MorphOS Developer
Posts: 588 from 2003/2/24
From: finland, the l...
(I promised myself to shy away from further comments here but I will break my promise to myself...)
This is exactly why you should not use AI for coding, especially if you're not yourself skilled enough to understand and judge the quality of the code produced. You will have little understanding what the code does and why it has issues.
LLM itself doesn't understand what it generates and why, either. This is beyond the capabilities of a language model. If you ask it to explain why the code crashes it will attempt to explain it (with confidence) and will likely be totally wrong in its explanation. Since it doesn't have actual understanding this is the best it can do.
LLMs are at a stage that they're quite convincing in producing totally wrong information. This is deceptive and dangerous.
That being said, as long as the limitations are understood the LLMs can be very useful tool for e.g, summarizing information. But always verify the results...
[ Edited by Piru 06.09.2024 - 17:22 ]