Artificial intelligence (AI) is emerging at a time of convergent existential crises. Governments around the world face the mounting specter of climate change and socio-economic inequality while navigating polarized political environments. AI has been presented as a new opportunity to improve and capacitize governance in the face of wicked problems, effectively deepening a longstanding commitment to technocratic and evidence-based intervention through more sophisticated tools for gathering and analyzing data. At the same time, social justice advocates are calling for AI to shed inherent biases and promote equity, even as the capacity of AI to meet these expectations is arguably overestimated. This talk discusses the ways in which AI presents plausible novelty but ultimately deepens existing ways of defining and solving policy problems that have led to ecological and social crisis. The talk also explores how AI’s incrementalist limitations point to the need for deeper epistemic reflection in confronting wicked problems, particularly amidst a ‘soft-collapse’ and the liminal state of transition.