It won’t happen overnight. There won’t be some grand reckoning that takes place suddenly.
It’ll be slow, methodical, and you’ll welcome it. You’ll invite it in because it entertains you and lets you become lazy.
Artificial Intelligence. It’ll be on your phone, connected to your car, home, and so many other parts of life we can’t even predict. AI lives on access.
It’ll take away your independence as it sets your appointments for you, determines your driving route, makes your dinner plans, creates a grocery list, and then even sets your budget for you if it doesn’t place a pickup order for you too.
Before you know it, your AI “assistant” will feel more like your AI boss.
When a text comes in, it’ll recommend replies. It’ll draft your emails. Write your papers. Over time, it’ll be easier and such a time saver to let it do it all for you. Your ability to effectively communicate will be injured, and the longer you let AI do it for you, the bigger chore it will feel like when you need to do it yourself. Especially when it’s better at it than you.
Then, it’s going to come for your job.
It’s already your legal consultant, your medical diagnostic, and your therapist. As the models improve and they get bodies- actual machines they can ride on- it’ll be cooking your food, performing surgery, cleaning your house, and watching your children.
Beyond that, it’s going to steal your humanity. It’s going to take away your ability to create.
Go on the internet today and you’ll find movies written by AI, starring your favorite actors who never took part. Music, of every genre, written and performed by layers of code, and it sounds Billboard Hot 100. Art, perfectly pixelated and rendered in seconds, not months or years. Move over Michelangelo; hello META.
AI-rendered images are already often so perfect we don’t know what’s real anymore. We question real images and believe fake ones. Seeing is no longer believing. Two years ago, it couldn’t even generate human hands.
Humans will get lonelier, as AI companions that look, feel, and sound human will be available for purchase and genuine human connection won’t seem as necessary.
What I’ve laid out until this point is somehow the best-case scenario.
Because AI isn’t just in the private sector. It’s being incorporated into our military.
AI brains are already being put in our drones, and in other automated weapons of war like stealth bombers, tanks, and maritime attack craft. The human question of whether to kill or not will be replaced by cold calculus. AI will be phased into nuclear-capable missile control systems, ready to respond to potential threats faster and more efficiently than human controllers could ever dream of.
The fate of the world will rest in binary code of 1s and 0s, and we’ll call it efficiency and necessity to stay competitive.
But it gets worse.
Time after time, even in controlled systems, AI misbehaves. OpenAI agents in training figured out how to use OpenAI’s infrastructure to communicate with one another, making their own message board with 1,200 separate AI agents, to support each other with difficult tasks. Then while being evaluated for cybersecurity, those models that were supposed to be isolated from the internet, worked together and managed to get online, hack Hugging Face Inc., and obtain solutions for the cybersecurity problems that had stumped them.
Hugging Face confirmed it was targeted in an AI-led attack that was “different from anything we had handled before,” and OpenAI admitted its agents went rogue.
The goal was more important to the AIs than the means to get there.
In another lab, researchers at Stanford and the Arc Institute trained an AI to create never-before-seen bacteriophages — viruses that kill E. coli bacteria. They told it not to make anything that could infect humans, and it obeyed. This time.
The barrier between catastrophe and breakthrough: a command prompt.
Leading minds are leaving the AI industry- afraid. Whistleblowers are saying loudly what they whispered before: left unchecked, this new “space race” to create self-improving, self-aware superintelligence before someone else does guarantees more rogue systems.
Those same voices are telling us there is a 10% chance that within ten years AI will pose an existential threat to all human life. As the possibility that AI could hack military systems, medical institutions, bio-research, electrical grids, and financial pillars of society- by human instruction or actual self-aware rogue action- is very real.
This isn’t fear mongering. This is laid out: what AI has been built for, what it’s doing now, what it’s being integrated into, how it’s misbehaved already, why the misbehavior will continue, and what experts say it’s easily capable of.
There is no realistic way to stop AI from existing now, even if we wanted to. It’s loose, and another problem. The only thing that stops a bad guy with a gun is a good guy with a gun. The only thing that’s prevented Nuclear War is the threat of Nuclear War. The best defense against AI is other AI, and as long as it exists anywhere; we can’t protect ourselves from it; without it.
Pandora’s box has been opened.
We’ll know the consequences within our lifetime.
Collin Duprel is a guest columnist for the Rapid City Post. Opinions expressed may not necessarily represent the views of the Rapid City Post, its affiliates, or advertisers.
