I genuinely don’t understand why everyone is so hyper polarized on LLMs. I find them to be fantastically useful tools in certain situations, and definitely one of the most impressive technological developments in decades. It isn’t some silver-bullet, solve all issues solution. It can be wrong and you can’t simply take things it outputs for granted. But that does not make it anywhere near useless or any less impressive. This is especially true considering how truly young the technology is. It literally didn’t exist 10 years ago, and the iteration that thrust them into the public is less than 3 years old and has already advanced remarkably. I find the idea that they are useless snake-oil to be just as deluded of a take as the people claiming we have solved AGI.