As AI automates narrow skills like writing code snippets, the ability to think at a system level becomes paramount. Designing how different components—including classical ML models, LLMs, and traditional software—fit together is a skill that is harder to automate and increasingly valuable.

The most effective use of AI is not in areas where you lack knowledge, but in your core areas of expertise. Your deep domain knowledge allows you to direct the AI with precision, discern quality output from mediocre results, and use it as a true apprentice.

Learning to work effectively with AI agents isn't about memorizing code syntax, which the agent handles. True technical skill is developing a deep understanding of the underlying systems' constraints and tradeoffs (e.g., different video libraries) to better guide the agent toward optimal solutions.

To make genuine scientific breakthroughs, an AI needs to learn the abstract reasoning strategies and mental models of expert scientists. This involves teaching it higher-level concepts, such as thinking in terms of symmetries, a core principle in physics that current models lack.

Sal Khan states that AI doesn't make knowledge obsolete; it makes it more critical. To create great work, humans must be able to judge AI outputs, direct the tools, and assemble the pieces. This requires a strong knowledge base, separating those who will thrive from those who get left behind.

The idea that AI makes engineering obsolete is wrong. Just as cloud computing created "leaky abstractions" that still required knowledge of networking, AI tools require engineers to understand underlying models and systems to be effective. The best AI-assisted engineers will be those with strong fundamental knowledge.

AI doesn't eliminate the need for fundamental skills; it heightens it. To use AI effectively, individuals need enough domain expertise—like basic coding—to ask the right questions, identify when the AI is wrong or "hallucinating," and understand the concepts behind its output.

To effectively apply AI, product managers and designers must develop technical literacy, similar to how an architect understands plumbing. This knowledge of underlying principles, like how LLMs work or what an agent is, is crucial for conceiving innovative and practical solutions beyond superficial applications.

With AI handling rote coding, education must adapt. The optimal approach is a "barbell" strategy: focus intensely on timeless theoretical concepts (compilers, databases) on one side, and on producing complex, high-level applications on the other. The middle ground of teaching the "craft" of coding is now obsolete.