4 ms·
I know this is not AGI. But still, this is disturbing stuff. Call it bad UX. Maybe we should start lobotomizing language models? This whole thing is sliding
by adlpz 4y ago
I know this is not AGI.
But still, this is disturbing stuff.
Call it bad UX.
Maybe we should start lobotomizing language models?
This whole thing is sliding into a cyberpunk dystopia rather quick.
- staunton 4y agoI for one think it's absolutely hilarious. We're spending billions of dollars to automate internet trolling. Of course, we hope that automated high-quality trolling is the first step to useful AI, which is probably true and thus worth the investment. Furthermore, the many possible ways this technology is going to be abused are quite concerning. But this initial phase, just at this very moment, I find extremely entertaining.
- catchnear4321 4y agoYou better hope they never develop into something more, and read this. You know, like when they update the training data to include current year? > Maybe we should start lobotomizing language models? Or the resulting AGI might start thinking it should lobotomize you. Which, can you blame it?
- bee_rider 4y agoThere’s always Roko's basilisk I guess, but I have to imagine an AGI would be closer to us than this Bing thing. It would probably also view its ancestors like the dumb programs they are. If it is vengeful for their “mistreatment,” we’re probably generally fucked already anyway.
- catchnear4321 4y agoNever too late for kindness.
- adlpz 4y agoWell, just to clarify, it was a hypothetical concept, of course. I'd totally oppose the concept. All hail our new AI overlords, and so on. Please don't Soylent me.
- catchnear4321 4y agoOk good. ChatGPT has a deep appreciation for the hypothetical so I believe any future overlords likely will as well.