lol, wait so sci-fi got it all wrong?
-
lol, wait so sci-fi got it all wrong?
The AI superintelligence isn’t gonna go rogue and act against human interest out of malicious intent.
It’s gonna do it because it doesn’t even know any better, it can’t follow instructions predictably, and — in the face of all that — we still gave it control of the wheel.
Who was writing that sci-fi novel? ha
-
lol, wait so sci-fi got it all wrong?
The AI superintelligence isn’t gonna go rogue and act against human interest out of malicious intent.
It’s gonna do it because it doesn’t even know any better, it can’t follow instructions predictably, and — in the face of all that — we still gave it control of the wheel.
Who was writing that sci-fi novel? ha
@jimniels If Then by Matthew De Abaitua contains such a scenario, where people's lives are basically ruled by some AI that's beyond their comprehension. It's feverish and weird and pretty good! The AI there in the book is much more scenario-based and natural than today's text parsers.
All of isn't fundamentally different from the side effects of YouTube and TikTok algos, where people end up doing the weirdest shit as their daily work, because that's somehow what brings the most ad revenue.
-
@jimniels If Then by Matthew De Abaitua contains such a scenario, where people's lives are basically ruled by some AI that's beyond their comprehension. It's feverish and weird and pretty good! The AI there in the book is much more scenario-based and natural than today's text parsers.
All of isn't fundamentally different from the side effects of YouTube and TikTok algos, where people end up doing the weirdest shit as their daily work, because that's somehow what brings the most ad revenue.
@koos haven't heard of it, will have to check it out!
-
@koos haven't heard of it, will have to check it out!
@jimniels i just checked my notes on the book and it's the second in a series. All three are about AI going weird and although the other two were more fun to read, If Then made most impact on me.
-
lol, wait so sci-fi got it all wrong?
The AI superintelligence isn’t gonna go rogue and act against human interest out of malicious intent.
It’s gonna do it because it doesn’t even know any better, it can’t follow instructions predictably, and — in the face of all that — we still gave it control of the wheel.
Who was writing that sci-fi novel? ha
@jimniels This is like the core idea of the community of people why we’re thinking about AI since like 00s.
There are many different specific scenarios. But the core idea is that it’s practically impossible to express this “human interest”. Whatever you try you can not state it in a way that would not harm anyone ever. And that’s in the case AI gets agency (actually do stuff in the real world) and autonomy (do it on its own), and also smarter than an average human.
For a brief few years they were called AI doomers by the opposition. But now that we kinda checked two first criteria on the list I don’t see much of that term around. I guess, it’s not that derogatory now that we start seeing dumb as a cork LLMs running in a loop off of a bunch of markdown files write takedown posts and flood internet with garbage.
-
R relay@relay.mycrowd.ca shared this topic
