What Did the Palantir CTO Say?
The idea that AI researchers might take up smoking because they expect future technology to cure cancer sounds almost satirical. Yet it became a talking point after Sankar discussed the behaviour during an interview with CBS News journalist Margaret Brennan.
Sankar claimed that researchers had begun smoking as a way of signalling their confidence in AI. He connected the alleged behaviour to the belief that future AI systems would be capable of curing cancer, suggesting that some people were willing to accept present-day risks because they expected technological advances to eliminate them later.
He described the broader phenomenon as a form of social delusion surrounding AI.
However, the distinction between a reported anecdote and a verified trend is important. Sankar did not name the researchers, quantify the behaviour or explain what evidence supported his statement. The CBS transcript confirms that he made the claim, not that the alleged trend has been independently established.
The remarks are therefore best understood as an example of the debate over AI optimism, rather than proof that smoking has become widespread among AI researchers.
Why AI Enthusiasts Are Betting on Future Medical Breakthroughs
The argument draws on a genuine development: AI is becoming an increasingly important tool in biomedical research.
Researchers are investigating how machine-learning systems can identify potential drug candidates, analyse complex biological data, improve medical imaging and help scientists understand disease mechanisms. These applications could make some stages of medical research faster and more efficient.
Cancer research is one area attracting significant attention. AI systems can help researchers examine patterns that may be difficult to identify through conventional analysis alone. They may also assist in developing hypotheses about how tumours behave and how treatments could be improved.
Nevertheless, identifying a promising target or predicting how a drug might behave is not the same as demonstrating that a treatment works in humans. Potential therapies must still undergo appropriate laboratory studies, clinical trials and regulatory review. Their effectiveness and safety must be established through evidence. Cancer also encompasses many diseases with different causes, biological characteristics and treatment responses. Progress against one type does not automatically translate into a cure for every other type.
Consequently, the expectation that AI will improve cancer research is scientifically plausible. Treating a future cure as guaranteed is a much stronger claim.
The Problem With Treating AI Predictions as Certainties
Sankar's remarks raise a broader question about how people respond to expectations surrounding transformative technology. When a technology is expected to deliver extraordinary benefits, people may begin treating those benefits as inevitable rather than conditional. In the case of AI, predictions about medical breakthroughs can become intertwined with beliefs about artificial general intelligence and the possibility of machines surpassing human capabilities across many tasks.
Such expectations can encourage investment and research. However, they can also blur the distinction between demonstrated capabilities and speculative outcomes. There is a further concern when anticipated future progress is used to justify accepting risks today. Even if AI eventually produces major advances in cancer treatment, the timing, scope and accessibility of those advances remain uncertain. Smoking causes cancer and other serious diseases. No credible promise about future AI capabilities removes the established risks of tobacco use. The same principle applies beyond healthcare: anticipated technological progress should not replace evidence-based decisions about present-day safety.
What This Means for AI Development
The controversy illustrates why responsible AI discussions need to consider both the technology's potential and the limitations of current evidence. AI could contribute to major advances in medicine, but those advances depend on high-quality data, rigorous scientific testing, appropriate oversight and the ability to translate research findings into effective treatments. Public claims about AI's future should therefore be evaluated according to their supporting evidence, not simply the confidence of the people making them.
For journalists, the episode also demonstrates the importance of distinguishing a provocative statement from an independently verified development. The smoking anecdote is attention-grabbing, but the more consequential issue is how expectations about AI influence real-world behaviour and public understanding of science. Ultimately, AI may help researchers develop better cancer treatments and improve patient outcomes. Whether it will cure particular cancers, how soon that might happen and who will benefit are questions that research must answer.
Until then, confidence in AI's potential should not be confused with certainty about its future achievements.
