Our Dystopia May Be Sam Altman’s Fantasy
AI CEOs love AI agents. But what about personal agency for human beings?
A few days ago, I posted a video clip of Sam Altman predicting what the worst case scenario would be for the proliferation of AI in society.
The background of the clip was as such: onstage at a conference, in a sort of fireside chat, Altman is asked about AI safety mechanisms. His response is that if we overreact and implement too many AI safety mechanisms that centralize AI under one control, humans will live “in service” of AI, and that we will have no privacy or agency. Essentially, he says, we’ll live in a surveillance state (of course I, Sam Altman, won’t surveil you—but the government or one of my competitors might.)
Several people criticized my reposting of this clip, claiming that I was “fear-mongering” and that I was sharing Altman’s response “out of context.” These respondents assured me—with a bizarre sort of moral authority for people who were defending AI oligarchs—that Altman was warning us of what could happen, not that he wanted it to happen.
He was merely responding to a question about the most terrifying AI scenario, they said—not promoting it.
Putting aside the fact that Altman has blown through all kinds of safety guardrails in his relentless pursuit of building OpenAI—for example, allowing advertisers to mine ChatGPT conversations, as well as building a relationship with the Department of War—I found it fascinating that people were obsessed with honoring the specific interview context of Altman’s response.
Because the context of Sam Altman—reported in several investigations, and through many testimonies—is that he is a power-hungry person, without a lot of regard for ethics, willing to steamroll anyone in his way. (As one board member put it to New Yorker writer Ronan Farrow: “I don’t think Sam is the guy who should have his finger on the button.”)
So with that context, it makes it difficult for me to believe that Altman was just innocently responding to a question about AI’s worst case scenario. It makes it impossible to think he is just a concerned individual wanting to ring the alarm about the surveillance state.
Because even if his response was stated in response to a posed question, Altman most certainly—in building what he built to begin with—must have already been considering people being spied upon or losing their critical thinking skills as a potential outcome. It didn’t terrify him enough, however, not to pursue it.
And say you could plausibly believe or argue that Altman was trying to warn us of becoming slaves to the AI. That he was, in good faith, trying to propose the worst case in order to prevent it from happening. That he wasn’t, like all of his AI competitor CEOs, trying to position himself as the one genius positioned to save us all.
Even if all of that were the case, how much stock should we put into any person who is posing as a savior to the very problem that they had a hand in creating?
And the idea of people losing their personal agency to AI—as he lays out in the worst case—is not actually a Black Mirror hypothetical. It’s already happening.
Case in point: the other day, Altman posted, in earnest, a “positive” use case for ChatGPT. The use case he proposed was: kids and parents can use ChatGPT to create podcasts about their days—feeding calendar information about soccer games and music lessons into the model so that it can create a machine-narrated version of storytime.
It’s hard to imagine something that strips a person of their agency more than reducing children’s experiences into a calendar-prompted machine-narrated podcast. Beyond limiting open and free-flowing communication between parents and their young children, this dynamic also zaps the creativity that so naturally springs from kids.
So yes, context is important. And that context may very well be that Sam Altman doesn’t find the worst case scenario so dystopian after all.
What else we’re paying attention to…
New York sued Kalshi, alleging it is an “illegal gambling operation.” The attorney general said platforms like Kalshi, Coinbase, and Gemini can “encourage problem gambling, including by people under age 21, and endanger people’s financial, emotional and physical health.”
Even babies are not free from surveillance tech. From a deep-dive into the baby monitoring industry: The comedian John Mulaney recently riffed on his toddler son’s sleep setup, saying his bedroom looked like a “C.I.A. black site,” with a blackout tent over his crib. “I don’t know why we have 24-hour CCTV footage of my son’s crib, he’s not going anywhere — he sleeps in a bag,” he said. “We then have three sound machines going at full blast. And because we read it once on Instagram, we keep his nursery at 68 degrees. Any warmer and your infant will die automatically.”
With the rise of vibe-coding, startups are seeing million-dollar valuations with just one employee.
Scammers in Cambodia were using ChatGPT “to create fake online personalities, translate conversations, draft messages to potential victims, and produce promotional material. Some users also generated forged passports, legal notices, stock purchase confirmations, and fake cryptocurrency trading interfaces.”
Uber’s Strategy for Fighting Sexual Assault Suits: ‘What Were You Wearing?’ Gross.
A former operations specialist in the tech industry describes the challenges of reentering tech in a first-person essay, I spent a decade being loyal to tech in Ireland. Then AI took my job: “To make matters worse, the interviewers rarely help you show your true potential. While some were kind, many others treated me with condescension. They looked down on me from a position of absolute power. They treated the interview like a cold interrogation rather than a mutual partnership.”