BTW: when I posed this question
> how do I use ffmpeg to slice out a video, I only want 3 seconds of video from the middle
to chatgpt4, it did not mention anything about accounting for keyframes. So I'm not sure what version you're using, but it's not the one I have access to :/ Another reason to assume good faith, because you seem to have access to something that I don't.
And, this is now (well before this specific conversation) an expired example because openai's model has been specifically training on this nuance, and thus even if it never made a mistake around it, it still wouldn't be impressive to me; simply because this nuance was directly added to it's model. Remember, we're talking about if the abilities of available models are impressive, what would be impressive to me is if they're able to generalize well enough to know things they haven't been directly taught.
> Why do so many assume we’re on the cusp of super-intelligent AI?
And how "smart" text to text engines can become.
My argument is actually pretty close to yours. People that are impressed, want to be. They're looking for ways to be impressed. Just like you think I'm looking for ways not to be impressed. Which, even if I am, it still should be easy to convince me (or others) that super-intelligent AI is coming soon to an API near you!
Right now, the LLMs that are the new hotness, aren't super intelligent. They're not domain experts, they can't play a simple game that children love without cheating. They're easily distracted if you know how to word a question correctly. While it's cool they can generate language that seems to be true, and useful. It's not as useful as what already exists. (For the context that I care about) And, all of these examples are markers of below average intelligence. What's the argument that would convince me that very soon, we'll take something stupid and make it hyper-intelligent? Because that's no something I've seen happen before.
Still, all of that said. Your comment that seems clearly meant to be insulting, doesn't actually make any other point. Did you mean to imply something else? Or was it just an attempt to throw insults by pretending to make an observation? Because the conversation is about if people should be impressed. Which when it comes down to it an opinion, are opinions bad?
So whatever you think you're criticising isn't the same thing that I'm using. Kind of demonstrates my point that you're either using an older version or choosing to ignore evidence for some reason.