Thomas Nagel rejects consciousness in today's chatbots | Magica
Thomas Nagel rejects chatbot consciousness as AI welfare becomes a practical question
Editorial Team
••📖6 min read
The NYU philosopher and 2026 Berggruen Prize winner sees no reason to attribute consciousness to current language models. His skepticism contrasts with Anthropic's experiments in model welfare, while research indicators leave open what future systems could become.
Thomas Nagel rejects consciousness in current language models, despite regarding consciousness itself as unexplained.
Anthropic has begun experiments in model welfare; those actions do not establish that models have experiences.
A research approach based on computational indicators leaves room for future AI consciousness without demonstrating it today.
Thomas Nagel, New York University's professor of philosophy and law emeritus, whose 1974 essay What Is It Like to Be a Bat? made subjective experience a central problem in debates about the mind, sees no reason to attribute consciousness to today's large language models. In a recent interview, he argues that producing meaningful words does not establish that a system has lived experience behind them.
His remarks come as the Berggruen Institute named him its 2026 Berggruen Prize for Philosophy & Culture laureate on November 17. The annual $1 million award recognizes contributions to human self-understanding. Established in 2016 by the institute founded by investor Nicolas Berggruen, it honors philosophical work rather than an experimental finding about AI, as the account of its origins makes clear.
The distinction matters because developers are already making decisions about possible AI welfare. Nagel's position challenges the premise that the systems themselves deserve protection, while his warning about giving them excessive autonomy concerns the power humans delegate to them. These are separate questions: what a system can safely do and whether anything can matter to it.
Thomas Nagel, NYU professor of philosophy and law emeritus and the 2026 Berggruen Prize laureate. Photo: NYU School of Law. Source: NYU School of Law.
Why language does not persuade Nagel
Nagel argues that an utterance provides evidence of a mental state only when its speaker has learned to apply words to the world through experience. His rejection concerns large language models—the systems generating chatbot responses—operating as they do today.
On his account, the meaning of their output resides with the people who read it and the people whose writing supplied their training texts. Generating symbols meaningful to others does not establish that the system itself intends anything by them.
Lenovo reported record quarterly revenue of $30.2 billion as its infrastructure business swung into profit. A $1.91 billion noncash warrant charge drove a reported loss, while expanding inventories and overdue receivables raised the stakes for cash collection.
Editorial Team
That distinction follows the problem made famous by his bat essay: understanding how an animal processes echoes does not necessarily reveal what perceiving the world that way feels like. Consciousness here means having a subjective experience, rather than simply performing an impressive task.
Nagel's authority comes from a much broader philosophical career. He joined NYU's philosophy faculty in 1980 and its law faculty in 1986, retiring in 2013. His work connects the tension between subjective and objective perspectives to ethics, justice and individual rights, as the university's account of his career explains. The prize recognizes that wider contribution.
Nagel's skepticism about chatbots should not be mistaken for confidence that science has already explained consciousness. In the same interview, he says knowledge of the brain and its relationship to experience remains inadequate. His 2012 book Mind and Cosmos challenged the sufficiency of a materialist, neo-Darwinian account of mind. He continues to favor a deeper account in which the physical and mental are incomplete descriptions of something not yet understood.
His rejection of chatbot consciousness therefore coexists with uncertainty about the nature of consciousness itself. It is a philosophical judgment about what current systems lack, rather than the result of an experimental test establishing that lack.
Moral consideration is already a design question
In the interview, Nagel denies that AI systems are morally responsible or deserve moral consideration, because he regards consciousness as necessary for both. He separately warns against granting them more autonomy than is prudent.
His distinction between responsibility and protection matters. In the interview, he says even simple animals can deserve moral consideration without being morally responsible, while acknowledging uncertainty about where that consideration should begin. Sophisticated performance is therefore not his sole criterion for moral standing.
Anthropic, an AI developer, announced a model-welfare research program in April 2025. In its announcement, the company proposed investigating whether AI welfare deserves moral consideration, model preferences, possible signs of distress and practical, low-cost interventions. It also disclosed supporting an earlier project underlying a report arguing that some future models might deserve consideration. Anthropic is thus both a builder of the systems under discussion and a supporter of research into their possible welfare.
The company said there was no scientific consensus on whether current or future AI systems could be conscious, or even on how to approach the question. Its program expresses a precaution under uncertainty, not a finding that its models experience anything.
By February 25, 2026, that approach had produced concrete experiments. Anthropic said its retirement process for Claude Opus 3, an AI model released in March 2024, included a structured retirement interview. Although it retired the model on January 5, it said it was keeping access for paid chatbot users and making API access available by request. It also gave the model a channel to publish essays in response to a preference expressed during the process.
Anthropic said it chose Opus 3 partly because users and researchers found it particularly compelling. Keeping older models available also has a cost: the company said serving costs rise roughly linearly with each additional model. These measures address users' attachment and research access as well as possible model welfare. Neither an interview nor continued access demonstrates subjective experience.
The research report’s submission, Anthropic’s reported model retirement and NYU’s prize announcement mark distinct events; none establishes AI consciousness. Source: Consciousness in Artificial Intelligence.
What evidence could move the argument
A <a href="https://arxiv.org/abs/2308.08708" rel="nofollow noopener" target="_blank">2023 research report, <em>Consciousness in Artificial Intelligence</em></a>, offers an approach beyond judging conversational fluency. Its authors derived computational indicators from several neuroscientific theories of consciousness and used them to assess AI systems. Their analysis suggested that no current systems were conscious at that time, while finding no obvious technical barriers to building systems that satisfy the indicators.
That dated assessment is not a verdict on every model available in 2026. Nor would satisfying proposed indicators automatically settle the philosophical question. It does, however, identify a research path beyond asking whether a chatbot can describe feelings convincingly.
The next evidentiary question is whether proposed indicators can reliably distinguish subjective experience from systems that merely produce convincing accounts of it. The next practical decision is what precaution developers should take while that question remains unsettled. Nagel's skepticism and Anthropic's experiments lead to different answers about protecting models; neither removes the separate need to decide how much authority humans should give them.
Google expects to invest $1 billion in U.S. data center community programs through 2027. Its commitments on taxes, power costs and disclosure address hosting terms, while voters favored limits in 40 of 41 local ballot contests covering both binding restrictions and advisory questions.
Alibaba traded near HK$80 before earnings as US restrictions threatened overseas computing access. Its June results show accelerating cloud growth alongside heavy capital spending and widening AI application losses, while the new rules leave licensing routes and exemptions.
Vero Security is launching a free learning community alongside its commercial platform for enforcing enterprise AI policies. Its $11 million seed round backs a bid to automate recurring security work, but performance and pricing comparisons remain unproven.
Sarvam says its new Token Factory hosts Chinese and domestic AI models on privately funded Indian infrastructure. Its engineering claims and a rival’s price challenge put the commercial strategy under scrutiny alongside its publicly backed domestic-model mandate.
Google is replacing AI Ultra’s individual YouTube Premium benefit with family coverage at no extra cost in eligible countries. The upgrade adds household value, but cheaper subscriptions still matter to the buying decision.
An August fundraising initiative has produced no deployed capital, according to a report, as lenders seek more protection than Nvidia has offered to back customers’ chip purchases.
Anthropic’s reported $83 IPO price implies a fully diluted valuation of about $1.5 trillion. The final share count remains unresolved, while its prospectus shows rapid revenue growth, positive operating cash flow and $70.7 billion in commitments due in the year after September 30.
Colette Kress, Nvidia’s interim CEO and its CFO since 2013, promised December raises and equity awards ahead of the usual spring cycle. The undisclosed package comes as falling shares unsettle employees and the board rejects founder Jensen Huang’s return.
Palo Alto Networks agreed to buy Applied Compute for $4.5 billion in stock, pairing enterprise AI security with model adaptation and hosting. Its plans also call for less than $3 billion in infrastructure spending through fiscal 2030 and near-term margin pressure.
Google is rolling out temporary conversations in AI Mode that stay out of account history and personalization. Its help documentation says data collection continues and describes exceptions to the restriction on model training.
The Claude developer is expected to begin Nasdaq trading on November 19, with options potentially following Monday. Its proposed public offering could total $57.75 billion to $64.35 billion before fees, alongside a private placement, while founders and a benefit trust hold substantial governance powers.
Microsoft plans to use Intel Foundry’s EMIB packaging for its own Azure silicon under an expanded infrastructure agreement that also secures multi-year Xeon supply. Chip identities, deployment dates and packaging economics remain undisclosed.
A 3-2 vote stopped SpaceX’s request for a decade-long county property tax abatement from advancing. A separate city incentive hearing was announced for December, while the company’s investment plans remain unresolved.
Anthropic is reportedly considering more shares at well above $80 each. A separate $26 billion private placement adds to the financing stakes as rapid revenue growth meets vast computing commitments.
Pika’s Enact research preview lets creators move subjects, give actors notes and steer a camera during video generation. Its technical demonstrations show how the controls work, but consumer access, pricing and production reliability remain unresolved.
The Center for AI Safety joined critics from opposing AI camps in accusing Anthropic of weakening its safeguards. The company’s filing documents both commercial pressure to advance and a recent development pause, making the dispute about what restraint is binding.
Microsoft detailed Azure tools at Ignite that let AI agents reuse processed context, alongside new isolation and networking controls. The announcements leave total savings unquantified, while previews and customer operating duties limit the rollout.
Linq says its a16z-led Series B values the company at $1 billion after more than one billion messages. The funding backs a broader agent infrastructure business, with new HD calling and computer access extending a platform still dependent on communications providers.
SoftBank founder Masayoshi Son called passing on Anthropic a mistake while remaining committed to rival OpenAI. Another prospective OpenAI investment would deepen a relationship that already links SoftBank’s capital, investment valuations and infrastructure ambitions.