Michael Hallsworth

We're seeing growing evidence of "agentic flooding", with public services and companies overwhelmed by AI agents. The risk is that pursuing reasonable individual goals suddenly leads to a collapse of access for all.

Managing this means recognizing that we're dealing with complex systems. In my new Substack post, I show how "system stewardship" could get us better outcomes in a world of self-organizing AI agents.

michaelhallsworth.substack.com/p/before-the-f…

I wrote about that strange feeling you get when someone criticizes AI, but clearly uses AI to do it. The rise of AI hypocrisy reveals that our attitudes on AI haven't caught up with our behaviors.
open.substack.com/pub/michaelhal…

Join an incredible lineup of speakers for the Behavioral Science & Policy Association's 2026 Conference! Early Bird tickets end on March 16th.

The Hypocrisy Trap is a finalist for the Foreword INDIES Book of The Year award. What's interesting to me is the category it's been nominated in!

In a way, the book IS self-help: it tries to help us be more consistent with our goals, using insights from behavioral science like cue-based planning and "induced hypocrisy". It show how companies can "design out" hypocrisy by rethinking their leadership and workflows, and how politicians can explain their intentions better.

Yet the overriding message is that the pursuit of total consistency is both futile and harmful. Pouncing on every slip feels good in the moment, but can set up an inhuman standard that breeds deep cynicism in the end.

So part of the self-help here is accepting imperfections and reserving our criticisms for when they really matter. Maybe not a typical message!

The INDIE Awards are judged by librarians and booksellers, so the shortlisting means a lot. Thanks again to @mitpress and my editors Catherine Woods Matthew Browne for their support.

“If you ask an AI what to do, and then do it, how much is it your decision? What if it’s something you would have done anyway?” A student asked me this during a break in my new AI class at UPenn.

That question bothered me because it highlights an increasingly urgent issue that I’ve started calling the “judgment gap”.

By this I mean: the growing distance between what we think judgment is and what it is in practice. In our perceptions it’s a personal act of reasoning, something we own; in practice, it’s often shared between us and AI, driven by factors outside our awareness.

This gap has always been there: for decades, behavioral science has shown how our judgments are influenced by our environment and relationships. But the rise of AI is introducing a whole new set of influences that are making the gap much harder for us to ignore. And it’s radically challenging our ideas of what “human judgment” even is.

I’ve been exploring these issues in The Judgment Gap, a new Substack that blends behavioral science, literature, and philosophy to shed light on how we judge one another, ourselves, and AI.

I started posting a few weeks ago, but wanted to build up enough material before letting people know about it. You can find highlights from the posts to date in the carousel below. If they interest you, please subscribe – I’ll be posting regularly.