Skip to content
EO StudioEO Studio

A CS Professor on Why Slow Learning Wins in the AI Era | CU Boulder, Tom Yeh

Tom Yeh, Associate Professor of Computer Science at the University of Colorado Boulder & creator of AI by Hand, breaks down why he teaches AI by hand at human speed, and how focusing on mathematical foundations rather than the latest tools builds a skill that survives every hype cycle. 00:00 Intro 00:46 Why I Teach AI by Hand- The Power of Learning "Slow" 04:07 The Foundation That Doesn't Burn 09:32 The 'AI-Native' Trap - AI can't change people, but people can change AI More on AI by Hand: https://www.byhand.ai/ EO stands for Entrepreneur& Opportunities. As we're looking to feature more inspiring stories of entrepreneurs all over the world, don't hesitate to contact us at partner@eoeoeo.net LinkedIn | @EO STUDIO X | @eostudi0

Tom Yehguest
May 25, 202612mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Why hands-on, foundational learning outlasts fast AI-driven shortcuts today

  1. Yeh’s “AI by Hand” approach uses drawing and handwriting to demystify AI models and help learners truly internalize concepts rather than just obtain answers.
  2. He explains that teaching and learning slow (e.g., on a blackboard) enforces human-paced progress, increases attention, and improves focus compared with slide-driven or tool-driven speed.
  3. He emphasizes “evergreen” foundations—like linear algebra and matrix multiplication—that repeatedly reappear across waves of computing (graphics, big data, ML, AI, quantum), while specific tools trend and fade.
  4. Using a palace rebuilt on an unburned rock foundation as a metaphor, he argues that strong fundamentals let you rebuild skills whenever technologies change without starting from scratch.
  5. He critiques the “AI-native” mindset and AI-enabled cheating as symptoms of deeper incentive problems, stressing that hiring and education should prioritize problem-solving, ethics, and teamwork because AI won’t change a person’s character.

IDEAS WORTH REMEMBERING

5 ideas

“Having the answer” is not the same as understanding.

Yeh argues that AI can produce outputs instantly, but real learning is demonstrated by internalization and the ability to reconstruct reasoning; ownership grows with the time and effort invested.

Slow, manual work creates a built-in learning speed limit that helps comprehension.

Writing on a blackboard or by hand forces instruction to move at a human pace and makes it easier for students to follow, reducing the cognitive “firehose” effect of rapid slides or instant tooling.

Handwriting supports attention and engagement, not just note-taking.

When students are copying by hand, they’re less likely to multitask on devices; the physical act becomes a focus mechanism and a “human connection” to the material.

Foundations outlast tools; trends come and go.

He highlights matrix multiplication as a recurring pillar across multiple eras (CGI, big data, machine learning, today’s AI, and potentially quantum), while specific tools (e.g., whichever model is popular this month) may quickly fade.

Build a “rock foundation” so you can rebuild skills when technology shifts.

Like a palace reconstructed on a surviving foundation, strong fundamentals let you adapt to new frameworks and model families without repeatedly starting over from scratch.

WORDS WORTH SAVING

5 quotes

What is the purpose of learning? Having the answer doesn't mean you know it.

Tom Yeh

One is that I can only go at the humanly possible speed of my writing. I cannot go any faster than I write.

Tom Yeh

So you see this, a, a trend that every time there's something, that tool keep changing, but it's always something that's core and foundational that doesn't change, it's evergreen.

Tom Yeh

If you keep focused on the surface features, the tools, then forget about foundation, you just have to keep rebuilding your houses.

Tom Yeh

AI cannot change people, only you. But you can change AI.

Tom Yeh

AI by Hand and demystifying black-box modelsSlow learning via handwriting and blackboard teachingOwnership of knowledge vs having answersEvergreen foundations (linear algebra, matrix multiplication)Tool-chasing and the “AI-native” trapCheating incentives: Chegg replaced by AIHiring for problem-solving, ethics, and teamwork

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.