How to explain Generative AI in the classroom

8 days ago

Copy Link

Generative AI is becoming a common tool across various fields, and teaching children about it involves understanding its workings, risks, and effective usage.
The approach includes six hands-on projects in Scratch to introduce generative AI, covering real-world applications, model behavior, prompt shaping, and reliability techniques.
Core themes focus on how models generate text, how to steer outputs, and when to distrust models, mirroring real-world AI practices.
Key jargon includes terms like 'language model', 'context window', 'temperature', 'hallucination', and 'retrieval-augmented generation (RAG)', introduced through practical experimentation.
Projects include 'Language Models' (foundations), 'Story Teller' (creative generation), 'RAG Time' (reliability), 'Personas' (prompt engineering), 'Translation Telephone' (semantic drift and bias), and 'Benchmarking' (model testing).
Students learn to recognize AI limitations like hallucinations, bias, and semantic drift, and techniques to improve reliability, such as RAG and few-shot prompting.
Challenges include hardware requirements for running models and downloading large model files, with advice to pre-download models for classroom use.
The goal is to equip students with critical thinking and practical skills to use generative AI effectively and safely.

Hasty Briefsbeta