What to teach when AI writes the code | Rainer Stropek | TEDxLinz

Overview

Rainer Stropek reflects on how AI-driven code generation is changing programming and what that means for teaching.

Personal trigger and apparent contradiction

After an evening preparing coding exercises for students, he questioned whether AI makes teaching programming pointless. This was painful because he is an AI expert and had observed AI progress firsthand—from simple autocomplete to generating full functions and features—seemingly confirming headlines that “programming is dead.”

The real issue is loss of process, not capability

Stropek argues that the problem isn’t that AI can write code better, but that it removes the part many people love: the hands-on craft and the mental/emotional journey of turning ideas into working software through their own effort.

He frames this as a grief/identity crisis: if AI can do the technical part, what remains for the human?

Why he personally doesn’t feel obsolete

Working with AI doesn’t make him feel replaced; it increases his sense of self-efficacy by shrinking the distance between idea and reality.

He concludes that he was never only “in love with coding,” but rather in love with developing—building mental models, shaping ideas, and making them real. In his view, AI is a new abstraction layer: it changes how he works, not the essence of his role (it expands his goals rather than shrinking his contribution).

Why coding still matters

• The experience matters. Coding shouldn’t disappear; like games or physical activities, the experience is more important than pure efficiency.
• It trains thinking skills. It builds precision, logic, and patience.
• AI still makes mistakes. The “last few percent” (verification, debugging, and closing the gap from almost-correct to correct) still requires deep understanding of code.

Broader “canary in the coal mine” framing

He claims the same emotional arc—denial → fear → grief → redefinition—will spread beyond programmers as AI gets capable in other professions, including designers, writers, translators, lawyers, accountants, architects, and doctors.

The common requirement becomes redefining what is essential and irreplaceable in each field.

Teaching implications and his thesis

Since identities and roles evolve, teaching must change too. He argues it still makes sense to teach coding (as general education like math, language, or music), but instruction should shift because:

• Natural language becomes the “programming interface.”
• The bottleneck becomes clarity, not syntax.

Learners must be able to:

• describe intent clearly,
• provide examples,
• set constraints,
• specify what “good” looks like,
• verify outcomes.

Universal relevance of clarity

Clarity applies across domains: lawyers using AI, designers prompting image generators, and managers delegating to AI agents. Turning vague ideas into precise specifications is becoming a universal “superpower.”

His own redefinition

He continues teaching, but reframes his role. He no longer teaches syntax as his central identity. His new focus is agency and clarity across abstraction layers—“I’m not just a coder, I’m a developer,” and the “new programming language is clarity.”

Presenters / Contributors

• Rainer Stropek
• Transcriber: Elif Tan
• Reviewer: Vaia Katsarou