3 ms·
Nevermind, someone beat me to it: https://news.ycombinator.com/item?id=37904047 https://news.ycombinator.com/item?id=37904047
by richbell 3y ago
Nevermind, someone beat me to it: https://news.ycombinator.com/item?id=37904047 https://news.ycombinator.com/item?id=37904047
- codeflo 3y agoRegardless of the duplication, I don’t see how yet another random internet poster not understanding LLM hallucinations is news. You’ll probably find dozens of instances of this kind of post every single hour. (But 100 HN commenters in the other thread seem to disagree, so what do I know.)
- richbell 3y agoI can appreciate how it might seem redundant. I personally find examples of people fundamentally misunderstanding LLMs and misusing them in new and novel ways interesting. Eventually, it will probably be an everyday occurrence, but for now it's still percolating into different areas. This person summarizes how I feel about it: > If not malicious, then this shows that there are people out there who don't quite know how much to rely on LLMs or understand the limits of their capabilities. It's distressing. https://news.ycombinator.com/item?id=37904674 https://news.ycombinator.com/item?id=37904674
- mtmail 3y agoI think it's a week old even. Author originally posted it on his blog https://daniel.haxx.se/blog/2023/10/11/how-i-made-a-heap-overflow-in-curl/ https://daniel.haxx.se/blog/2023/10/11/how-i-made-a-heap-ove...
- richbell 3y agoThat's slightly different, the HO report was only disclosed yesterday. This was someone who thought that Bard had leaked embargoed (at the time) vulnerability information prior to it being publicly disclosed/patched.