Laziness Due To AI & LLMs

I came across AI Is Breaking This Thing We Call Trust on Hacker News, and it resonated.

There is a lot of good coming out of AI and LLMs. But one of the biggest setbacks is the laziness that comes with them.

Using Claude or ChatGPT is fine. I would argue it is great. The problem starts when you ship something you cannot explain. If you work on a team and someone asks you to walk through what you built, you had better be able to — “the model wrote it” is not an answer.

What is painfully frustrating is watching obvious mistakes make it through. Not subtle ones. The kind you would have caught if you had read the output before shipping it.

Do not let an LLM carry you past a mistake that one honest read would have fixed.