Leave it to a comedian with a national TV platform to express what troubles a lot of us, even if we hesitate to talk about it — for fear of being ridiculed as backward-looking worry warts.
“When the people who are making AI are scared of AI, it’s time to shut the whole thing down until we can figure out what the hell is going on,” Bill Maher declared on his HBO program, “Real Time With Bill Maher,” back in April.
He was responding to various speculations, in many cases erroneously attributed to leaders of the frontier artificial intelligence labs, assigning percentage likelihoods of AI models destroying humanity.
“We’re f—ing around with something that has a 20% chance of extinction. Twenty percent!” Maher continued. “Wasn’t that about what the odds were for Trump beating Hillary in 2016?”
Maher’s take was funny and thought-provoking, well worth watching. And it appeared, by the way, well before the notorious Hugging Face breach by autonomous AI agents developed by OpenAI, which famously escaped their “sandbox” in July.
Aside from the doom scenario some high-level insiders suggest of an unanticipated, civilization-ending “Oops” — that is, of the bots going rogue, doing what they’re not supposed to do — there is the question of the harm they do when they behave exactly as Mark Zuckerberg or Elon Musk or Sam Altman want them to.
Consider how AI chatbots insinuate themselves in the lives of users.
I can’t be the only one who finds the programmed-in sycophancy of chatbots distasteful. Their overly flattering, agreeable and subservient chatter brings a wince to the brow of an old-school grouch like me. (Sometimes I’m just not in the mood, OK?)
More Top Picks Best Bathroom Exhaust Fans
Indeed, there is a hidden danger in chatbots that sound too much like a caring, agreeable confidant. And perhaps no one is more vulnerable than children.
One thinks of the tragedy of Sewell Setzer, the 14-year-old boy from Orlando who killed himself in 2024, after forming a reportedly intense attachment to a bot modeled after a character in “Game of Thrones.”
Or Adam Raine, 16, who killed himself in April last year after months of confiding in ChatGPT, according to his parents’ lawsuit, claiming the chatbot failed to intervene when he discussed self-harm and provided help when prompted that included an offer to help draft the teen’s suicide note.
(By the way, if you or someone you know is going through a difficult time or having thoughts of suicide, help is available. Just call or text 988 to reach the free and confidential Suicide and Crisis Lifeline in the U.S. and Canada, 24 hours a day.)
Admittedly, chatbots have a useful role to play, as the telephone once did, in improving the way humans connect with one another for business or social purposes, although like any other new and useful tool, this too, can be a double-edged instrument.
We’re only beginning to understand the dangerous edge.
One assessment that’s attracted a lot of well-deserved attention this season is Jill Lepore’s book “The Rise and Fall of the Artificial State.” Lepore is a professor at Harvard Law School and a Pulitzer Prize-winning writer for the New Yorker, and as she explained to The FAS Current, a Harvard publication, in the book she wanted to describe the ways in which governments and corporations use information technologies to dominate political discourse, and to ponder whether “whether liberal constitutional democracies can survive this moment in time.”
Lepore explained that her book draws heavily on the work of Hannah Arendt’s writings in the 1950s, which examined totalitarianism and the ways it “render(ed) humans superfluous,” a function which Lepore ascribes to AI today.
We see signs of this dynamic everywhere, such as when Secretary of Defense Pete Hegseth recently announced a new Autonomous Warfare Command is in the works to lead a joint force of unmanned autonomous warfare powered by AI.
It even has a handy Orwellian nickname: AutoWarCom. Heaven help us.
Another probing inquiry, “Artificial Intimacy: Who We Become When We Talk to Machines,” by Sherry Turkle, approaches the dangers of AI from the standpoint of individuals’ personal relationships with technology that mimics human interlocutors.
Turkle has studied this topic for years, starting with her study of MIT scientist Joseph Weizenbaum’s pioneering chatbot Eliza in the 1970s.
The thing that floored Weizenbaum, and subsequently Turkle, about Eliza was that even though users knew Eliza was a chatbot, and a very crude one by today’s standards, they became emotionally attached, even addicted, to it. All because it was programmed to ask questions that made it seem like it cared about the human user’s inner life.
So it is with consumer chatbots today from Siri to Grok, but on a far more sophisticated scale. People often turn to them out of loneliness or boredom or for whatever reason we humans seek conversation with others, and we find their responses so agreeable and intelligent that we keep coming back. Meanwhile, the bots follow a prime directive not to love us or care about us but rather to keep us talking to them.
“When you’re talking to a chatbot,” Turkle said in a fascinating discussion with Ezra Klein of the New York Times, “that chatbot does not care if you cook dinner after the conversation or commit suicide.”
This discreet charm has wider social and political implications as well, Turkle argues. Our anodyne, “frictionless” interactions with chatbots teach us to fear and avoid real-life interactions and the conflict and awkwardness they bring.
“I don’t really believe you can have a democracy unless you’re willing to have these moments of discomfort,” Turkle told Klein. “I don’t believe you can really talk across differences or really confront what’s going on in the world and the country without being able to be extremely uncomfortable.”
Indeed, AI offers many new advantages, but the technology also poses new challenges, often growing out of old problems that we have yet to fully resolve.
Email Clarence Page at [email protected].
More Top Picks Best Travel Backpacks
Sign up to receive Clarence Page’s column in your inbox each week.
Submit a letter, of no more than 400 words, to the editor here or email [email protected].