Last night I watched and episode of Star Trek: The Original Series and I'm still can't stop thinking about how well the episode (The Ultimate Computer) maps onto contemporary issues with AI. The theme of the episode was "unforeseen problems with advancing technology and the fear of humans being replaced with automation" - a classic sci fi theme that was relevant for the time. But while a story like this could have easily aged poorly based on technology of the era, it's instead scarily precognizant about the specifics of our problems today.
It opens with the Enterprise being voluntold that they're testing out a revolutionary breakthrough in computing science, and 95% of the crew is put on leave because this new computer is going to do all of the work for them
(Little side note here that Bones expresses concern about sailing with a skeleton crew because the limited numbers will make manning the ship difficult if the computer doesn't work; his concern is barely acknowledged before being brushed aside)
Kirk doesn't like the new computer system from the beginning, but is struggling to figure out whether his gut-instinct concern is valid, or if he's allowing his fear of being replaced to stand in the way of genuinely helpful technological progress
The scientist who created the computer explicitly brings up the argument that by replacing people, the computer will save time and resources, freeing people to do undefined other things
Instead of saving resources it turns out the computer is actually drawing massive amounts of power from the ship
Obviously, the computer ends up making bad decisions, but now it's too intertwined with the Enterprise and they can't shut it down
Then the computer starts killing people in its efforts to carry out its objectives (at this point I am forcibly reminded of the AI-assisted suicide cases)
As Spock points out, this is not the 'logical' behavior that would be expected from a computer
Turns out, the only way to make a computer complex enough to replace human decision making is to map it to a human mind
Which means that it also includes all the logical and personality flaws that are intrinsic to being human
Therefore there is no way to create an infallible or perfectly logical computer that is meaningfully capable of replacing humans
The computer scientist ends the episode in pitiable disgrace, having caused the problems in the first place with his desperate attempts to prove that he's still making relevant and groundbreaking advancements
The parallels to today's AI issues are frankly uncanny. Like, if you wrote this today, people would accuse it of being too blunt. I don't really have conclusion or message of this mini-essay, I just needed to acknowledge that 60 years ago at least one Star Trek writer was predicting the issues of today to a degree that had my mouth hanging open.
