Post
It is currently no different with GPT-5. You can even clearly see it in videos trying to hype it up as much as possible: youtu.be/BUDmHYI6e3g?... When getting limitations on fictive requirements (demos) we just adjust our requirements. In real life, we don’t have that luxury.
youtu.beGPT-5 Fully Tested (INSANE)YouTube video by Matthew Berman
As I’ve said before, this doesn’t mean LLMs are unusable even for programming. But it means we need to be honest (IMO) what its strenghts are and not. Stop trying to out-yell your next YouTuber on how this time everything really has changed. Demos are not the best parameter for evaluating tech.
0 reposts1 like