Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?
This is your opportunity to ask questions. No question too simple or too silly.
Culture war topics are accepted, and proposals for a better intro post are appreciated.
Jump in the discussion.
No email address required.
Notes -
(Mildly) Interesting observation:
Recently, people on Twitter have claimed that Claude 3.5 Sonnet is stunningly good at guessing a user's ethnic background from any substantive amount of example text, even examples which have no glaringly obvious tells.
I decided to randomly throw in 2 to 3 comments >500 words each in two tries to see how it did.
In both cases, it correctly guessed I was Indian. The first try had a comment which tangentially mentioned the fact that I was a doctor who had relocated to the UK for training and some tendency to use British spelling, which immediately made it jump to South Asian/Indian. I decided to omit that one and go off more esoteric comments, and once again it got it bang on. I won't share the full chat since it would be around 90% my own comments copied and pasted, but for the last, stunning, example, Claude noticed:
I'm blown away. I had no idea that you could even make that kind of derivation, none of these strike me as Indian™ tropes in the least. All LLMs are excellent world modelers (and by extension at modeling the user), but that's better than I expected and by far.
I'd be curious if anyone else wants to give it a try and is willing to report back. Just copy two or three substantive comments and throw 'em in the pot.
Tried a few of my comments here on a blank prompt; it's either a testament to my mimicry or a consequence of little substance but it mostly fails, especially memes and/or chudisms seem to throw it off and it defaults to American. Weirdly enough, the failure rate is lower when I paste multiple comments at once (even when individually it judges every comment as American), the main mechanism at work indeed seems to be pattern-matching. ...Man, an AI-driven police state would be some shit, huh?
It's still mildly spooky with some of my drafts and longer writeups - Claude has none of my shit and consistently guesses right across multiple regens, even standing its ground when I wink-wink-nudge-nudge it if it's really really sure. Its explanations are also sometimes funny:
I think I just got dissed by a machine, send help.
Really? I thought it's a common idiom, point taken.
For what it's worth, 4o indeed fails 100% of the time on the same prompts. Don't have o1 to try but 4o seems to get sidetracked by the content almost immediately so I don't think the CoT layers would help much.
I don't have consistent access to o1 either (I do nothing that warrants the expense that Flash 2 Thinking can't wrangle), but I agree it was a prude about it.
There's truesight and there's truesight. I have very little doubt that most people would never clock me as Indian if it weren't for all the times I've mentioned it intentionally.
(At least it never had the balls to call me an ESL speaker lol)
Hmmm
:(
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link