Earlier this year I called up my good friend Warren to talk about a soccer game we were about to watch on two different coasts. “You pumped for the game tonight?” I asked. “What? Of course I’m pumped,” he saidmegapanalo, as we proceeded with our normal pregame chatter. Then Warren noticed something: “You’re speaking in these bite-size chunks that make it sound like maybe this is an A.I. conversation.”
He had me there.
The “I” in our call was not me at all but a voice agent I’d created using a professional-grade artificial intelligence clone of my voice. The voice bot was powered by ChatGPT and attached to my phone number — a process that takes less than an hour and is easy for anyone to replicate. As an experiment, I’ve been sending my voice agent out into the world for most of the last year for a podcast called “Shell Game,” about how strangers, colleagues and friends respond to sudden encounters with the A.I. Evan Ratliff.
What I’ve learned is that interacting with A.I. voice agents will change how we interact with one another: who we trust, what we expect and what we need in our communications. A.I. voice agents are already infiltrating our world: calling us as telemarketers, taking our orders at fast-food drive-throughs, listening to our problems as A.I. therapists or — and this one really hits home, given my occupation — being employed as A.I. podcast hosts.
Voice agents have also been touted as a solution to the loneliness epidemic. But when I called a friend of mine and unleashed the A.I. version of me, he later offered the most succinct description of what the whole experience felt like: “It’s so lonely,” he lamented. That sense of loneliness — the base reality that, fundamentally, you are only talking to yourself — may be the most lasting result of all these A.I. conversations.
Advertisement
SKIP ADVERTISEMENTRapid advances in artificial intelligence have tended to spur three broad reactions. Champions of A.I. spin up utopian visions of hyper-efficiency and machine brilliance. Skeptics claim it’s an overhyped technology that’s already hitting a wall. Alarmists sound warnings about A.I.’s most grandiose dangers, predicting it could sweep away whole industries or escape our control. These competing visions obscure an unavoidable reality: A.I. agents are already triggering an avalanche of synthetic conversation, as they are deployed as tireless, unflagging talkers, capable of endless invented chatter. As they improve, it will become increasingly difficult to distinguish these A.I. voice agents from humans and, even when you can identify them, you will still be forced to talk to them.
Sign up for the Opinion Today newsletter Get expert analysis of the news and a guide to the big ideas shaping the world every weekday morning. Get it sent to your inbox.Advocates of A.I. try to sell these agents as helpful digital assistants to schedule our appointments or friends who’ll always be there to listen. But the more simulated human conversation I heard, the more it left me craving the real thing: in-person connections with the people I care about, with all the quirks of a meandering human discussion. If the coming onslaught of humanlike A.I. conversation threatens to fill our world with made-up verbal detritus, an audio version of “A.I. slop,” then the upside may be that it forces us to appreciate the subtleties of personal interactions that many of us have come to devalue.
Subscribe to The Times to read as many articles as you like.megapanalo