I decided to hit the pavement at Salesforce’s gigantic Dreamforce conference to ask tech-minded people about AI’s catastrophic risk. Turns out, it’s not all AI gloom and doom — that’s only if you’re looking online.
Last week, Anthropic researcher Jacob Coxon announced he was resigning from the company, saying that AI labs are “gambling with our lives.” Debates about an AI-led apocalypse proliferated across the internet, and the leaders of Anthropic and OpenAI posted at length about new safety measures.
Around two-thirds of Americans now think there’s at least a moderate chance that AI will destroy humanity, a Politico poll taken this week found.
It can be hard to tell, though. On Tuesday at Dreamforce, I asked some attendees about their so-called “p(doom),” or where they place the probability that AI will cause a major catastrophe. As I navigated a flood of attendees in suits and lanyards, people told me they were worried about AI but not about an apocalypse. Instead, they’re concerned about other issues, such as impacts on labor, mental health, and the environment.
Dean Levita, a local tech worker attending the conference, said he was mostly unconcerned about an AI-driven extinction. Being aware about catastrophic risks isn’t the same as thinking they’re right around the corner, he said, likening earthquake warnings from his lifetime in the Bay Area to the discussion about AI risk.
“It’s cautionary, it’s not imminent,” he said.
Dreamforce attendees brought up data centers
Henry Sandles and Sara Jamison had come to learn more about Salesforce since they regularly use the software at their nonprofit, which preserves a historic railroad. When I asked about the fears of extinction, Sandles cracked a grin and said, “We’ve been here before.”
He said he trusts that smart people will shut down any AI before it wreaks real-world havoc.
“I just can’t see computers being smarter than humans,” Sandles told me.
Jamison said she is far more concerned about the environmental impacts of data centers. She’s happy to see AI automating boring work, she said, though she doesn’t want to see it used for art.
Ashleigh Kaye — the name she uses with friends, which she shared for fear of retaliation by her employer — is visiting from Oklahoma for the conference. She also raised concerns about data centers, especially regarding electricity prices. She told me she’s not worried about any kind of mass extinction from AI. Instead, she thinks it’s dangerous to weave the tech into so many applications that could then exert more control over us.
“It’s hard to get away from AI,” she said. “Everything’s AI-driven.”
Some say AI can impact mental health
Deep Karia, in town representing his Indian software development company at the conference, told me that his worries for AI cluster around privacy and mental health. He said a disaster stemming from AI is more likely to be because someone grew dependent on the tech and had some kind of psychotic break.
Karia is “not at all” worried about AI posing a risk of extinction.
“We know that’s not going to happen,” he said.
He said that, amid the discourse around AI’s risks, he’s reassured by the belief that humans will remain in control. That could change, Coxon and other AI researchers said this week, though for now, it’s a common rebuttal to the existential fears.
Likewise, Soumya Nimmu, a law school student passing by the conference, said she’s been getting sent videos about the AI fears. It’s a distraction, she said.
She’s more troubled by deepfakes running rampant in schools and the use of AI in military technologies. Nimmu said that the here-and-now issues of AI should get more attention, like people getting traumatized labeling horrific content for AI’s training: “Why is that not what we’re talking about?”
AI extinction fears still loom
That’s not to say AI extinction fears weren’t present. I ran into none other than Valerie Sizemore, a co-leader of the Stop AI nonprofit, on my way to the conference. Her group fears “human extinction within a matter of years,” and advocates for a ban on developing superintelligent AI. As of this month, Stop AI’s ideas feel markedly less fringe: Sen. Bernie Sanders has proposed a ban until a federal regulator makes safety rules.
Sizemore said it’s been “really nice” to see AI risk discussed on the news and hear from family and friends who now better understand her yearslong crusade against the tech. She spent the day handing out flyers next to a drawing of the Grim Reaper that said: “AI – Either we stop it, or it stops us!”
Have a tip? Contact this reporter via email at scouncil@businessinsider.com, or over text, Signal, Telegram, or WhatsApp at 415-757-8198. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.
