Source Engineering: Grounding AI-based Decisions with Integrity
I explore mechanisms to build verifiable, traceable origins for “why” a model chose a particular path in it decision after solving the problem of guidance and the limited state of models awareness. This is a significant challenge to Human-AI collaboration, and is prerequisite to shift from thinking of how to accept output to actively understanding and managing the foundations of it. The “hesitation factor”, the reluctance to trust a model’s answer, is a bottleneck that may perhaps be a short-lived problem, but requires new engineering scenarios or possibilities.