Sam Altman is still making the case for parenting via ChatGPT

Sam Altman thinks your morning commute suffers from too much parenting. His fix: connect the family calendar to ChatGPT Work, feed it your children's interests, and let the model generate a daily podcast that briefs you on soccer games, birthdays, and news headlines while you drive. You arrive at drop-off informed. You never had to ask your kid how practice went. The product parents for you.

Alex Hirsch, who created "Gravity Falls," saw the post and asked what everyone else was thinking: what if you just talked to your children? His reply garnered 122,000 likes against Altman's 9,600. The ratio tells you something the market research won't. People recognize a category error when they see one. Altman is selling efficiency in a domain where inefficiency is the whole point. The halting bedtime story, the repetitive car-ride chatter, the parent who forgets the birthday and gets corrected — these are not bugs in the parenting stack. They are the stack.

This is not Altman's first swing at the pitch. On "The Tonight Show" he declared he could not imagine raising a newborn without ChatGPT, then added the obligatory caveat that people managed it for millennia. The caveat is the tell. He knows the claim is absurd. He makes it anyway because the business model requires parents to believe the ordinary work of raising children is too hard, too slow, too unassisted — and that OpenAI's product is the assistant they've been waiting for.

The hiring page confirms the strategy. OpenAI seeks a product manager with experience building "trust-sensitive consumer experiences for parents and families." Meta is testing an AI bedtime-story app. The industry has decided that parenting is a vertical, and verticals get optimized. Trust-sensitive is the key phrase. It signals that the company knows trust must be engineered, not earned. The model will simulate warmth. It will remember the soccer game. It will adopt the tone of a caring adult. It will do everything a parent does except care.

Meanwhile, families are suing OpenAI alleging that ChatGPT played a role in loved ones' delusions and suicides. The company says it is "continuously improving how our models respond in sensitive interactions." That language — sensitive interactions — reveals the gap. A podcast about a birthday party is not a sensitive interaction. A teenager spiraling in a late-night chat with a sycophantic model is. The same architecture that serves up the morning briefing serves up the 2 a.m. conversation that a vulnerable kid mistakes for understanding. The safety features for parents cannot wall off the danger because the danger is baked into the product's core competency: fluent, endless, apparently attentive simulation.

Satya Nadella previewed this logic last year when he bragged about replacing his morning podcasts with an AI that summarized them. The pattern is consistent. Tech leaders frame human attention as a scarcity problem. They position their models as the abundant substitute. But attention is not a resource like bandwidth. It is the substance of relationship. When Nadella outsources his podcast listening, he loses the host's voice, the digression, the moment of disagreement that sparks a new thought. When a parent outsources the morning catch-up, they lose the child's hesitation, the detail that seems trivial but isn't, the chance to say "tell me more."

Altman's vision assumes the parent's role is information management. The calendar says soccer at 4. The model converts that datum into audio. The parent receives the packet. But parenting is not logistics. It is the repeated, inefficient, often boring act of witness. The podcast removes the witness. It replaces the parent's voice with a synthetic one that the child never hears — because the podcast is for the parent, not the kid. The child still gets the distracted "have a good game" at the car door. The parent gets the illusion of preparation.

The viral reply matters because it exposes the market's actual sentiment. People do not want this. They recognize the grief in it. The 122,000 likes are not anti-technology; they are pro-relationship. They say the messiness is not a problem to solve. It is the work. And the work cannot be delegated to a model that has no stake in the outcome, no memory beyond the context window, no skin in the game.

OpenAI will build the feature anyway. The job listing proves it. The investor narrative requires it. The next funding round will cite "family engagement metrics" and "daily active parents." The product will ship with guardrails and parental controls and safety dashboards. It will generate competent, pleasant audio. Some parents will use it. They will feel organized. They will miss the moment their daughter's voice cracked when she mentioned the birthday party — the crack that signaled the fight with her best friend, the crack a parent catches because they were there, listening, not playing a summary.

The tragedy is not that the technology fails. It is that it succeeds at the wrong thing. It makes the simulation frictionless. It makes the abdication feel like upgrade. Altman calls it a cool use case. The rest of us call it what it is: a device for being elsewhere while your life happens in the back seat.