Some AI Researchers Are Taking Up Smoking as a 'Flex', Believing AI Will Cure Cancer, Palantir CTO Says

Some artificial intelligence researchers are reportedly taking up smoking to signal their confidence that AI will eventually cure cancer, according to Palantir Technologies chief technology officer Shyam Sankar. But he offered no names or evidence to substantiate the claim, leaving the anecdote as a striking illustration of his concerns about excessive faith in the technology.

Speaking on CBS's Face the Nation on 11 October, Sankar described the behaviour as a sign of what he called 'mass psychosis' and 'social delusion' around AI. He argued that some people in Silicon Valley treat artificial general intelligence (AGI), a hypothetical system capable of performing a broad range of intellectual tasks at or beyond human capabilities, with almost religious devotion.

The more important question is not whether the smoking anecdote represents a widespread trend, which has not been established, but how the industry can separate confidence in AI's future from evidence of what it can deliver today.

Sankar's Criticism of AI's 'Machine God'

Sankar invoked philosopher Blaise Pascal's idea of a 'God-shaped hole' in human hearts to argue that some technology developers have placed their faith in AGI. 'I think it's important to understand that many of these folks in Silicon Valley who are building this technology have filled that hole with the machine god of AGI,' he said.

His criticism cuts in two directions. Sankar challenged people who believe AI could bring catastrophic consequences, while also warning against utopian expectations that the technology will solve major human problems without sufficient evidence.

He said technology companies have two responsibilities: demonstrating that AI can strengthen national prosperity and security, and engineering safe products from the outset. That shifts the debate from sweeping predictions towards measurable outcomes and accountability.

What AI Can Actually Do for Cancer Research

AI already has potential applications in cancer research, including analysing complex datasets, identifying biological patterns, and supporting clinical decision-making. The US National Cancer Institute outlines how AI may contribute to cancer research and care, but such applications do not establish that the technology can cure cancer.

A research tool may help scientists identify a promising treatment target or analyse data more efficiently. That is different from demonstrating that a therapy is safe and effective in patients. Treatments must still undergo appropriate testing and regulatory review.

Cancer also encompasses many diseases with different biological characteristics. Progress against one type does not mean a universal cure is imminent, and a breakthrough in research does not automatically translate into a treatment that reaches patients. This distinction matters when public expectations are shaped by predictions about what future AI systems might achieve rather than evidence from current medical studies.

A Real-World Test of AI's Medical Promise

A collaboration announced by Tempus AI, Moderna, and Merck on 7 October illustrates the more concrete work taking place in personalised cancer treatment. The companies are preparing for the potential commercialisation of intismeran autogene, an individualised neoantigen therapy being evaluated with KEYTRUDA (pembrolizumab) in patients with completely resected stage IIB–IV melanoma and other cancer types.

Tempus will coordinate the collection and transfer of tumour tissue and blood samples for next-generation sequencing. Subject to regulatory approvals, it will also provide sequencing services to support the therapy's design and manufacturing.

The treatment remains under evaluation. The collaboration is not evidence of a cancer cure, but it shows how specialised data services can support the development of targeted therapies while clinical evidence determines their value.

Why Confidence Must Not Replace Evidence

Sankar's smoking anecdote is provocative because it imagines people discounting present-day health risks on the assumption that future technology will eventually remove them. However, his remarks do not establish that researchers broadly hold this view or that smoking has become a widespread practice in AI circles.

Smoking remains a major cause of preventable illness and death, and no AI breakthrough has eliminated its established health risks. The broader lesson is that optimism about AI should be judged against demonstrated results, not promises.

The technology may accelerate parts of cancer research, but its eventual medical impact will depend on validated discoveries, safe treatments, and evidence that patients benefit. For an industry often defined by bold predictions, that is a more meaningful measure of progress than the strength of anyone's belief in what AI might one day achieve.