The deeper problem is not just automation but ambiguity about who owns decisions, accountability and outcomes.
Google's SRL framework provides a step-by-step "curriculum" that makes LLMs more reliable for complex reasoning tasks.