>What gave me goosebumps about this was not GPT-4’s ability to define nouns, or even its generically correct but unimaginative take on Yakuza thug Japanese, but rather the recognition that this was a bit, that it was supposed to be funny, and that having our thug jump from explaining “book” and “pen” to “ontology” represented an opportunity to tell a joke
This reminds me a lot of when someone asked ChatGPT to generate an imaginary set of top Hacker News articles 20 years in the future. Every one had a quality of being a little too on the nose to the point of feeling like wink-wink in-jokes.
The author is crediting this as a goosebump-raising insight into what is "really" being asked for. I think it reveals that beneath the surface people really want when they pretend they want serious predictions or serious non-fiction commentary.
But what if you actually wanted what you said you wanted? I think there's a very real aspect of human nature that is ready to credit non-responses as on topic if they are funny enough. And I find it disappointing when people are able to be pushed off topic in that way. I think a lot of meme-inspired misinformation exploits this kind of thing on purpose, leading to a lot of "well I don't know if it's true or not, but you have to admit it's funny!"
So at least in the fake HN headlines example I felt it succeeded at reading a subtext, but a frustrating one that reflected a basic limit to curiousity that I find frustrating to have confirmed.
>What gave me goosebumps about this was not GPT-4’s ability to define nouns, or even its generically correct but unimaginative take on Yakuza thug Japanese, but rather the recognition that this was a bit, that it was supposed to be funny, and that having our thug jump from explaining “book” and “pen” to “ontology” represented an opportunity to tell a joke
This reminds me a lot of when someone asked ChatGPT to generate an imaginary set of top Hacker News articles 20 years in the future. Every one had a quality of being a little too on the nose to the point of feeling like wink-wink in-jokes.
The author is crediting this as a goosebump-raising insight into what is "really" being asked for. I think it reveals that beneath the surface people really want when they pretend they want serious predictions or serious non-fiction commentary.
But what if you actually wanted what you said you wanted? I think there's a very real aspect of human nature that is ready to credit non-responses as on topic if they are funny enough. And I find it disappointing when people are able to be pushed off topic in that way. I think a lot of meme-inspired misinformation exploits this kind of thing on purpose, leading to a lot of "well I don't know if it's true or not, but you have to admit it's funny!"
So at least in the fake HN headlines example I felt it succeeded at reading a subtext, but a frustrating one that reflected a basic limit to curiousity that I find frustrating to have confirmed.