To do things like do accurate math, you need a different kind of model, one that is based on having actual facts about the world, generated by a process that is semantically linked to the world.įor example, Wolfram uses the example of asking ChatGPT the distance from Chicago to Tokyo it gives an answer that looks plausible but is wrong. Wolfram's point, which is valid, is that ChatGPT can't do the damn math. (The monologue also kinda sucked, but that one I'll give it a pass on, since Milch's writing was pretty incredible.) I asked it if it would generate a slightly more profane example at whatever level it was OK with, and it did add some cursing, but not nearly matching Swearengen's potty mouth. It refused to use that level of profanity. The first thing return contained not one curse word, so I told ChatGPT that it should include some more instances of 'fuckin' to better match Swearengen's filthy-mouthed yet lyrical style of speech. I also just recently watched Deadwood (which is phenomenal, btw) and asked it to generate a monologue in the style of Al Swearengen on the topic of a good night's rest. It informed me that it wasn't comfortable writing a song that glorified substance abuse. I'm pretty bullish on ChatGPT and its ilk, but I _really_ dislike when ChatGPT lectures me because my request is against its 'moral values.' I recently pasted in the lyrics from Sleep's titanic song 'Dopesmoker' and asked it to generate a song with similar lyrics. Do the damn math, don't wag your finger at me and crush my dreams! It's obviously ridiculous but is also a fairly simply dimensional analysis problem. I'm almost offended by the 'cubic light year of ice cream' answer from ChatGPT.

Math wolfram online