fascinating study from a 9th grader (!!) that tested whether the major LLMs recognize Gen Alpha phrases.

reason for the study was rooted in content moderation curiosity: if LLMs can’t recognize GenA slang, how can they protect users from being bullied?

the answer? they can’t. they understand GenA slang as poorly as parents.

even Meta’s Llama 3 didn’t recognize the slang, despite the 9th grader referencing Insta comments as important.

candidly, if TikTok had an LLM i think it would do very well at this test. the only thing that evolves faster than Gen Alpha slang (the first post-iPhone generation btw!) is TikTok microtrends.

oh also, i want to be friends with this 9th grader!!! she seems so cool!!!

fascinating study from a 9th grader (!!) that tested whether the major LLMs recognize Gen Alpha phrases.

reason for the study was rooted in content moderation curiosity: if LLMs can’t recognize GenA slang, how can they protect users from being bullied?

the answer? they can’t. they understand GenA slang as poorly as parents.

even Meta’s Llama 3 didn’t recognize the slang, despite the 9th grader referencing Insta comments as important.

candidly, if TikTok had an LLM i think it would do very well at this test. the only thing that evolves faster than Gen Alpha slang (the first post-iPhone generation btw!) is TikTok microtrends.

oh also, i want to be friends with this 9th grader!!! she seems so cool!!!

https://www.404media.co/ai-models-and-parents-dont-understand-let-him-cook/

building farcaster | surfing @ venice, ca | eating @ gjusta / gjelina | “lucky me, lucky mud” 🤠

ah man, this reminds me of adolescence (the show)

the topic raised in Adolescence. Not AI related directly but roots back to the same problem

Code words as concealment against AI is interesting

That's interesting. Language as a fluid construct continually evolves, so the youngest generations will have to train the LLMs on their adaptations.