Key facts
- Microsoft CEO Satya Nadella acknowledged significant internal "tokenmaxxing" among employees.
- Nadella stated that employees should not use frontier AI models for non-frontier problems.
- He emphasized matching AI tasks with appropriate models to ensure value and manage economics.
- Nadella highlighted Microsoft Copilot's auto mode as a system designed for task-model matching.
- He discussed his ongoing efforts to transform Microsoft into an AI-centric company.
Microsoft CEO Satya Nadella has acknowledged a significant amount of "tokenmaxxing" occurring within the company, referring to the excessive use of AI models. Speaking at a live taping of The New York Times' "Hard Fork" podcast, Nadella stated that while the novelty of AI can be addictive, employees need to step back and consider the purpose and value creation of their AI usage.
Nadella emphasized that not every problem requires the most powerful AI models, advocating for the use of the right model for the specific task. He pointed to Microsoft Copilot's auto mode as an example of a system designed to match tasks with the most appropriate model, ensuring both desired outputs and economic efficiency. "It can't be a race to doing things that just don't add value," Nadella remarked.
He also shared an anecdote about a tool that keeps software projects updated by monitoring workplace conversations, enabling AI to create plans and make code updates autonomously. Nadella has been actively working to reshape Microsoft into an AI-focused company, competing with smaller, agile rivals. This includes appointing a new CEO for Microsoft's commercial business and a dedicated AI advisor in late 2023 to rethink the company's business model for the AI era.