Risks Before Rewards
Most AI conversations in a boardroom die the same way: someone leads with the upside, and every head in the room quietly files objections until nothing lands. This module runs the other direction. Name the risks candidly, address each one with its mitigation — and only then does the rewards section unlock. The order is the point.
Risks first, rewards last — always. Address the concerns before the upside and the room is actually open when the upside arrives, instead of stuck on "but what about…". And because rewards come last, the section ends with the upside fresh in everyone's mind — not the fear.
How to run this — facilitator guidance
The sequencing principle, stated as a principle
Inoculation before persuasion. An audience that has heard its own objections named — by you, candidly, before they had to raise them — stops defending and starts listening. If you lead with rewards, every risk you skip becomes a silent counter-argument running in someone's head. Say the rule out loud at the start:
Timing for a 35-minute run
- 0–5 minRead every risk card aloud, unflipped. No solutions yet. Let the room sit with the full list. Ask: "What's missing?" — add it to the deck.
- 5–20 minFlip one card at a time. Before marking it addressed, ask the room: "Would this mitigation actually satisfy you?" If yes, capture what it means for this organisation in the note field. If no, do not mark it — an unearned green tick defeats the whole exercise.
- 20–25 minThe rewards unlock. Read them in board language — loss ratio, expense ratio, retention. Point out that each one only lands because of a risk the room just addressed.
- 25–35 minThe risk-landscape talk track below. This is the board-level insight of the session — deliver it standing up, off the diagram.
Rules of the room
- Never skip ahead to rewards, even if the room asks. Especially if the room asks — say why: "the lock is the lesson."
- Don't rush the discomfort. The risk cards are written to sting a little; that's what makes the mitigations credible.
- Always end the section on the rewards screen. Whatever runs over, the last thing on the wall is the upside.
Two threads to keep pulling
First: every mitigation in this deck is a management discipline, not a technical feat — the board can govern all of it without writing a line of code. You do not need to be an expert in a thing to understand how it works. Second: several mitigations (sign-off rules, approved tools, sampled error rates) only work if people actually use the tools — which is the argument for everyone in this room, board included, becoming a competent user.
The risk deck
Seven risks, written the way a candid board member would put them — no vendor softening. Flip each card to see the mitigation, decide whether it would actually satisfy this board, and mark it addressed. Add anything the deck is missing.
What the upside looks like on our statements
The risk landscape itself is shifting
Everything above treats AI as a risk to this organisation. The larger point for an insurer is different: the risk environment you underwrite in is changing. The obvious layer is growing digital and cyber exposure — in the lines you sell and in your own operations. The deeper layer is this: technology now changes faster than the institutions around it can adapt, and the gap between those two lines is where new insurable risk is born.
The thesis
Schools, regulators, municipalities, courts — the institutions society runs on — adapt at the pace of budget cycles and election terms. Technology no longer waits for either. That gap is not a technology story; it is a risk story, and risk is your business.
"For two hundred years, when technology changed, institutions caught up within a generation. This time the technology moves in months and the institutions still move in decades. Everything you insure sits inside institutions — so the question isn't whether this gap touches your book. It's where first."
Knock-on one: education, then your workforce
Schools have not figured out how to thrive with these tools — many are still deciding whether to ban them. A cohort is moving through the system right now during that confusion. The plausible consequence is a future labour shortage of a new kind: not fewer bodies, but entrants with weaker fundamentals — writing, numeracy, judgment formed by doing the work unassisted.
"The adjusters, inspectors, and underwriters you hire in 2035 are in classrooms today, in schools that haven't worked out what these tools are for. If they arrive with weaker fundamentals, that's not an education-policy problem — that's your claims accuracy, your inspection quality, and your training budget."
Knock-on two: government, then your loss runs
Insurers underwrite on a quiet assumption: that governments do their part. Sewers get maintained, development gets controlled, permits get inspected, codes get enforced. During the adaptation period those services will be inundated — by change, by demand, by their own tooling transitions — and services under pressure have a real chance of adapting badly. Every one of those failures lands somewhere on a loss run.
"Every policy you write assumes somebody else's institution did its job — the sewer was maintained, the inspection was real, the development wasn't on a floodplain. If the next decade inundates those institutions, some of that work gets done badly or late. The water backup claim doesn't say 'institutional lag' on it. But that's what it is."
"So the question for this board is not just 'how do we use AI.' It's 'what does our book look like in a world where the institutions around it run ten years behind — and are we pricing for that world, or the old one?'"