Undiagnosed-ADHD builder finds Claude Code breaks task paralysis but quickly triggers dopamine-loop spending, burning through Pro, Max, and API credits on a single game project.
Key Takeaways
Task paralysis (brain stops) differs from analysis paralysis (brain loops); AI lowers the activation energy for the first step, not the strategy.
Author used Claude Code plus /modus opusplan to reduce token burn while building an iOS app and a game.
The idea-to-result cycle is now so short that dopamine hits before willpower catches up, escalating spend from Pro to Max to extra API credits.
Author draws a hard line: no AI for artistic work due to harm to creators, but treats coding acceleration as a legitimate productivity tool.
Hacker News Comment Review
Multiple ADHD-diagnosed commenters confirm the pattern independently: Claude Code removes start-inertia but the novelty wears off within months, leaving frustration and a desire to return to low-level technical depth.
A recurring concern is the structural mismatch: asking people who already struggle with dopamine regulation to adopt a tool that compresses reward cycles is a recipe for compulsive use, not productivity.
Commenters flag a compounding risk where LLM sycophancy pairs badly with the relief from first-step paralysis, potentially masking poor architectural decisions before the dopamine high fades.
Notable Comments
@pllbnk: frames AI token spend as a slot-machine model, arguing the end game for AI companies is gambling infrastructure, not productivity.
@albert_e: notes a paradox where 10x throughput per session creates its own resistance to starting, because the brain anticipates being overwhelmed by output volume.
