And this was a debug session from hell, enormously helped by an AI doing much of the grunt-work.
I’d like to call it my tireless helper, but the AI several times stated flat out that this was impossible and unsolvable and that we should just write a report about it.
I suspect those things have been trained by people who may not be quite as stubborn as I am.
But while the AI was ready to give up several times, it did keep adding debug code and analyzing it faithfully when I pushed. So credit where credit is due and I let the AI write the commit message above.
This is basically a one-liner fixing a bogus “round_up()” to a “round_down()”, but there were 24 patches adding more and more debug information to this, and 18 kernel boot to finally narrow it down to this.
We need more of these kinds of honest accounts that clearly show how LLMs aren’t magical “thinking” beings. They are pieces of software that produce their best work when supplied with intent and direction from us, either during training or in user sessions.
I think this part is rather interesting:
“I suspect those things have been trained by people who may not be quite as stubborn as I am.”
I read stubbornness as a proxy, a diagnosis of what is more likely a product of how the models are trained to favor quick solutions. If the right solution isn’t immediately obvious or available after a few tries, most models (at the moment of writing) won’t stop and ponder whether taking a step back is necessary to find the root cause. Other problems that fall into this category are ones where the solution isn’t more code, but less.
We need more of these kinds of honest accounts that clearly show how LLMs aren’t magical “thinking” beings. They are pieces of software that produce their best work when supplied with intent and direction from us, either during training or in user sessions.
I think this part is rather interesting:
“I suspect those things have been trained by people who may not be quite as stubborn as I am.”
I read stubbornness as a proxy, a diagnosis of what is more likely a product of how the models are trained to favor quick solutions. If the right solution isn’t immediately obvious or available after a few tries, most models (at the moment of writing) won’t stop and ponder whether taking a step back is necessary to find the root cause. Other problems that fall into this category are ones where the solution isn’t more code, but less.