AI Safety Advisor Joe Rogero: Stop Taking Courses, Start Building Things

In a viral post on the AI forum LessWrong, advisor Joe Rogero delivers a blunt message to aspiring AI safety and governance professionals: stop accumulating courses and start building tangible projects. He observes a common pattern where individuals endlessly cycle through programs like MATS, ARENA, and CAIS's AI Safety course, seeking validation through credentials rather than creation. Rogero argues this creates a bottleneck of applicants waiting for permission to contribute, while urgent problems in AI policy, technical governance, and interpretability remain unaddressed.

Rogero's core advice is to 'Just Do The Thing'—the project idea you've been putting off until you 'learn more.' He provides a concrete example: a policy memo exploring how Open-Source Intelligence (OSINT) could be used to enforce the EU AI Act, comparing it to OSINT's role in nuclear nonproliferation. This project, completed as a course final in about a month of part-time work, demonstrates the accessible bar for meaningful contribution. The post concludes by reframing the goal: become 'the sort of person who makes that sort of Thing' and use those creations as proof of initiative and capability when seeking future opportunities.