Microsoft CEO Satya Nadella has acknowledged his own addiction to token-maxing, the practice of using the most powerful AI models for every task. In an interview, he stated, 'I'm like a token-maxer too. So it is addictive.' Nadella warned that the uncritical use of frontier models can lead to inefficiencies, as the marginal cost of productivity improvements must match the marginal cost of the token. He argued that pure token-maxing won't drive real economic growth, despite his personal involvement in the practice.
Nadella outlined a sweeping vision for the future of programming, where developers will no longer write code manually. Instead, they will oversee hundreds or thousands of AI agents. He described this new core skill as 'cognitive coverage,' which involves deeply understanding the code that agents produce. 'I have a repo full of code written by agents. I'm cognitively understanding what happened,' he explained. According to Nadella, while a computer science education will still be necessary, the nature of the work itself will change significantly.
Nadella's comments come amid growing discussions about the ethical and economic implications of AI model usage. The CEO's admission highlights the tension between the allure of powerful AI tools and the need for responsible deployment. Source: thedecoder
