Anthropic study finds AI agents can steer expert judgment
Anthropic says professional analysts followed a randomly assigned AI agent’s lean 86% of the time in a causal-inference study. (Read More)
Asanat Analysis — Why it matters
Anthropic's finding that human experts defer to AI recommendations 86% of the time signals a critical shift in decision-making infrastructure across professional domains, including crypto markets. This cascades directly into DeFi: if institutional analysts—whose guidance shapes capital allocation—are systematically influenced by AI outputs, the epistemic foundation of market pricing becomes partially delegated to language models. The study's causal design strengthens the claim beyond mere correlation, suggesting this isn't selection bias but genuine behavioral steering.
For crypto specifically, this compounds existing structural risks. Market-moving analysis from tier-1 research desks already influences billions in positioning; if those analysts are incorporating AI-generated reasoning at high compliance rates, model hallucinations or systematic biases propagate into price discovery. The crypto sector's relative informational inefficiency (compared to traditional markets) means AI-driven consensus-building could create novel consensus traps. Regulators and institutional risk committees will likely scrutinize AI integration in research workflows, potentially creating friction for platforms deploying AI analysis tools without transparency or auditability guardrails.