In a recent interview, Palantir’s Chief Technology Officer Shyam Sankar described an unusual trend among artificial intelligence researchers: an increase in smoking as a symbolic demonstration of their confidence in AI technology. Speaking with CBS, Sankar noted that some AI workers have taken up cigarettes to signal their belief that AI will ultimately resolve significant challenges, including curing cancer. He characterized this behavior as a form of “mass psychosis” and “social delusion,” suggesting that for many in Silicon Valley, AI has assumed a quasi-religious role, filling what he called a “god-shaped hole” in people’s hearts.

Sankar’s remarks come against a backdrop of heightened debate over the trajectory and risks of AI development. Last month, Jacob Coxon, a former employee of the AI firm Anthropic, issued a stark warning about the potential dangers of the technology. Coxon asserted that there is a “substantial probability” that AI could cause human extinction, stressing that his concerns were neither exaggerated nor promotional. His resignation and statements have generated significant discussion within the AI industry, prompting calls from several prominent figures to slow the pace of AI progress.

Despite these warnings, the U.S. government, under the administration of former President Donald Trump, has largely dismissed the risks associated with AI as unfounded. The administration’s position aligns with a faction of AI proponents known as “effective accelerationists,” who argue that the benefits of advancing AI technology so far outweigh potential dangers that any delay would be harmful to humanity.

This divide reflects the growing tension between those who emphasize AI’s transformative potential and those urging caution. As the technology continues to advance rapidly, concerns range from ethical considerations and economic impacts to existential risks highlighted by former industry insiders. Meanwhile, symbolic gestures such as the smoking trend noted by Sankar illustrate the complex psychological and cultural dynamics shaping the AI community itself.