AI & Society
What should we ask of ourselves as we ask more of our machines?
I approach AI safety and governance from the human side: the judgment, attention, and independence we need to properly use increasingly capable systems, and whether the institutions responsible for governing them can actually keep up.
Oversight depends on people.
AI governance often asks whether a human remains “in the loop.” But I think there's another question we need to ask first: what happens to the human who stays there?
As we rely more on AI to interpret information, weigh options, and make decisions, we may become better at working with it while becoming
less prepared to question it.
Keeping a human in the loop only means something if that human can still disagree appropriately.
The human variable is not fixed.
Four questions I keep coming back to.
Human oversight capacity
If humans are expected to supervise AI, how do we make sure they remain capable of doing it well?
Cognitive offloading
What should we happily hand over to AI — and what abilities might we want to keep for ourselves?
The evaluation paradox
What happens if we get better results with AI while becoming worse at judging those results on our own?
Institutions and education
What can schools, workplaces, and institutions do to preserve the human capacities responsible AI use depends on?
From question to proposal.
Exploring what these ideas could become.
Research & policy
Putting the idea to the test.
I’m developing a policy proposal around sustained AI use, cognitive offloading, and human oversight: what we should measure, what the evidence can actually tell us, and what governance should do about it.
Public conversation
Bringing more people into the question.
Some of these ideas belong outside research and policy circles too. Through essays, talks, presentations, or conversation, I want to find ways of making these issues accessible without making them simplistic.