A developer argues AI coding erodes skill, produces only liabilities, and is propped up by ~$1T in debt against $0 in profit.
Key Takeaways
Skill decay feedback loop: outsourcing to AI atrophies expert judgment, making AI output harder to evaluate over time.
“Skill collapse” scenario: no junior engineers hired today means no senior engineers in 10-15 years; AI then trains on AI output and collapses.
Code is a liability, not an asset; the asset is team understanding. AI generates only the liability side.
The AI supply chain – colocation firms, compute providers, labs, and service providers like Cursor and Windsurf – has ~$1T in debt against roughly $37B revenue and $0 profit.
New NVIDIA Kyber racks (1 MW) obsolete Vera Rubin-based data centers currently under construction before they go online.
Hacker News Comment Review
Commenters largely rejected the absolutist framing: the dominant counter is that AI works well as an ergonomic accelerator when the developer can fully read and judge each line committed.
A recurring split emerged between engineers who enjoy coding intrinsically and those who enjoy problem-solving – the latter group reports AI fits naturally; the former resists it.
Several commenters flagged the “never” stance as a career liability, noting that refusing AI is increasingly a non-starter in employment contexts regardless of personal preference.
Notable Comments
@georgemcbay: separates personal projects (no AI, for enjoyment) from paid work, where not using AI is “increasingly a non-starter.”
@wyderkat: “I noticed that my friends who like AI, don’t like coding much” – concise empirical observation supporting the author’s intrinsic-enjoyment thesis.
