3 ms·
This is a classic problem with LLM (misuse). The users wants to analyze something, but it's too long, so they chop it up into multiple requests. And by the time
by juujian 3y ago
This is a classic problem with LLM (misuse). The users wants to analyze something, but it's too long, so they chop it up into multiple requests. And by the time they get to the point, the original code snipped is long out the window and the model confidently spews up something that was never there (though it looks plausible).
- rdtsc 3y agoIt seems at some point a human (Dinesh B?) had reviewed and posted it. So, it's fine to use an LLM to summarize and analyze some piece of code, but then after Daniel responded that the analysis was junk, they persist at it, assuming Daniel was wrong, and just re-prompted LLM to paraphrase the crap again. Let's assume this person is doing it for clout, it's surprising they don't see how this behavior would hurt their own reputation.
- tamimio 3y ago> but then after Daniel responded that the analysis was junk, they persist at it, assuming Daniel was wrong, and just re-prompted LLM to paraphrase the crap again. Lack of self awareness, you will be surprised by the amount of people that have that, that’s why I prefer to work with smart emotionally intelligent individuals than ones who supposedly are “smart” with higher GPA/degree and such, because that behavior will be their everyday job and eventually turning the none-sense argument into a political one.