the biggest weakness of current LLMs is that they can't learn from experience like humans do we improve by learning from mistakes, but AI doesn't, at least not yet GPT-5 might be smarter than humans in many areas, but without this ability, its usefulness stays limited
7,82K