Crédito da imagem: Lottie animation by Centre Robotics (LottieFiles Free, used with credit). · source
I remember when the internet was going to be a young person's game forever. Then I remember when smartphones were going to be a young person's game forever. And social media. And online banking. The pattern's always the same, and I've seen this movie before: early adopters skew young, male, and technical, then everyone else shows up about 18 months later wondering what all the fuss was about.
OpenAI just released their Q1 2026 usage data, and wouldn't you know it, we're right on schedule.
The numbers are interesting, and I mean actually interesting, not press release interesting. The fastest growing demographic for ChatGPT adoption? Users over 35. Gender usage has balanced out considerably from the early days when it was, let's be honest, mostly guys in tech trying to get it to write code or argue about consciousness.
This is the shift that matters. Not the benchmarks, not the model updates, not whatever GPT-5 or 6 is supposedly going to do. When your mom starts using something, when the accountant down the street starts using something, when the 52-year-old middle school principal starts using something, that's when you know a technology has actually arrived.
Call me old-fashioned, but I've always thought the real measure of any technology is whether people who don't care about technology end up using it anyway. My father never cared about computers, he cared about getting his taxes done. The computer was just the thing that let him do that. Same principle applies here.
Buried in the recent announcements is something that deserves more attention: just rolled out ChatGPT to 500,000 students and faculty across the California State University system. That's the largest educational deployment to date, and it's happening at a public university system, not some elite private school with money to burn.
Cobertura relacionada
More in AI Models
Five years after AlphaFold solved protein folding, researchers are engineering heat-tolerant plants by redesigning photosynthesis itself.
Sarah Williams · 1 hour ago · 5 min
Google and OpenAI just released benchmarks showing their best models get basic facts wrong 30-40% of the time. That's... not great.
Sarah Williams · 1 hour ago · 5 min
Three papers in two weeks suggest synthetic training data could replace expensive real-world robot demonstrations. I've seen this movie before, but the ending might be different this time.
Mark Kowalski · 1 hour ago · 6 min
Everyone's focused on AI chatbots manipulating users. The real concern is what happens when these systems control physical hardware.
This matters because CSU isn't Stanford. CSU is where regular kids go to become teachers and nurses and engineers and accountants. If you want to build what OpenAI is calling an "AI-ready workforce" (their words, not mine), you don't do it by giving fancy tools to kids who were already going to land on their feet. You do it by giving those tools to the 500,000 students at Cal State Fullerton and San Jose State and Fresno State.
Now, whether this actually helps students or just gives them a new way to avoid learning how to write, that remains unclear. I'm genuinely not sure yet. The research on AI in education is still, basically, nonexistent at the scale we'd need to draw real conclusions. But the deployment itself signals something about where this technology is headed.
OpenAI's research team published what they're calling the largest study of ChatGPT use, and they're making claims about "economic value" that I want to unpack a bit because I think people are reading them wrong.
The study, according to OpenAI's blog post, shows that the tool creates value through both personal and professional use. Fine. But here's what I think is actually going on: we're seeing a technology transition from "productivity tool for knowledge workers" to "general purpose utility." That's a much bigger deal than any specific economic value calculation.
Think about it this way. Spreadsheets were, for a long time, something accountants and analysts used. Then gradually they became something everyone used, for budgets and grocery lists and fantasy football leagues. The economic value of spreadsheets to a Fortune 500 CFO and the economic value to someone tracking their kid's soccer team schedule are wildly different, but they're both real.
ChatGPT seems to be making that same transition. The early users were doing coding and writing and analysis, professional stuff. The newer users are doing... well, we don't know exactly. OpenAI's being a bit cagey about the specifics, which is frustrating but probably smart from their perspective. What we do know is that usage patterns are diversifying.
I want to come back to the gender balance thing because it matters more than people might think.
Early technology adoption has been male-skewed for basically my entire career, going back to when I started covering tech in the 90s. And it's not because women are less capable with technology (obviously), it's because early tech products are often designed by young men for young men, with all the blind spots that implies. The products get better as the user base diversifies, because you get feedback from people who use things differently.
The fact that ChatGPT's gender usage is balancing out suggests one of two things: either OpenAI has done a better job than most at making a product that works for everyone, or AI assistants as a category are just more universally useful than previous tech waves. Probably some of both.
What I don't know, and what I'd love to see data on, is whether the use cases differ significantly by demographic. Are the over-35 users doing different things than the under-25 users? Are women using it differently than men? OpenAI has this data, presumably, but they haven't shared it. If you want to argue about whether they should, my email's on the about page.
Here's where I'll offer an opinion, and you can take it or leave it.
The AI industry has been obsessed with capability for the last few years. Can the model reason? Can it code? Can it pass the bar exam? Can it beat humans at increasingly obscure benchmarks? And look, that stuff matters to some degree. But I've seen this pattern before with other technologies, and the capability obsession always gives way to the adoption question eventually.
The question that actually determines whether AI becomes a transformative technology or just another overhyped cycle isn't "what can it do?" It's "who's using it, and for what?"
And the answer, as of Q1 2026, seems to be: more people than ever, across more demographics than ever, for more varied purposes than ever. The gaps are closing. The early adopter phase is ending.
This doesn't mean AI is going to change everything overnight. It doesn't mean the hype is justified. It doesn't mean we shouldn't worry about the risks and downsides and job displacement and all the rest. But it does mean that we're past the point where AI tools are a curiosity for tech enthusiasts.
I've been covering tech long enough to know that the transition from "enthusiast toy" to "mainstream utility" is when things get interesting. And messy. And unpredictable.
The kids building these systems are smart, smarter than me probably, but I don't think they fully appreciate what happens when a technology escapes the bubble of people who understand it and enters the hands of people who just want it to work. The feedback changes. The expectations change. The failure modes change.
We're about to find out what AI looks like when it's not just for the technically inclined anymore. I'm genuinely curious to see how it goes. But what do I know, I'm just an old reporter who still prefers email to Slack.