this post was submitted on 22 Nov 2024
112 points (80.1% liked)
Technology
60052 readers
2821 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I can answer that. We won't.
We'll keep iterating and redesigning until we have actual working general intelligence AI. Once we've created a general intelligence it will be a matter of months or years before it's a super intelligence that far outshines human capabilities. Then you have a whole new set of dilemmas. We'll struggle with those ethical and social dilemmas for some amount of time until the situation flips and the real ethical dilemmas will be shouldered by the AIs: how long do we keep these humans around? Do we let them continue to go to war with each other? do they own this planet? Etc.
Assuming we can get AGI. So far there's been little proof we're any closer to getting an AI that can actually apply logic to problems that aren't popular enough to be spelled out a dozen times in the dataset it's trained on. Ya know, the whole perfect scores on well known and respected collage tests, but failing to solve slightly altered riddles for children? It being literally incapable of learning new concepts is a pretty major pitfall if you ask me.
I'm really sick and tired of this "we just gotta make a machine that can learn and then we can teach it anything" line. It's nothing new, people were saying this shit since fucking 1950 when Alan Turing wrote it in a paper. A machine looking at an unholy amount of text and evaluation based on a new prompt, what is the most likely word to follow, IS NOT LEARNING!!! I was sick of this dilema before LLMs were a thing, but now it's just mind numbing.
AI developers are like the modern version of alchemists. If they can just turn this lead into gold, this one simple task, they'll be rich and powerful for the rest of their lives!
Transmutation isn't really possible, not the way they were trying to do it. Perhaps AI isn't possible the way we're trying to do it now, but I doubt that will stop many people from trying. And I do expect that it will be possible somehow, we'll likely get there someday, just not soon.