Maybe we should just not allow LLMs to be used by the general public and make AI hyperspecialized on academic articles instead of training it with every fucking idiot on Reddit’s opinion? Like just try and ask the Google AI about Hollow Knight Silksong. It won’t give you any reliable answers because the subreddit is super into shitposting

2

You can do something useful. Many useful use-cases. I just built an agent that automatically transcribes meetings, puts it in a txt file, uploads that to a database, then summarizes it to one page, puts it on company letterhead, then emails all participants a copy. If you say AI isn’t useful, you’re ignorant to its capabilities. Now stupid chatbots aren’t useful yes, but the LLMs behind it are.

1