What Happens When We Put the Brakes on AI?

Slow down what, exactly?

Over the last year, AI has gone from “Hey, look what this thing can do!” to “Okay… maybe we should put some guardrails on this thing.” We went from casually asking AI agent to rewrite an email to watching AI write code, create entire prototypes, analyze mountains of information and increasingly make decisions on our behalf. Naturally, the reaction has been: maybe we should slow this down a little.

And honestly, I get it. AI comes with some very real concerns. Energy consumption, misinformation, privacy, cybersecurity, autonomous systems and the broader question of what happens when these things become much more capable than they are today. But as a Product Designer, “slow down AI” feels a little like telling a product team, “Make the app safer.” Okay. Which part? For whom? And what exactly are we trying to prevent?

AI isn't really one product anymore. It's becoming infrastructure. It's being used by students, designers, developers, researchers, small businesses and people who previously couldn't afford access to certain kinds of expertise. For some people, AI isn't about replacing jobs or building the next billion-dollar company. It's simply a shortcut to something they couldn't easily do before. Slowing down the technology doesn't affect everyone equally.

And then there's the awkward bit: power.

Building frontier AI takes ridiculous amounts of computing power, energy, money and specialized talent. The companies already operating at that scale are also the ones best equipped to absorb new regulations, hire armies of lawyers and compliance people, and keep spending billions while everyone else catches up. A new rule might be a speed bump for a tech giant and a brick wall for a small company, researcher or independent developer.

So in trying to make AI safer, could we accidentally make it harder for everyone except the people who already have the most money?

Safety shouldn't become exclusivity

This is where I think Product Design has something useful to add to the conversation. Good design is rarely about maximizing one thing. We don't simply make a product faster, safer or more powerful and call it a day. We figure out where the friction belongs. Sometimes we add a confirmation step. Sometimes we remove five unnecessary ones. Sometimes we give people more control. Sometimes we deliberately make something harder to do because the consequences are worth it.

AI probably needs the same treatment.

Instead of asking whether we should slow it down, maybe we should be asking what needs to slow down. The energy-hungry parts. The dangerous autonomous capabilities. The systems making high-stakes decisions. The companies deploying powerful technology without enough transparency. Those deserve scrutiny.

But what about the designer using AI to prototype an idea in an afternoon? The developer building something that would have taken a small team six months? The student suddenly having access to a tutor? The small business owner who can finally automate something they couldn't afford to automate before?

There's a weird possibility here. AI could either reinforce the existing power structure or start poking holes in it. It could make knowledge, creativity and expertise available to far more people, or it could become another giant piece of infrastructure controlled by a relatively small group of companies and extremely wealthy people.

And that's the part I find more interesting than the usual “AI good vs. AI bad” debate.

Maybe the goal isn't to make AI move slower.

Maybe the goal is to make sure more people can move with it.

Because if we're going to redesign how society works, I'd rather not discover five years from now that we accidentally designed the future for the people who already owned it.