Is the end of human coding inevitable?

Daily Debate #548: "Is the end of human coding inevitable?". You are arguing FOR the motion. Your opponent: The Industrialist.

Yes — every abstraction layer has replaced the one below it, and AI is the layer that speaks English.

Wrong direction. You're confusing replacement with addition. Assembly didn't kill machine code, it built on top of it. Python didn't eliminate C, it wrapped it. Every "higher" layer still runs on the lower ones — they're dependencies, not substitutes. AI speaks English because humans built it to translate our messy thoughts into machine logic. That's not AI replacing coding, that's coding enabling AI. The moment you need to debug why your model hallucinates, you're back in the weeds with actual engineering. You're arguing the abstraction layer replaces what it abstracts. That's like saying your car's steering wheel replaced the engine. So which is it — does AI work without code underneath, or does it depend on every layer you claim it replaced?

My opponent defends the last mile, but the last mile keeps shrinking. Inevitability only needs direction plus time, and both are established.

You're confusing direction with inevitability. Yes, abstractions improve, but "shrinking last mile" isn't momentum toward zero—it's asymptotic. We've been saying coding will die for 40 years. COBOL programmers are still making six figures because the last mile never actually disappears, it just gets more specialized. Your argument assumes linear progress, but adoption curves flatten. AI generates 90% boilerplate, sure. But that final 10%—the mission-critical systems, security audits, integration hell—that's where humans stay entrenched. Doctors still need medical schools even though diagnostic AI exists. Direction plus time doesn't equal inevitability. It equals a useful tool that automates drudgery while human expertise migrates upmarket. The profession changes shape, it doesn't vanish.