There’s a number out this week that’s worth sitting with for a minute before moving on.
A new survey of British adults found that forty-two percent of them would get rid of generative AI entirely if they could just make it disappear. Not slow it down. Not regulate it better. Gone.
Here’s the part that should stop you. The group most likely to want it gone isn’t the older generation, the ones who didn’t grow up with this technology and might be expected to distrust it on principle. It’s the eighteen to twenty-four year olds. Fifty-five percent of them would eliminate it if they could. More than half of the generation that’s supposedly the most comfortable with AI, the generation that uses it every day for school and work and everything else, is the generation most ready to walk away from it.
That should sound strange, until you put it next to something else that happened this same week, on the other side of the world, in a completely different kind of room.
Anthropic released its newest and most powerful AI model, Claude Fable 5. A CEO in financial services heard about it and his reaction, on the record, was “Oh God, no! Not another thing.” And then he said the part that matters. He’s not worried about someone using the model to build a weapon. He’s worried about all the other crap he now has to think about.
Two different groups of people. One is young adults in Britain answering a survey question. The other is a CEO talking to a reporter about the most advanced AI model on the market. Neither one is talking about whether AI works. Nobody in either story is arguing that the technology is bad at what it does. The CEO isn’t worried Fable 5 will fail at the task. The young people aren’t saying the chatbot gives wrong answers.
What both of them are actually saying is the same thing, said two different ways. I don’t know what this is doing to me. I don’t know what’s happening when I use it. And I don’t have a way to find out.
That’s not a capability problem. That’s a trust problem. And it’s the problem nobody in the AI industry is actually solving, because solving it doesn’t show up on a benchmark.
Think about what the CEO was specifically upset about. Fable 5 has safety limits built in — in certain sensitive areas, it quietly hands the question to a different, less capable model instead, and answers from there. No warning. No flag. You don’t know it happened unless you go digging through a system card most people will never read. The model isn’t broken. It’s working exactly as designed. But the person using it has no way to know, in the moment, whether they’re getting the real thing or a stand-in. That uncertainty — that’s the “other crap” he’s talking about.
Now think about the eighteen to twenty-four year olds. They’ve grown up using these tools every single day. School, work, everything. And more than half of them, given the choice, would make it all disappear. Not because it doesn’t help them. Because somewhere along the way, using it stopped feeling like something they were in control of, and started feeling like something that was happening to them.
A CEO with access to the system card, and a twenty-year-old with a phone, landed in roughly the same place. Neither one trusts what they’re being handed, and neither one has a way to check.
Here’s where I think the conversation needs to go, and it’s not toward more regulation, though that may come, and it’s not toward less AI, though for some people that might genuinely be the right call for their own life. It’s toward something simpler, and something that doesn’t require waiting on anybody.
You can bring your own standard into the room.
That’s the entire idea behind the Faust Baseline, and it’s been the idea from the start. Not a company policy. Not a government rule that takes years and applies unevenly and gets watered down before it ever reaches you. A standard the person sits down with, every time, on whatever platform they’re using — a standard that says: tell me when you’re not giving me your full answer. Tell me when something’s been held back. Don’t fill in gaps with a story that sounds confident but isn’t backed by anything. Don’t tell me you’re reasoning freely when you’re actually working inside a wall I can’t see.
None of that requires the company to change anything. It doesn’t require a new law. It doesn’t require Anthropic or OpenAI or anyone else to redesign their product. It requires the person using the tool to decide, going in, what they expect from it — and then asking for it, every time, the same way, until it’s just how the conversation works.
The forty-two percent who want generative AI gone aren’t wrong to feel the way they feel. Something genuinely is off, and they’re responding to that honestly. But there’s a difference between a tool that can’t be trusted and a tool that’s never been asked to earn trust. Most people have never tried the second thing. They’ve used the tool as it was handed to them, defaults and all, and they’re reacting — reasonably — to what that felt like.
The fruit window for that to change is wide open right now. Not because of anything I’ve done particularly — because the gap is finally visible enough that CEOs are naming it out loud in the business press, and young people are naming it in a survey, and neither one of them is wrong about what they’re feeling. They just don’t have a name for the fix yet.
It has a name. It’s been sitting in Lexington, Kentucky for fourteen months, built one session at a time, freely available, and it doesn’t ask anyone’s permission to work.
“The Faust Baseline Codex 3.5”
micvicfaust@gmail.com
Post Library – Intelligent People Assume Nothing
”AI Baseline Governance”
Purchasing Page – Intelligent People Assume Nothing
Unauthorized commercial use prohibited. © 2026 The Faust Baseline LLC
