I use AI every day now. It drafts content, helps me think through pricing, and even helped write parts of this blog. Forty years ago, that sentence would have sounded like science fiction. In 1983, WarGames asked whether a computer could take control of something it shouldn’t. I didn’t expect to be asking a version of that question about my own business.
So I’ll ask it straight. Should we actually be worried?
Three different fears, usually lumped together
The first is already here. Jobs shifting, misinformation spreading, scams getting more convincing, a handful of companies gaining outsized influence. Real, and happening now, no speculation required.
The second is autonomy. Picture AI systems running the electrical grid, financial markets, transportation networks, all deeply interconnected. The risk there isn’t AI turning malicious. It’s AI making decisions faster than humans can review or reverse them. That’s an engineering and governance problem, not a movie plot.
The third is what people usually mean when they say “taking over.” A system that develops goals in conflict with ours and gains the capability to pursue them on its own. Could that happen? Nobody actually knows. Serious researchers land on both sides of that question, and I’m not going to pretend I have the answer they don’t.
What WarGames actually got right
The lesson of that movie was never that computers turn evil. It was that people shouldn’t hand over decisions of real consequence without understanding the system or keeping a hand on the wheel. That part hasn’t aged at all.
Current AI doesn’t want anything. It has no ambition, no survival instinct, no plan it’s running between our conversations. The actual concern researchers focus on is narrower: future systems that operate continuously, use tools, write their own code, connect to real infrastructure. Hand a system like that broad authority with a poorly defined goal, and unexpected behavior gets a lot more likely. That’s the real work happening in AI safety right now. Building safeguards, limiting authority, testing behavior, and making sure humans can still intervene.
Where I actually land, running a small business on this stuff
Electricity transformed everything it touched. It also introduced electrocution and house fires. Nobody decided electricity was a mistake. We built insulation, breakers, and codes instead.
I expect AI follows the same arc. Not because the risks are identical, but because every technology this size forces new rules into existence eventually.
Here’s the version of that lesson I actually apply. I don’t use AI to replace judgment, mine or my clients’. I use it to extend what I can get to in a day.
At DADO, AI might help me explore a campaign idea, organize information, or improve a first draft. It doesn’t decide what fits a client’s brand, what should be produced, or what I ultimately recommend. That’s still my job.
It helps me draft faster, think through more options, and catch things I might have missed on my own. The final decision still belongs to me. So does the responsibility for the outcome.
That’s probably the healthiest way to use any of this, whether you’re worried about superintelligence or just wondering if you should let a tool write your next email. Stay in the loop. Keep asking whether the thing you built still does what you meant it to do.
I don’t have the big answer. I’m not sure anyone does. But I do know this: the moment we stop questioning how much authority we give AI is the moment we should start worrying.
