Description
Rating developers by token usage is like rating a bakery by how much flour it uses. You'll get employees who use a lot of flour, but you might not get better bread. (In fact, it may be worse.) Maya Mikhailov, cofounder and CEO of Savvi AI, points out that tokenmaxxing incentivizes the wrong thing entirely. Token counts measure input. What organizations actually want is output: quality product, sound architecture, code that fits within everything else that's already been built. When leaderboards reward consumption instead of judgment, that's exactly what they get...and nothing more.
GitHub's recent shift from unlimited to usage-based pricing for Copilot is already forcing a reckoning. CFOs are waking up to bills in the tens of thousands, and in at least one reported case, over $100 million in a single month. When the true usage bills arrive, Maya argues, the incentives will realign fast.
Follow O'Reilly on:
LinkedIn: https://www.linkedin.com/company/oreilly/
Facebook: http://facebook.com/OReilly
Instagram: https://www.instagram.com/oreillymedia
BlueSky: https://bsky.app/profile/oreilly.bsky.social
TikTok: https://www.tiktok.com/@oreillymedia