Week 5 – Why AI Safety Is So Hard (and What You Can Do)
IN THIS LESSON
Incentives, Power & Action
AI safety is not just a technical problem, it is political and economic. Large firms lobby to delay regulation. Capability research receives orders of magnitude more funding than safety research. Development cycles run in months; regulation takes years. Meanwhile, geopolitical competition punishes restraint.
This week explores:
Why safety teams lack veto power
Why labs both deny risks and restrict access
The prisoner’s dilemma of AI races
Disagreements among experts on AGI timelines and severity
We conclude pragmatically: AI safety is a field you can enter. From red-teaming and evaluations to governance, research, and policy, there are concrete paths to contribution.
Final takeaway: AI safety is not about fear, it is about making progress survivable.
