I use ChaptGPT now and then and it works great except for one task: “find me this pic or video”. One would think after training on all the things it would excel at this first and foremost. Nope. Trying to find an old cracked.com article took 25 minutes, but it finally found it in a university archive. It kept giving me links to wrong articles or 404s. Looking for an old, popular meme and it had a fit. I finally prompted, “show me whatever you think it might be”, you stupid fuck. Couldn’t do it, finally found it on Google image search.
I honestly can’t figure out vibe coding. How are people making even the simplest app?! I fed a very short PowerShell script of mine into Claude to see if there was a way to speed it up. Nope. It hosed it all the fuck up. ChatGPT has never spit working code for me, only good for getting a new idea or reminder of a different tack to try.
That’s because the text part and the image part are two separate systems. ChatGPT can’t actually ‘see’ the pictures. It passes it off to a different system, like Stable Diffusion.
Imagine you are talking to a blind guy in a box, but you don’t know he’s blind, and you ask him to find you a picture of whatnots. He says ‘sure, I can do that, of course’ and then looks up in his braile dictionary what whatnots are. After this he knocks on the box next to him where his friend is, who can see, and describes whatnots.
His friend passes him the picture and he passes it to you. It’s actually a picture of whatchamacallits.
“This isn’t a whatnot. This picture has a dongle, it’s supposed to have a dangle.”
“Oh, of course, you’re so right to call this picture out, let me try again.”
And so on. That’s why it constantly prompts you to describe more so it can pass fine tuned descriptors to the imager, but it has no idea what the picture is you want or if what he gives you is correct.
I use ChaptGPT now and then and it works great except for one task: “find me this pic or video”. One would think after training on all the things it would excel at this first and foremost. Nope. Trying to find an old cracked.com article took 25 minutes, but it finally found it in a university archive. It kept giving me links to wrong articles or 404s. Looking for an old, popular meme and it had a fit. I finally prompted, “show me whatever you think it might be”, you stupid fuck. Couldn’t do it, finally found it on Google image search.
I honestly can’t figure out vibe coding. How are people making even the simplest app?! I fed a very short PowerShell script of mine into Claude to see if there was a way to speed it up. Nope. It hosed it all the fuck up. ChatGPT has never spit working code for me, only good for getting a new idea or reminder of a different tack to try.
That’s because the text part and the image part are two separate systems. ChatGPT can’t actually ‘see’ the pictures. It passes it off to a different system, like Stable Diffusion.
Imagine you are talking to a blind guy in a box, but you don’t know he’s blind, and you ask him to find you a picture of whatnots. He says ‘sure, I can do that, of course’ and then looks up in his braile dictionary what whatnots are. After this he knocks on the box next to him where his friend is, who can see, and describes whatnots.
His friend passes him the picture and he passes it to you. It’s actually a picture of whatchamacallits.
“This isn’t a whatnot. This picture has a dongle, it’s supposed to have a dangle.”
“Oh, of course, you’re so right to call this picture out, let me try again.”
And so on. That’s why it constantly prompts you to describe more so it can pass fine tuned descriptors to the imager, but it has no idea what the picture is you want or if what he gives you is correct.