The H.A.C.K.E.R. Project

This article is solely an opinion being expressed by the author and does not reflect the opinions of The H.A.C.K.E.R. Project as a whole.

AI In Media: Destroyer of Worlds

In many different movies, TV series, video games, and books, Artificial Intelligence (AI) is portrayed as the enemy. Ultron tried to drop an entire city on the planet to induce a mass extinction event in Avengers: Age of Ultron. AUTO refused to let the humans living in The Axiom return to Earth in the movie Wall-E. M3GAN, HAL 9000, GLaDOS, The Terminator; the list goes on. When people think of AI, one thought that often crosses their minds is world-domination.

What AI Can Actually Do

We have to remember, though, that these are all fictional examples of AI. AI in the real-world is quite different: they can generate images, text, or code; play and win at games like chess, Go, or tic-tac-toe; sweep our floors; or predict the likelihood that a given person has a specific disease, among other tasks. As it currently stands, a single AI won’t be able to do all of these things. Your Roomba isn’t going to be your next chess opponent. ChatGPT isn’t going to (accurately) start predicting my likelihood of developing hyperthyroidism by the time I’m 50. Unlike humans, who are able to excel at multitudes of tasks or hobbies, AI are built to efficiently perform usually just a select few functions. They don’t really do much beyond these tasks.

But couldn’t an AI learn how to do more things? Well, it’s important to mention that not all AI even incorporate machine learning. Some forms of AI are more or less just simple reflexive agents. Take, for example, a digital thermostat. You tell it what temperature you want the room to be. Maybe you tell it a specific time you want the room to be cooler or warmer, and perhaps you tell it how humid you want the room, too. Then, you let the thermostat do its thing. It reads in the current temperature, humidity levels, and time, and then acts accordingly. If it finds that the temperature is too low, it knows to active the heating function. Too humid? Then dehumidify. It doesn’t do any sort of learning, but it is able to do these tasks autonomously, just as if a human or other intelligent being were doing them. The ones that do incorporate machine learning usually learn the rules and strategies related to their specific tasks, either based on given feedback, their own experiences, or both. A chess-playing AI will first learn how to play chess, then learn basic strategies, then advanced strategies, and soon will be able to contend with grandmasters. It won’t spontaneously learn how to drive a car, nor will it choose to seek out that sort of information, either. Even if it is directed to learn these things, it won’t be able to do much with car driving information unless it is also reprogrammed to drive cars. Knowing when you should brake vs accelerate won’t be useful information in a game of chess.

Can AI Take Over the World?

As it currently stands, AI doesn’t really have the capacity to take over the world. Most of the time, AI tends to stick to what it was created to do. If I built an AI that generates images based on a prompt, that is what the AI will do. If I want it to also generate stories, then I would need to manually add that in. If I want it to drive a car, I need to program that in separately as well. Artificial Intelligence, in a lot of cases, is not really as intelligent as we think it is. Just ask any AI art generator to draw a picture of a handshake. It looks terrifying.

Will AI ever reach a point in the future where it could take over the world? Maybe, but that would require a tremendous amount of time, effort, and resources. By the time AI gets to that level, I think an AI takeover would be the least of our issues (if it even is an issue; I’d like to imagine a future where AI and humanity live in harmony). So, for now, no. AI isn’t going to take over the world, and it definitely won’t anytime soon. That “AI-takeover” cake is a lie.