When Retirement Stopped Working

Alphabet Inc. (GOOG) co-founder Sergey Brin retired roughly a month before COVID-19 upended everything. His plan was simple: sit in cafés, study physics, enjoy some downtime. Then the pandemic hit, routines evaporated, and Brin found himself "kind of stewing" without a technical outlet.

Speaking recently with Stanford School of Engineering Dean Jennifer Widom, Brin described the experience bluntly. He felt himself "spiraling" and "not being sharp." When Alphabet began allowing limited numbers of employees back into offices, he started showing up more often. That informal return evolved into deeper involvement with internal AI projects, including work on Google's Gemini AI models. He called it "very rewarding" and said staying retired "would've been a big mistake."

How Google Let the AI Lead Slip Away

Brin offered a surprisingly direct assessment of how Google fumbled its early AI advantage. The company published the Transformer research paper in 2017, a breakthrough that became the foundation for modern large language models. Then it hesitated.

According to Brin, Google underinvested after that pivotal research. Internal concerns centered on the risk that chatbots might "say dumb things," so the company held back on broader public releases. Meanwhile, competitors didn't wait around. Brin said OpenAI "ran with it," accelerating the public adoption of generative AI tools while Google watched from the sidelines.

Still, he argued that Google's long game positioned it well. The company had invested heavily in neural-network research, custom-built chips, and massive data centers. Brin pointed out that very few companies operate across the full AI stack at that scale, combining in-house research, proprietary semiconductors, and global computing infrastructure. That advantage matters, even if the company was late to the chatbot party.

Advice for the AI Generation

When asked how students should prepare for careers shaped by artificial intelligence, Brin pushed back against the idea of abandoning technical work. He warned against switching fields simply because AI can now write code.

"I wouldn't go off and switch to comparative literature because you think the AI is good at coding," he said. His reasoning: AI already performs well across many nontechnical tasks, so avoiding coding won't provide any real shelter. Besides, coding remains valuable and widely used inside AI development itself.

Brin also took a moment to reflect on past failures, singling out Google Glass as an example. "Everybody thinks they're the next Steve Jobs," he said. "I've definitely made that mistake."

These days, Brin is closely involved with AI models again, and he says the pace of progress keeps him engaged. "If you skip the news in AI for a month, you're way behind," he noted. It's a far cry from sitting in cafés studying physics, but apparently that's exactly what he needed.