Most organizations govern AI the same way cities govern speeding. They set a rule, then punish people who break it.
Dr. Eugene Chan thinks that's the least interesting tool available. He's a behavioural scientist - founder of the consultancy Behavieural and Professor of Business at Tyndale University - and his argument is that punishment is a lagging response to a problem you could have designed out.
His analogy is a speed camera versus a speed bump. Both reduce speeding. The camera does it by fining you after the fact, and it needs maintenance forever. The speed bump does it by making the behaviour physically inconvenient, and you install it once. One punishes. The other just makes the wrong thing harder to do.
Apply that to shadow AI, or to the human-in-the-loop review that everyone claims to have and nobody actually does, and the conversation changes shape entirely.
Host Susan Diaz and Eugene also get into why universities are harder to govern than corporations, why a single institution-wide AI rule can't work when the philosophy faculty and the accounting faculty need opposite things, why the white-font trick some professors are using is a symptom rather than a solution, and what a provost should map before writing a single line of policy.
Susan and Eugene are building a governance mapping engagement for higher education institutions together. If that's you, get in touch.
About Dr. Eugene Chan
Dr. Eugene Chan wears two hats. He's the founder of Behavieural, a consultancy that uses behavioural science to help organizations solve trust challenges - AI adoption, customer loyalty, brand and PR. He's also Professor of Business and Marketing at Tyndale University, and has taught at TMU as well as in Australia and the United States. His PhD in management is from the Rotman School.
What we get into
(00:37) Two hats: behavioural science consultancy and the business school
(01:34) Policy versus governance, and why even the consultants can't define it clearly
(02:43) Susan's working definition: policy is the best practice, governance is the daily behaviour
(03:00) Confusing access with literacy - why handing out licences isn't adoption
(04:14) AI as the internet 25 years ago, and why there's no chief internet officer
(04:55) Who actually owns AI governance? "Talk to IT" isn't an answer
(06:07) The case for the CTO, the case for legal, the case for HR
(07:38) Why universities are structurally harder than corporations
(08:14) Three groups, three different toolsets, one institution
(08:55) Why "use AI and you fail" is a more rational position than it sounds
(10:47) But banning it is not enforceable. It's like banning the calculator.
(11:38) The better question: under what circumstances should it be used?
(12:21) Why this can't be one rule for a whole university
(12:41) Susan's classroom approach - use it, then defend it
(14:53) Why that teaches a skill most working professionals don't have
(16:34) Susan on delegation, and what "senior human in the loop" means
(18:24) The white-font trick professors are embedding in assignment briefs
(19:34) Turnitin, AI detectors, and what a 30% match actually tells you
(21:35) Governing by consequence, and where that runs out
(22:42) Speed cameras, speed bumps, and a Toronto argument
(24:22) Friction instead of punishment
(25:23) Applying it directly to shadow AI
(26:19) Cameras need maintenance. Speed bumps don't.
(27:31) Susan's reframe: leading indicators and lagging indicators
(28:53) Why human-in-the-loop fails when everything's on one screen
(30:12) Five screens instead of one
(31:28) The student problem - personal devices, personal subscriptions
(34:13) What higher ed leaders should be thinking about beyond cheating
(35:22) Map it, survey it, treat it as a situation analysis
(37:38) Silos, and why this has to come from the provost level
(38:50) Why an honest map is cathartic rather than damning
(39:16) The questions to answer before you write any governance
(40:14) If AI frees up staff time, where does that time go?
(41:29) The flywheel, and why the first turn is the hardest
Quotes
"You cannot ban the use of AI. At least you cannot enforce it. It's like avoiding the use of a calculator." - Dr. Eugene Chan
"The challenge is not banning the use of AI. It's finding the right use cases." - Dr. Eugene Chan
"On paper, there is human in the loop. But in practice it fails." - Dr. Eugene Chan
"Speed cameras do work, but they require maintenance. Speed bumps? You install it once, and it's pretty much lifetime." - Dr. Eugene Chan
"Can we somehow change the environment, or change the structure, so that it still produces the desired result - without fear of fines being the primary motivator?" - Dr. Eugene Chan
"It's basically like a SWOT analysis. Understanding where you are at." - Dr. Eugene Chan, on mapping AI readiness
"People are confusing access with literacy." - Susan Diaz
"There's something deeply cathartic about mapping it out. It's neither good nor bad. It has no sentiment. It's just a picture of things as they are." - Susan Diaz
Resources
Dr Eugene Chan's consultancy: behavieural.com
Get host Susan Diaz's book, Swan Dive Backwards: wearenorthlightai.com/swan-dive-backwards
Working in higher education? Susan Diaz and Dr Eugene Chan are offering a governance readiness mapping engagement for institutions - a picture of how AI is actually being used across faculties, where the gaps are, and what governance should address first. Reach either of them on LinkedIn - Susan Diaz, Dr Eugene Chan
Enjoyed this? The podcast isn't sponsored or funded. Ratings are how people find it. Please leave us a rating
