Item — HN Reader

"AI" has been doing that since the 1950s though. The problem is that each time we define something and say "only an intelligent machine can X" we find out that X is woefully inadequate as an example of real intelligence. Like hilariously so. e.g. "play chess" - seemed perfectly reasonable at the time, but clearly 1980s budget chess computers are not "intelligent" in any very useful way regardless of how Sci Fi they were in the 40s.

So why's it different this time?

1 comments

yeah im in agreement, ai will eventually do everything a human being can

Perhaps, but why are you so convinced we're so close when we weren't all the other times?

Not OP, but I do think that because "humanlike abstract thinking and informal reasoning is completely unnatural to how computers work, and it's borderline impossible to make them do that" was by far the biggest AI roadblock, in my eyes.

And we've made it past. LLMs of today reason a lot like humans do.

They understand natural language, read subtext, grasp the implications. NLP used to be the dreaded "final boss" of AI research - and now, what remains of it is a pair of smoking boots.

What's more is that LLMs aren't just adept at language. They take their understanding of language and run with it. Commonsense reasoning, coding, math, cocktail recipes - LLMs are way better than they have any right to be at a range of tasks so diverse it makes you head spin.

You can't witness this, grasp what you see, and remain confident that "AGI isn't possible".