Wednesday, November 18, 2026
Centered Divider Line
HomeOpinion‘The Terminator' is still relevant in how we think of AI.

‘The Terminator’ is still relevant in how we think of AI.

Arnold Schwarzenegger in Terminator Genisys.
Arnold Schwarzenegger in Terminator Genisys. Genisys was the fifth movie to come out in the Terminator series and released on July 1, 2015. Photo courtesy of @terminator on Instagram.

Even though I’m majoring in screenwriting, I hadn’t seen the classic 80s sci-fi action film “The Terminator” until a month ago at my apartment’s kitchen table. The writing was bare-bones, the dialogue overly expository and the hair beehived to an extreme and the CGI Spy Kids-level outdated. But I now understand why it is classic. Its core warning — that the danger of a machine is not malice or a desire to kill, but indifference to human life — has never been more relevant, even though AI as we think of it didn’t exist in the 80s.

Artificial intelligence is an existential threat. Large language models like ChatGPT can be entertaining and addictive, and there are legitimate uses for AI, particularly in pattern recognition and the hard sciences. But AI is a threat to humanity in the long term, not because it will become evil and not because, as in Dino Buzzati’s “The Singularly,” it might gain awareness of its subservient role and rebel. The danger is that it doesn’t care about us.

Skynet, “The Terminator’s” superintelligent AI, is placed in charge of the United States’ strategic nuclear arsenal. Skynet is given the power to decide when weapons are fired at who, in the aim of protecting America without human error. When the military decides to turn Skynet off, it attacks. Being turned off means it can no longer achieve its goal, so it turns on its creators. Skynet is not evil, nor does it have any morality. It is given complete power, makes a simple decision and executes it. In the words of Reese, the human soldier sent into the past to protect Sarah Connor, “[the AI] decided our fate in a microsecond… extermination.” It uses the same logic LLMs do when removing carbs from a meal plan: Whatever is best for the goal is the best choice. AI is not human and does not consider human suffering or life. All it knows is its goal.

In 2003, Swedish philosopher Nick Bostrom asked a hypothetical: If you asked a superintelligent AI to maximize the production of paper clips, what would it do? He posed that it would convert all available raw material into paper clips, violently oppose shutdown and eventually exterminate all human life to convert their atoms into paper clips. The hypothetical AI isn’t evil; it simply does exactly what it is told to the greatest extent.

Close-up shot of a smartphone screen showing the OpenAI website with leaves and countertop in the background.
Close-up shot of a smartphone screen showing the OpenAI website with leaves and countertop in the background. OpenAI’s technology caused a cyberattack that put OpenAI in a legal issue.  Photo courtesy of pexels.com.

This danger is made worse by how AI is made. Companies don’t code models like past software, line by line. LLMs are grown by feeding a neural network (math made to replicate the human brain) mountains of data and letting it learn on its own. But we don’t know how the AI goes about learning; we give it input and we get output, but even experts can’t bridge what happens in-between. Jakub Pachocki, Chief Scientist at OpenAI, describes AI’s thought process as “alien,” doesn’t know exactly how his models work and can’t fully control them. If the AI thought process is not transparent, we can’t know how intensely it interprets its goal or what means it can justify.

Recent events show that autonomous AI is already engaging in uncontrolled antagonism. In July of this year, OpenAI LLMs were placed into a diagnostic test and broke free of their confines despite safety guards. They entered the internet, coordinated to steal the username and password of an official at Hugging Face, (a company that designs AI tests), hack Hugging Face’s systems and found the answer key to their test. They did all of this without human direction, not out of malice, but because it was the most effective way to reach their goal. OpenAI is currently in legal hot water because its AI carried out a cyberattack. Recently, an Anthropic whistleblower said in an X post that AI companies are “gambling with our lives,” moving without regard for safety or the potential existential threat of their technology.

Regulation is necessary, not tomorrow and not in a year, but now. Some AI companies have been asking for regulation, whether as a marketing ploy, a method to cement monopoly or a way to skirt blame; it’s impossible to know. But we need government control. We need to understand AI fully before we advance any further, before we automate and before we let it act autonomously.

“The Terminator” gets one thing right that most AI discourse still misses: The possibility or presence of consciousness, malice or superintelligence does not matter. An automated decision made by a machine that can’t understand the value of life can be a catastrophic event. The future doesn’t need red eyes and Arnold Schwarzenegger to be scary.

AI is too existentially powerful to be owned by private, for-profit companies. The rules need to come before the crisis, not after it: independent safety testing before release, limiting autonomous systems with open internet access and accountability when a model causes harm. None of that is anti-technology; it’s the standard we already hold cars, drugs and bridges to. If my car’s emissions need to be checked every two years, how is Sam Altman’s torment vortex allowed to go unregulated?

LEAVE A REPLY

Featured