Ask HN: Why are we humans still prompting to make Agents better?

Posted by shubhamintech on Sunday June 28, 2026 @12:20AM
from the ask-hn dept.

Often times I think, why aren't we using more and more agents to often improve themselves. Surely they're smarter than most of us. What do you think is stopping them?

A bunch of companies have raised insane amount of capital as well for this, but it's still not productised/generally accessible?

Sort of baffles me.

( 3 comments | 2 points | on HN )

Re:Ask HN: Why are we humans still prompting to make Agents ... by andsoitis on Sunday June 28, 2026 @12:28AM (#48704354)

> Surely they're smarter than most of us. What do you think is stopping them?

They are, in fact, NOT smarter. They have more vast more knowledge, but that is different from being smart or wise.

[ Parent | on HN ]

Re:Ask HN: Why are we humans still prompting to make Agents ... by shubhamintech on Sunday June 28, 2026 @12:32AM (#48704386)

If that's the case, how does one make them smarter for their specific use case?

I'd like to believe that vast knowledge is still more than what I have in my mind?

[ Parent | on HN ]

Re:Ask HN: Why are we humans still prompting to make Agents ... by lordkrandel on Sunday June 28, 2026 @02:44AM (#48704993)

No, it isn't. Machines still don't have judgement. They don't have taste, they don't feel pain, they don't have empathy. I guess that calling it AI or LLM is too broad and unspecific, making them appear as gods, while a cat has all the initiative to survive and a machine does not have a single clue about "meaning" "feelings". Even if they can copy some text here and there about humans describing those concept, they cannot elaborate on that, they just can regurgitate knowledge, not create anything new. Give them a wrong direction, they will never go back and re evaluate, and even if you tell them to do it, YOU are the one who feels the need to verify. Not those giant perceptron masses.

[ Parent | on HN ]