What part of the AI future do you think people are still completely underestimating?

A viral Reddit thread with over 5,000 comments reveals that AI researchers and industry professionals believe society is fundamentally misjudging where AI will cause the most disruption. While public discourse focuses on job displacement and AGI timelines, experts identify three underappreciated areas: synthetic biology acceleration (where AI could enable engineered pathogens), energy grid instability (as AI compute demands overwhelm infrastructure), and infrastructure collapse from poorly coordinated autonomous systems. These systemic risks represent second-order effects that could manifest before AGI arrives, challenging current regulatory frameworks.

Technical experts point to specific vulnerabilities: AI-driven protein folding models like AlphaFold 3 are already reducing biological engineering barriers by 100x, while training runs for models like GPT-5 consume energy equivalent to small cities. The convergence of AI with other exponential technologies (quantum computing, nanotechnology) creates compound risks that existing governance structures aren't designed to address. Unlike job displacement, these infrastructure-level disruptions could occur suddenly and at scale, with recovery measured in years rather than months. The discussion suggests we need to shift from debating AI's direct capabilities to modeling its complex systems impacts.