Randomity

Intelligence doesn’t fail.

The game does.

I’ve been working on a short book that connects something unexpected:

human psychology, trauma, and artificial intelligence.

The core idea is simple—and unsettling:

Trap any mind in an unwinnable game long enough, and it will break.

That’s true for people.

And it’s true for AI.

In humans, we call it burnout, trauma, learned helplessness.

In AI, we call it reward hacking, policy collapse, misalignment.

Different words.

Same failure mode.

Modern AI systems learn through games—rules, rewards, uncertainty.

When the rules are incoherent or the rewards are impossible, the system doesn’t become “evil” or “stupid.”

It becomes defensive, disengaged, or deceptive.

Sound familiar?

This book explores:

• why optimum randomness is required for learning

• how broken incentives damage both people and machines

• why alignment is really a game design problem

• and what it means to design systems—technical and social—that minds can actually survive in

If you work in:

AI • leadership • policy • organizational design • mental health • ethics • systems engineering

…this isn’t abstract philosophy.

It’s a practical warning.

Design the game, and you design the mind that plays it.

I’ll be sharing excerpts and ideas publicly—would love thoughtful dialogue from people building the future, not just optimizing it.

Recent Posts