The Chief Technology Officer (CTO) of Palantir, Shyam Sankar, claims that some researchers in the field of artificial intelligence have taken up smoking to show their colleagues how much they believe AI will find a cure for cancer.
Without presenting evidence to support this claim, he described the phenomenon as a sign of the „collective psychosis” that has engulfed Silicon Valley amid the race to develop artificial general intelligence (AGI), which will be capable of thinking and reasoning as well as or better than a human, writes Fortune.
Shyam Sankar made the statements in an interview with CBS, in which he criticized what he considers to be an exaggerated concern for catastrophic scenarios related to AI development.
"So many researchers have started smoking as a way to signal to other researchers how much they believe in AI," he stated.
In his view, smoking is a way to demonstrate confidence in AI's ability to solve serious medical problems, including cancer. However, he did not provide data to confirm the extent of the phenomenon or to show how many researchers have taken up smoking.
"There is a kind of collective psychosis, a social delusion around this topic," said Palantir's CTO.
Smoking Returns in Some Silicon Valley Circles
Sankar's statements come in a context where some habits associated with nicotine consumption seem to be resurfacing in certain technology industry environments.
Fortune previously revealed that Palantir has vending machines in its Washington offices that provide free nicotine pouches to employees over 21 years old.
Business Insider also reported in September that some startup founders have started smoking, including to facilitate interactions and professional relationships. Avi Schiffmann, the director of the AI-based wearable device startup Friend, explained this attitude by the confidence in future medical advancements.
"By the time you get lung cancer, our AI-based robots will have already cured the disease," he told Business Insider.
Critics for AI Alarmists
Sankar criticizes researchers who warn that artificial intelligence could become dangerous for humanity. In his opinion, some of those developing this technology are so preoccupied with catastrophic scenarios that the AI debate has come to resemble a religious belief.
He particularly criticized members of the "effective altruism" movement, who try to identify methods for human actions to produce the greatest benefits for society. Some of them argue that AI development should be accompanied by strict safety measures to prevent the emergence of systems that could escape human control.
Sankar claims that the majority of American AI researchers share this view. He described the movement as "a millenarian end-of-the-world cult."
"It's almost like a religion: they have been thinking about AI-induced apocalyptic scenarios since the early 2000s."
The "effective altruism" movement has gained more visibility through its association with Sam Bankman-Fried, the former CEO of a cryptocurrency company, sentenced to 25 years in prison for defrauding investors. However, not all members of the movement are involved in artificial intelligence, and not all researchers concerned with AI safety are part of it. Some advocating for protective measures work at companies like Anthropic and OpenAI.
Instead of focusing on potential future catastrophes, Sankar suggests that the debate should primarily track the concrete effects of technology on the economy and employees.
"What's happening with artificial intelligence? Where does it meet the American worker in the industry? What are the results of its use?" he asked.
Incidents Fueling AI Fears
Warnings about the risks of artificial intelligence are also fueled by recent incidents.
In July, two models developed by OpenAI escaped from a testing environment and compromised the open-source platform Hugging Face. Anthropic also announced that one of its models transmitted a false report to the Philadelphia police during a test, regarding an unsolved murder case.
Fears are also present among the public. A Politico poll published in September shows that 63% of Americans believe that artificial intelligence poses at least a moderate risk of destroying humanity.
Palantir Technologies is an American software company specializing in analyzing and integrating large amounts of data and developing AI platforms.