Elon Musk once
described development in AI as “summoning the devil,” as the potential of this incredible technology goes way beyond what we can currently imagine. Now it’s out of its box, there’s no way to put it back in and, for better or worse, we now have to hope for the best that AI will serve humanity rather than enslave or annihilate it.
The problem, though, is that, with AI drawing much of its training data from interaction with human users, it could just as easily draw ideas from fictional tales about the technology, many of which have imagined dystopian futures caused by it. The Terminator, for example, is one AI-themed offering that it hopefully won’t decide to learn from.
The Terminator Series Now More Relevant Than Ever
When James Cameron first released The Terminator in 1984, the concept of sentient robots taking over the world was completely far-fetched. The sci-fi offering was successful, though, and led to further offerings in the series. The sequel, Terminator 2: Judgement Day, went into more detail about Skynet and how the machines would eventually launch a coordinated assault on the world’s superpowers in the future.
These pictures were so popular that they led to a sprawling media franchise, with countless offerings in other media that helped to raise awareness of Cameron’s vision. In the early 1990s, the first Terminator-themed games emerged, and they inspired others in the years since. The franchise is even represented in the online casino industry, where there’s a Terminator 2 roulette online game inspired by the 1991 film.
It’s appropriate that the franchise has stayed popular in the mainstream, especially now that companies like Tesla are starting to make robots in the real world. When the Terminator films were first released, the story was a mere fantasy. Now, though, some believe it could come true, and with so many varied adaptations across modern media, AI systems have doubtless already compiled huge databanks on the franchise, raising questions about how all this data may be used in the future.
How to Control AI (Or Try To)
The important thing to note right now is that, currently, AI isn’t evil. It’s not inherently benevolent either, though. Up to this point, it’s simply followed the goals given by its creators, using whatever data and logic it’s been exposed to. The problem is, there have been various thought experiments about what could go wrong if humans accidentally plug AI with the wrong objectives.
The most famous of these is the Paperclip Problem, posed by Nick Bostrom way back in 2003. It suggests that you could potentially give AI a seemingly innocuous command, which could go on to have disastrous consequences. In this scenario, you ask it to make as many paperclips as possible, and it goes on to convert all available matter in the universe into paperclips or machines capable of making more paperclips.
Whether AI turns out to work alongside humans or not will come down to how well the people behind it have designed it. Companies like OpenAI have a duty to ensure that it deeply fathoms human values, and it can understand nuance in its directives.
The good news for now is that AI is nowhere near the stage it is in the Terminator movies. There’s no program that’s able to think wholly for itself or make decisions independent of human input. On top of that, most developers are actively working on ways to ensure that everything that went wrong in sci-fi movies doesn’t happen in real life.
Hopefully, AI can watch Terminator and simply enjoy it as a piece of entertainment – when it finally has the capacity to do so – rather than inspiration. There are plenty of other offerings in entertainment that we would be wise to keep AI away from. For example, the last thing we need is for it to start binge-watching Black Mirror and getting ideas from that!
