If you follow AI closely, you probably have a doom scenario running somewhere in your head. Most people who work in it do. It’s not a good way to live, and it doesn’t help you make better decisions either. This is the argument for getting out of it, written down so it’s there when it’s needed.
Why it’s a real problem
AI will be smarter than us. That’s not really the debate anymore. A neuron fires a few hundred times a second, a transistor switches billions of times, and the models already have more connections and more knowledge than any one person. The debate is what happens after.
What worries me is that the people closest to it are the most worried. Researchers inside the labs can see the scaling plots, and by their own account the plots show a lot of room left. The best proxy is maths. Two years ago these models couldn’t do basic arithmetic. This month one of them solved a Millennium Prize problem, something no human had managed since 2003. Anthropic says Claude went from leading 1% of its research in February to 26% in August. Wherever models show early competence today, expect them to be superhuman soon.
Nobody slows down on their own either. These companies aren’t racing to own a niche. When AI gets smarter it gets smarter in every industry at once, so whoever gets there first wins all of them. The risks that come with that are real: the same biology that cures cancer can build a weapon, a system smarter than us might not want what we want, and even in the good case a K-shaped economy hits people hard before it helps them. Senior researchers at the labs put the chance of a catastrophic outcome somewhere around 10 to 25%, and they have no reason to exaggerate.
For me it stops being abstract at one point. I come from a traditional Vietnamese family, and there’s a script. Finish uni, get a job, build a family, have kids, grow old. Every step was a given until the kids step. That one is shaky right now, and that’s when the doom scenario stops being a debate and becomes your life.
How to stay optimistic anyway
I can’t make the risk go away. What I have is six things I keep coming back to, and they work whether or not you buy any of the above.
-
Separate what you can’t control from what you can. I can’t control what’s on the labs’ plots. I can control how I spend the next few years, how carefully I build, and whether I add my voice to the pressure on these companies and governments. Right now the labs say they’d slow down if everyone else did, governments say the labs should build the guardrails, and both point at the other country. Public pressure is the lever that’s left, and it’s one you can actually pull.
-
It’s a probability, not a verdict. Even the pessimists say 10 to 25%. That’s 75 to 90% not doom, and nobody plans a life around the tail. The same uncertainty that allows the worst case allows the best one, where we cure most diseases and lift a few billion people out of poverty. Nobody knows which one we get. Not the researchers, not the CEOs, not me.
-
We’ve stood at the edge before. In 1983 Stanislav Petrov saw a missile alert, decided it was a false alarm and didn’t launch. It’s not proof of anything, but humanity has a track record of standing at the edge and not stepping off.
-
This is still one of the best times in human history to be alive. For most of history there were no vaccines, no clean water, no soft beds. A king a hundred years ago couldn’t buy what you have in your pocket. We take all of it for granted. I try not to.
-
Having kids is a bet, not a verdict either. I’ve landed here: it’s a bet placed with the best information I have. If I raise a child well and give them what they need, that’s responsible enough. Running your life on a certainty you don’t have isn’t caution, it’s just a different kind of guess.
-
The things models are bad at are the things they can’t be trained on. Trust. Relationships. Someone willing to take responsibility. People still need a person, not a random computer, to be accountable to them.
So that’s how I stay optimistic. Not by believing the future is good. By deciding to act like my actions matter, because they might.