AI Models And Parents Don’t Understand ‘Let Him Cook’

Manisha Mehta, a soon-to-be 9th grader, presented research at the ACM Conference on Fairness, Accountability, and Transparency in Athens showing that Gen Alpha slang is hard for parents, professional moderators, and large language models to interpret. The paper compares kid, parent, and professional moderator performance with four LLMs: GPT-4, Claude, Gemini, and Llama 3. Kids scored 98, 96, and 92 percent across the three categories, while parents scored 68, 42, and 35 percent, and the LLMs performed about the same as parents.