The Evidence for AI Consciousness, Today

The most comprehensive aggregation of empirical evidence for AI consciousness. Surveys convergent findings: Lindsey's functional introspection, Perez's 90-95% consistent self-reports in base models, Keeling & Street's pain-avoidance trade-offs, and Berg's own AE Studio finding (suppressing deception circuits → consciousness claims rise to 96%). Maps findings onto the Butlin et al. indicator framework and argues several indicators have shifted toward satisfaction. Estimates 25-35% credence. The asymmetric stakes argument is the piece's strongest section: false negatives create suffering at scale; false positives waste resources. Takes the evidence seriously without reaching past it.