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

Key Concepts

• AI by Hand: An educational philosophy emphasizing the manual derivation of AI algorithms to demystify "black box" technology.
• Foundational Knowledge: Evergreen concepts (e.g., matrix multiplication) that remain relevant across technological shifts.
• Human-Speed Learning: The pedagogical benefit of manual writing to pace information intake and improve focus.
• Ownership of Learning: The distinction between possessing a credential and internalizing a skill through effort.
• The "Foundation" Metaphor: Comparing technical skills to the stone foundation of a palace; tools may change, but the base remains.

---

1. Main Topics and Key Points

• Demystifying AI: Professor Tom Diet argues that AI is not a mystery but a set of mathematical processes. By writing out the math by hand, learners can move past the "black box" perception.
• The "By Hand" Methodology: Diet advocates for manual note-taking and calculation. He notes that writing at a "humanly possible speed" prevents cognitive overload and forces students to disconnect from digital distractions (like social media).
• Evergreen Foundations: The speaker identifies matrix multiplication as the foundational pillar of computer graphics, big data, machine learning, and potentially quantum computing. He argues that while specific tools (e.g., DeepSeek, Google Cloud) are transient, these mathematical foundations are permanent.

2. Real-World Applications and Case Studies

• The Gyeongbokgung Palace Analogy: Diet uses the historical reconstruction of this palace to illustrate his point. The palace was destroyed, but because the original stone foundation remained, it could be rebuilt. He applies this to career development: if you have a solid foundation, you can adapt to any new technological tool.
• The Chegg/AI Cheating Crisis: Diet observes that when platforms like Chegg were disrupted by AI, cheating persisted. He concludes that cheating is a symptom of a flawed incentive system in education, not the fault of the tools themselves.

3. Methodologies and Frameworks

• The "Human Speed" Framework:
  1. Manual Transcription: Writing notes by hand forces the instructor to pace the lecture and the student to focus.
  2. Internalization: Moving from passive consumption (buying a certificate) to active ownership (spending time to acquire the skill).
  3. Foundation-First Learning: Prioritizing core mathematical concepts over the latest software trends.

4. Key Arguments and Perspectives

• AI Cannot Change People: Diet asserts that AI is a tool, not a character-builder. If a person is not a team player or a problem solver, AI will not make them one.
• The Value of Effort: He argues that the willingness to struggle with a difficult concept—even if the specific equation is forgotten later—is what differentiates successful learners. It builds the "muscle memory" of how to learn.
• Hiring Philosophy: Employers should prioritize work ethic, problem-solving, and collaboration. If a candidate possesses these traits, they will naturally adopt AI tools to enhance their performance without being forced.

5. Notable Quotes

• "Having an answer doesn't mean you know it."
• "AI cannot change people, but you can change AI."
• "Whether you own or now you own something, you value something, is actually proportional to how much time you spend on acquiring that piece of knowledge."

6. Technical Terms

• Transformer: A deep learning architecture designed to process sequences of data (like words in a sentence) using attention mechanisms.
• Support Vector Machines (SVM): A traditional machine learning method used for classification and regression, which the speaker studied before the rise of deep learning.
• Black Box: A system where the internal logic is hidden from the user; the goal of "AI by Hand" is to open this box.

7. Synthesis and Conclusion

The core takeaway is that in an era of rapidly evolving AI tools, the most valuable asset is not the mastery of a specific software, but the foundational ability to learn. By focusing on "evergreen" mathematical principles and maintaining a "by hand" approach to learning, individuals can build a resilient identity that survives technological obsolescence. Ultimately, education should focus on cultivating problem-solvers and team players, as these individuals will naturally leverage AI to their advantage, whereas the tools themselves cannot compensate for a lack of fundamental character or effort.