
AI Can Be Brilliant and Completely Wrong, Here’s Why (Middle School and up)
Episode 3
Send us Fan Mail AI can write poems, answer questions, and sound confident, but does it actually understand what it’s saying? We sit down with Reed Koch, Director of Engineering at KungFu.ai, to unpack what “intelligence” means in machine learning and why computers can be both impressive and hilariously confused at the same time. Reed shares his unusual path into AI through languages, linguistics, and natural language processing (NLP), then makes big ideas feel simple with stories kids (and adults) can remember. We talk about generative AI and how tools like ChatGPT often work more like “singing the next line” than thinking like a person. Along the way, Reed breaks down key concepts like pre-trained transformers, classification vs generative models, and the difference between supervised learning, unsupervised learning, and reinforcement learning. We also get practical: where kids can start learning Python, how community groups and camps help you stick with it, and why prompt engineering is becoming a real skill. Reed tells a wild example of an AI system that struggled with “fake apple” reviews and Halloween costumes, then connects it to a bigger point about data quality, bias, and responsible AI design. If you’re curious about AI for kids, coding for beginners, NLP, and AI safety, hit play, share this with a parent or teacher, and subscribe so you don’t miss the next conversation. What’s one job you wish AI could help with, and what’s one job it should never do alone?
Publisher
AI for Kids
Published
October 13, 2026
Duration
34 min
Format
Podcast
Category
Lifestyle · Kids & Family
Source
RSS feed
Host
Amber Ivey (AI)
AI for Kids
@aiforkids
#brilliant #completely #middleschool #educationforkids #parenting
