For the last four years, this is what LLMs have been doing as modus operandi:
1. here's thing you asked for. And here's a lot more you didn't ask for.
Then I notice there are mistakes, so I point them out. And I get this pattern:
2. Oh yes, sorry for that and thanks for pointing out my mistake. I think a better thing to do that would have been XYZ.
Then I see that the output is far from what I originally had in mind. I steer the LLM back by saying they have strayed and made the thing far more complicated than it should have been. And then I get this pattern of responses:
3. Oh right, you are right, I unnecessarily complicated it. Let me tackle it differently, simply just as you said I should.
By then I have run out of tokens, so I have to wait. When I have tokens again, I restart the convo, and once I get my 'solution', I notice that the LLM has done additional things in there that make it clear, at a glance, it was done by AI. I don't mind that so much as these create clutter and are totally unnecessary. So I ask it to clean it up and remove all those frills I never asked for. Then I get:
4. Got it, getting rid of those extra things.
Now the outcome I have is still a far cry from what I originally had intended to have. I would have been more efficient but definitely happier had I hand-coded it all by myself in the first place. It's not like I have to type everything out anyway, good IDEs already have made me fast.
There, now don't worry about AI taking your job. They can't and they won't.
PS. Pro-AI midwits will downvote this (it's a pattern on HN when you say sth against AI you get downvoted, so if you think I am right, please upvote so as to counter those midwit downvotes. It's a shame one has to change one's opinions in order for the opinion to be available to read. But I won't. I will say what I think, regardless of the downvotes).
jan_m_savage · · focus · HN ↗
1. here's thing you asked for. And here's a lot more you didn't ask for.
Then I notice there are mistakes, so I point them out. And I get this pattern:
2. Oh yes, sorry for that and thanks for pointing out my mistake. I think a better thing to do that would have been XYZ.
Then I see that the output is far from what I originally had in mind. I steer the LLM back by saying they have strayed and made the thing far more complicated than it should have been. And then I get this pattern of responses:
3. Oh right, you are right, I unnecessarily complicated it. Let me tackle it differently, simply just as you said I should.
By then I have run out of tokens, so I have to wait. When I have tokens again, I restart the convo, and once I get my 'solution', I notice that the LLM has done additional things in there that make it clear, at a glance, it was done by AI. I don't mind that so much as these create clutter and are totally unnecessary. So I ask it to clean it up and remove all those frills I never asked for. Then I get:
4. Got it, getting rid of those extra things.
Now the outcome I have is still a far cry from what I originally had intended to have. I would have been more efficient but definitely happier had I hand-coded it all by myself in the first place. It's not like I have to type everything out anyway, good IDEs already have made me fast.
There, now don't worry about AI taking your job. They can't and they won't.
PS. Pro-AI midwits will downvote this (it's a pattern on HN when you say sth against AI you get downvoted, so if you think I am right, please upvote so as to counter those midwit downvotes. It's a shame one has to change one's opinions in order for the opinion to be available to read. But I won't. I will say what I think, regardless of the downvotes).
Pannoniae · · focus · HN ↗
jan_m_savage · · focus · HN ↗