Build systems where humans and AI make each other sharper

If you're reading this, you've probably already watched AI get bolted onto workflows you're responsible for — and noticed that the results were uneven in ways nobody on the team could fully explain. The model performs well on the benchmark, users are technically "in the loop," and yet something is off: people are deferring when they shouldn't, errors are slipping through at the handoff, and no one is quite sure whether the human or the AI is actually accountable for the output. That discomfort is not a soft concern. It's a precise signal that the system wasn't designed — it was assembled.

That's the gap this program is built to close. HHAI is a genuine scientific discipline, and it gives you the vocabulary and the frameworks to stop guessing. You'll learn to reason about human-AI collaboration the way a systems engineer reasons about reliability: by understanding the capability profile and failure modes of each component, and designing the interfaces and decision rights around those realities. That means engaging seriously with the human factors research on cognitive load, automation bias, and trust calibration — not as background reading, but as the foundation for concrete design choices.

The curriculum is structured around the full arc of the problem. We start with the complementarity argument — why the goal isn't to replace human judgment with AI inference, but to construct systems where each compensates for the other's failure modes. We move through the three core collaboration architectures, task allocation and handoff interface design, multi-dimensional evaluation frameworks, and the failure modes that no accuracy metric will surface. We close by connecting everything to the organizational reality: writing product specifications that actually get built, and leading adoption in teams where "we're adding AI" is not self-implementing.

I designed this program because I've seen what happens when capable technologists and researchers try to navigate this space without a shared framework. Every concept in this curriculum is earned — grounded in evidence, tested against real system contexts, and presented with the precision the problem demands. There's no hand-waving about "augmenting human intelligence" without specifying the mechanism.

If you want to understand hybrid human-AI systems at the level where you can design, evaluate, and lead them, I'd like to work with you. The problems in this space are hard, the stakes are high, and the field needs practitioners who are equipped to think clearly about them. Come build that capacity with us.