In the wake of Meta’s decision to shut down its internal “tokenmaxxing” dashboard after reports of the AI leaderboard leaking to the media, Reid Hoffman, LinkedIn co-founder and venture capitalist, voiced his support for the concept, which has recently captivated Silicon Valley.
An AI token represents a segment of data processed by an AI model when understanding prompts and generating responses. It also serves as a unit to measure AI usage and cost.
Consequently, numerous companies have begun internally monitoring which employees utilize the most tokens, viewing it as an indicator of those embracing AI tools. This concept, known as “tokenmaxxing,” is derived from Gen Z slang for optimizing specific aspects, akin to “looksmaxxing” or “sleepmaxxing.”
However, tech company engineers are debating whether this metric is a legitimate workplace productivity measure, akin to ranking individuals by expenditure.
In an interview at Semafor’s World Economy summit, Hoffman shared his insights on AI adoption, expressing a positive outlook on the practice. Although he didn’t use Gen Z terminology, he supported tracking employee token spending.
“Engaging and experimenting with AI should involve people across various functions,” Hoffman stated. “A useful dashboard to observe—though not a perfect productivity measure—is how much token usage occurs.”
He further explained that while some might use many tokens randomly or exploratively, tracking “tokenmaxxing” should be paired with understanding their applications.
“Failures in experiments are acceptable. You want diverse, collective use,” Hoffman noted.
Hoffman also advised companies on AI strategies, advocating for organization-wide AI integration and regular check-ins to share findings.
“Weekly group check-ins can illuminate what we’ve tried with AI for productivity, personally and collectively, and showcase amazing discoveries,” Hoffman suggested.