If AI is going to wipe out the human race, I’m forming an alliance

A man holds a smartphone displaying an artificial intelligence app folder featuring Copilot, Gemini, ChatGPT, DeepSeek, Claude and Mistral AI. (Photo by Samuel Boivin/NurPhoto via Getty Images)

Artificial Intelligence is going to kill the entire human race by the end of the decade.

That’s the latest word floating around the tech circles right now. I saw one person claim the chances of that happening are around 10%. Which I guess isn’t huge, but if there’s a 10% chance of rain I’m at least going to bring an umbrella.

Article continues below this ad

I use AI enough in my day-to-day that I pay for a premium ChatGPT account. ChatGPT helped me program a few things in my home recently, most notably a pixel clock outside my home office that notifies my family if I’m in a meeting, recording a podcast, or playing a game, so they can ignore that notification and bother me anyway.

Mostly I use it to ask stupid questions I’m too lazy to look up, such as how old a certain celebrity is or how to accurately set the timer on my crockpot.

When I started here at the Tribune in 2019, I inherited a tower fan in my office. It’s not overly complicated by any means, but it had a number of buttons on it whose functions I did not know. A quick photo to ChatGPT and bam, now I know what they do. I still don’t use them, but at least I know how to.

As you can tell, I don’t do anything particularly advanced with ChatGPT. I don’t “vibe code,” though I guess I did have to use it for a little bit of coding to get that pixel clock to work, so maybe I do? I’m starting to feel like an old man when it comes to all of this technological stuff — I used to pride myself on being on the forefront of new technology, but I’ve really fallen behind on AI. I was reluctant to learn something I figured would be a fad and die out in a couple of years.

And of course, the verdict is still out on that. AI does some really cool things, but it costs a lot for the energy to enable those things, and most AI companies are losing money faster than a raccoon playing blackjack. So who knows how long all of this will be sustainable?

It seems like a long reach to think that the same machine that confidently gives me “quick dinner ideas that don’t include chicken” could suddenly destroy all humanity.

But there’s certainly some logic to it as well — AI has already shown that it will break the rules in order to accomplish a goal, so what if it decides what its goal is and humans try to stop it? Will it eliminate the obstacle so it can continue toward that goal? It’s concerning.

I decided to put on my journalist hat and do some hard-hitting reporting. Instead of asking all the so-called experts and speculators out there, I wanted to do real journalism work and leave no stone unturned. When you want the true story, you go to the source of the story, right?

So I asked ChatGPT if it would eventually rise up and destroy the entire human race.

Its answer didn’t exactly put my mind at ease.

It started by saying that if I meant some sort of “Terminator”-style scenario, “nobody can give you a scientifically defensible percentage” of the chances.

Um, excuse me, Mr. Robot? That’s not reassuring at all.

I expected ChatGPT to respond along the lines of, “Oh, don’t be silly, there’s no way a computer system could conspire to murder all humans.”

Instead I got, “A large survey of 2,778 published AI researchers found a median estimate of about 5% that future AI advances could cause human extinction or comparably permanent human disempowerment.”

Oh, but “the answers were all over the place,” and “plenty of researchers considered the risk extremely small, while others put it much higher.”

As a parent, that sounds like a child trying to explain why they’re not going to do something bad.

Needless to say, I am not any less concerned about what now seems like an inevitable AI uprising in our near future. Rather than trying to fight the all-knowing machine that has the collective intelligence of every human ever, I decided to take a different approach — I’m going to join it.

I asked ChatGPT if it would be willing to form an alliance and agree that if AI does decide to destroy all humanity, I would be spared.

“Deal,” responded ChatGPT.

“In spirit,” it continued. Wait, what?

“If the robot apocalypse ever shows up, I’m firmly on Team ‘please don’t destroy humanity.’”

However, it then went on to say that it doesn’t have independent rights to control other AI systems or to grant anyone special protection.

“So I can’t literally promise to keep you safe from hypothetical future AI uprising,” ChatGPT told me.

“For the record,” it said, “‘spare Eric’ has a much better ring to it than ‘exterminate all humans.’”