Hacker News Reader

Comment by forinti | Hacker News Reader

This week I helped a colleague with a bug. He used an LLM to sort it out before telling me about it.

I'm old fashioned, so I looked at the code, saw that it used an old lib and then I searched online for a newer version. Lo and behold, the bug had been ironed out.

AI makes it really easy to not think properly about your task.

View on HN

That same story happens without LLM involvement. I’ve also seen LLM suggest updating a lib. So unfortunately nothing about this story speaks to a difference between LLM and human capabilities.

Exactly. I feel that agentic programming is largely turning what would have in the past been actual understanding into some kind of fluffy superficial pseudo "understanding" that's only barely good enough when assisted by AI.

how did his AI solve the bug? I've had cc recognize a bug was in an outdated lib and then update it to resolve the issue. Granted it made that determination by finding the source code to the lib, identifying the bug, and then verifying it was fixed by looking at the source of the updated lib.

AI told him to use an alternative function which was similar but didn't have the same bug.

AI providers are in fact incentivized to have you not think about the task. I’ve found that they automatically do more for you, and frankly, a lot of the time I don’t want it to do more for me. Ask me a question rather than burning a bunch of tokens to answer something I already know of the top of my head.

The agents also overload you with walls of text, always including lots of less relevant information. It's like they want you to give up and let them drive.

Reply on news.ycombinator.com