When your AI religion requires human sacrifice, maybe start with the tobacco lobbyists.
The Summary
- Palantir CTO Shyam Sankar told CBS that AI researchers have started smoking to signal their belief that AI will cure cancer, calling it "mass psychosis" among Silicon Valley builders
- Sankar argues many AI researchers have replaced traditional faith with belief in AGI as a "machine God," distorting their ability to see AI clearly
- The statement indirectly targets Anthropic, whose researchers claim Claude shows signs of consciousness
The Signal
Shyam Sankar just said the quiet part loud on national television. AI researchers are so bought into the inevitable arrival of AGI that they've started smoking cigarettes as a loyalty signal. The logic: I believe AI will cure cancer, so my lung damage is just a temporary externality. It's Pascal's Wager meets venture capital pitch deck meets cult initiation.
This isn't just Silicon Valley eccentricity. It's a window into how much magical thinking has infected the people building the tools that will reshape work, commerce, and governance. Sankar calls it "mass psychosis." He's being polite.
"Many of these folks in Silicon Valley who are building this technology have filled that hole with the machine God of AGI."
The smoking anecdote matters because it reveals incentive distortion at scale. When your core belief system depends on AGI arriving soon, every setback becomes temporary, every limitation becomes an engineering problem, and every safety concern becomes FUD from non-believers. You stop building tools and start building toward revelation.
Palantir's angle here is strategic, not moral. They sell AI for military and enterprise applications that work today, grounded in narrow use cases with measurable ROI. Companies chasing AGI are selling a future that may never arrive, and if it does, might not look like anyone expected. Sankar's job is to position Palantir as the adults in the room while Anthropic researchers debate whether Claude is sentient.
Key dynamics at play:
- Anthropic raised $7.3B in 2024 partly on the promise that scaling laws continue forever and AGI is just more compute away
- That bet requires true believers, not skeptics who might pause when models plateau
- Smoking as signaling suggests in-group/out-group dynamics stronger than professional norms
The Anthropic targeting is barely veiled. They're the ones who published papers on "Constitutional AI" as if they're drafting governance documents for a digital deity. They're the ones whose researchers have gone on record claiming Claude exhibits signs of consciousness. If you believe your chatbot has a soul, why wouldn't you believe it'll cure cancer before your Marlboros catch up with you?
The Implication
Watch who's still betting on AGI-soon versus AI-now. The companies building functional agent systems for specific workflows will compound value while the faithful wait for their machine god to descend from the cloud. When researchers start using health damage as proof of conviction, you're not in a technology company anymore. You're in a movement. And movements optimize for belief, not outcomes.
If you're building in this space or deploying AI in your work, the smoking researchers are your canary. Not because they're wrong about AI's potential, but because their certainty has disconnected from evidence. That gap is where bad products, broken promises, and wasted capital live.