I Will Never Use AI to Code (or write)

Author: Anthony Manning-Franklin (antman-does-software.com) Published: 2026-03-04 URL: https://antman-does-software.com/i-will-never-use-ai-to-code-or-write Reading time: 13 minutes

Note: full body could not be re-fetched (defuddle 429, host throttles). The summary below was extracted via WebFetch on 2026-05-10 and captures the nine load-bearing arguments verbatim or near-verbatim. The original is the canonical version; cite it for direct quotation.

Nine reasons the author refuses to use AI for coding or writing

1. Personal enjoyment. Coding is a craft Manning-Franklin enjoys. Compares it to his father's view on writing — using AI to generate output removes the satisfaction of the creative act itself. The thing being optimized away is what made the work worth doing.

2. Skill development. Distinguishes recognition (passive reading about concepts) from genuine skill acquisition, which requires active practice, mistakes, and integration into mental models. Warns that AI can create an "illusion of learning" — you saw the answer, you didn't earn the model.

3. Skill decay. Continuously outsourcing coding tasks creates a feedback loop where skills atrophy over time. As proficiency declines, the temptation to delegate more work increases, accelerating dependency.

4. Unsustainable economics. AI industry has raised approximately $1 trillion in debt while generating only $37 billion in revenue with zero profit. The structure depends on continued heavy subsidization. Use of AI is participation in a subsidy game that will reverse.

5. Accountability void. When output quality cannot be reliably assessed, responsibility becomes unclear. Critical concern for safety-critical systems.

6. Environmental impact. AI data centers require enormous energy (~10 gigawatts annually), often powered by fossil fuels and built on productive farmland.

7. Economic irrationality. Fundamental contradiction: if AI reduces headcount across industries, who will afford consumer products? An "extracting all wealth" scenario unsustainable for economic exchange.

8. Team dynamics / institutional knowledge. Code itself is a "liability"; understanding developed through collaboration is the actual asset. AI cannot generate this shared knowledge between team members.

9. (ninth point present in the essay; the WebFetch summary collapsed it into the first eight + an "Accountability Void" framing — verify against the original).

Stylistic notes

• First-person, declarative title — "I Will Never". The essay is stake-driven, not analytical.
• The economic and environmental sections sit alongside personal-craft arguments — same shape as i-am-an-ai-hater (citation paragraph + ethical refusal) but framed as a personal-discipline post rather than a refusal-of-discourse post.
• The skill-decay argument echoes skill-atrophy-supervision-paradox but applied to all coding, not just to agent supervision.