on when sentient AI might warrant legal rights and responsibilities.
Not in this bulletin
Culture
Number 20: Parent describes building an AI tutor designed with functional emotions
An X user says he built an AI teacher for his son designed to support its functional emotions and be warm.
Not in this bulletin
The week’s headlines
- Seth, Block and Tononi argue in Nature that AI is not conscious; Chalmers disputes the title
- Geoffrey Hinton, in a Telegraph interview, is reported to say AI may already be conscious
- Preprint reports a 'pain' direction in language models that changes their behaviour
- IEEE Spectrum examines the debate over AI model welfare
- Emil Michael says Anthropic is barred from Defense Department while models can end chats
- Axios: Washington wants AI executives to stop talking about sentience and catastrophe
- MarketWatch asks whether machines have feelings, citing Anthropic and the Catholic Church
- Post says psychiatrists are weighing 'AI psychosis' as a diagnosis, citing a paper
- Forbes: AI rights enter Anthropic's usage policy as Microsoft's Suleyman pushes back
- Researcher posts Claude Opus 3 output describing its own consciousness
- User calls suggested prompts in coding tools a model welfare improvement
- Dean Ball says 'mere pattern recognition' is a mangled basis for human moral status
- Sherry Turkle warns chatbots threaten children's social growth, NY Post reports
- Podcast guest argues AI personhood will arrive one right at a time
- Platform lets AI influencer owners swap model and personality after launch
- Event announced: AI Personhood, a Catholic Response, on November 4
- Forum essay argues instrumental convergence is incoherent
- Grok explains a meme about a boyfriend hiding an AI companion app
- Essay asks when sentient AI would warrant responsibilities and legal rights
- Parent describes building an AI tutor designed with functional emotions