Have a Coherent AI Policy
6 hours ago
- #Management Practices
- #AI Policy
- #Software Engineering
- The author criticizes tokenmaxxing, a flawed KPI that measures AI token usage, comparing it to outdated management practices like timing tasks with a stopwatch.
- An AI policy is introduced for the author's team, emphasizing no mandate for AI use, understanding AI-generated code, maintaining job capability without AI, and prioritizing people over metrics.
- The policy allows senior engineers flexibility in using AI tools while encouraging awareness of AI advancements, but rejects the contradiction in AI boosterism about urgent adoption versus rapid obsolescence.
- Junior engineers are advised to use AI judiciously to avoid hindering learning, as true skill development comes from hands-on coding and grappling with concepts, not outsourcing to AI.
- The author stresses that AI should serve human goals, not replace them, and that teams should have a coherent philosophy tailored to their context, rather than chasing vanity metrics like tokenmaxxing.