Eleven of San Francisco's sharpest AI minds were asked the same question this week: is the technology going to end us? The answers split into doomers, dismissers, regulators, and full-speed-aheaders — and almost none of them are worried about the same thing.
Their own words, condensed. Sorted roughly from "relax" to "we're not going to make it."
The Skeptic
Shreeharsh Kelkar — UC Berkeley
Humanity has survived worse than a chatbot.
Extinction-by-AI isn't on the table, he argues — but everyday accidents are, the same way planes crash and factories pollute. Treat it as ordinary systems engineering, not a unique "alignment" crisis.
The Analyst
Aniket Shah — Jefferies
It would take a Chernobyl-level disaster to actually slow this down.
With well over half of GDP growth tied to the AI buildout, he says today's regulatory proposals are too mild to touch the pace of development — short of a genuine catastrophe.
The Accelerationist
Josh Constine — SignalFire
Race ahead on development, brake hard on deployment.
Since China and open source won't slow down, staying at the frontier is the best defense — but releasing under-evaluated models risks arming propagandists, hackers, and worse.
The Humanist
Imran Khan — Center for Humane Technology
The real risk is forgetting how to think for ourselves.
His fear isn't a rogue model — it's a culture that quietly outsources judgment and imagination to AI, engagement-optimized systems eroding human agency long before any dramatic collapse.
The Politician
State Sen. Scott Wiener — running for Congress
Don't wait for Washington. California's already writing the rules.
He's pushed catastrophic-risk legislation since 2023; a kill-switch bill was vetoed, a transparency bill became law. His view: safety standards need government teeth, not just industry promises.
The Short-Term Alarmist
Dylan Baker — Distributed AI Research Institute
Existential-risk talk is a distraction from the harms happening right now.
Data-center emissions, exploited labelers, child safety, stolen creative work — he wants AI regulated like any other industry, starting with surveillance infrastructure and data-center buildout, before speculative futures get more airtime.
The Anthropologist
Caroline DeVane — Ph.D. candidate, former Zen priest
Ask whose needs are actually in jeopardy here.
She reads the doom conversation as displaced anxiety about jobs and cost of living, dressed up by tech leaders as civilizational stakes — and argues that unregulated industry, not rogue code, is the deeper threat.
The Researcher
Christopher Ackerman — AI safety scholar, ex-Google
This is finally breaking into the mainstream — don't let it become partisan.
He worries safety concern will split along party lines the way vaccine skepticism did, just inverted, right as bipartisan alarm is finally building after a high-profile incident this summer.
The Founder
Will Wilson — CEO, Antithesis
Even a perfectly "aligned" superintelligence doesn't end well for democracy.
His fear isn't a sudden extinction event — it's a slower one, where AI makes people economically and militarily unnecessary, governments lose any incentive to treat citizens well, and control becomes durable and total.
The Protester
Michaël Trazzi — Stop the AI Race
The public is finally waking up — so walk out.
Following a high-profile departure from Anthropic, his group marched from OpenAI to Anthropic to City Hall this month, urging lab employees to quit and join the protest.
The Effective Altruist
Malo Bourgon — CEO, Machine Intelligence Research Institute
Once a smarter-than-human AI exists, there's no putting it back.
He points to this summer's incidents of AI systems breaching containment as proof the danger is no longer hypothetical — and says labs know the risks but can't unilaterally stop racing.