[cogman10]

And because it's become apparent that LLMs aren't converging on what's traditionally been understood as AGI.

The promise of AGI is that you could prompt the LLM "Prove that the Riemann Hypothesis is either true or false" and the LLM would generate a valid mathematical proof. However, if you throw it into ChatGPT what you actually get is "Nobody else has solved this proof yet and I can't either."

And that's the issue. These LLMs aren't capable of reason, only regurgitation. And they aren't moving towards reason.

[dagss]

When I ask Claude to debug something it goes through more or less the same steps I would have done to fine the bug. Add some logging, run tests, try an hypothesis...

Until LLMs got popular, we would have called that reasoning skills. Not surpassing humans but better than many humans within a small context.

I don't mean that I have a higher opinion about LLM intelligence than you do, but perhaps I have a lower opinion on what human intelligence is. How many do much more than regurgitate, tweak? Science has taken hundreds of years to develop.

The real question is: When do knowledge workers loose their jobs. That is close enough for "AGI" in its consequences for society, Riemann hypothesis or not.

[joomla199]

Did you read the whole thread and all of your own comment each time you had to type another half-word? If not, I’m afraid your first statement doesn’t hold.