AI in Movies has nothing to do with the AI we have now
When it comes to AI, we need to focus on the facts, and not let our views get influenced by popular media like Film, TV, and books. LLMs are not as smart as us they are just very sophisticated word guessing programs. Even if they were as intelligent as us, they can't exist without us. So calm down.
In this post ill explore why sci-fi movies are flawed in their reasoning.
Suspension of Disbelief
In order to enjoy any movie you have to suspend your disbelief. In the real world Criminals don’t run rampant while the police do nothing, superheroes don’t exist, evil killers are not immortal, and AI is not likely to want to kill us!
How does this apply to AI?
Every sci-fi movie either has no cybersecurity or it’s not very good. If they had proper security, the movies would be very boring.
- All the Terminator movies – Skynet does not get very far if standard security measures are in place. Firewalls, encryption, and air locked systems can not be beaten by an AI on the network. Also, nuclear weapons have physical safeguards with lots of redundancy; Skynet would never get control, humans are not that dumb! Also, the police and army would have gotten involved in under 20 minutes at a terminator sighting. With modern facial recognition, those T-800s are easy to spot. Power disruptions from time travel can be spotted in the grid. Knowing our future, we would fight back, ban AI, and win by doing nothing!
- 2001: A Space Odyssey – Why does the AI control all of the ship? That's just dumb! Why is it so hard to turn it off? Just add an off switch in the mess hall.
- The Matrix – Why make self-replicating machines? That’s dumb. Also, machines don’t need air; why not move to space? This movie does not hold up to scrutiny. You have hover ships but no way to get into space. Even if you don’t have the fuel, blow up the planet and ride the asteroids out of there. No longer have to look after humans. Win-win.
- Westworld – Why give them real guns? Also add a kill switch.
- WarGames – Backdoors, really? Also, delete your ruddy games section before release; that’s just bad programming.
So there are no dangers?
No, there are certainly risks, think of an LLM as a power tool. A drill isn't "evil," but in the hands of someone without proper training, it is dangerous. Similarly, an AI doesn't need to be "sentient" to be a threat; it just needs to be a capable tool in the hands of someone with bad intentions.
All the "AI hacks" we see today aren't the result of a machine "waking up"; they are the results of humans failing to secure their systems, failing to monitor their data, or choosing to prioritize speed over safety.
What are the real dangers of humans misusing LLMs?
When we strip away the sci-fi, the real-world dangers are much more grounded but no less important. We should be focusing our energy on:
- Privacy Erosion: The harvesting of personal data by "smart" devices.
- Cybercrime: Phishing, hacking, and identity theft facilitated by automation.
- Information Warfare: The rise of deepfakes and the manipulation of truth.
These are all critical issues that require immediate attention and better policy. We aren't fighting a "Terminator" plot; we are fighting a lack of oversight for the tools we have already built.