AI disruption. Argh!
“Don’t get left behind. Everyone will lose their jobs. Money will become worthless. You must move now. Find the market gap. Become an AI-first organisation. Lead, lead, lead.”
There is a lot of fear, uncertainty and doubt here. But what is actually true? And, before that: what do we actually care about?
We’ve danced this dance before
Big shifts have come before. The telephone. The internet. The web. Smartphones. Social media. Crypto. VR. The details change, but the dance is predictable and familiar.
There are winners. Sometimes enormous winners. There are losers. Sometimes enormous losers. New entrants appear. Incumbents disappear. Markets fragment, then consolidate. Early adopters arrive, then the mass market, then the laggards. You choose your plays.
But consider something else. How many technological revolutions stopped spending time with people you love from being worthwhile? Did forest walks stop being relaxing? Did we stop missing people we’d lost? Did sitting beside a river or looking out over the ocean stop bringing calm? Did people stop needing help? Did trust stop mattering in a relationship?
The world has an extraordinary amount of persistence. The things we care about most have remained relatively unscathed by successive waves of technology, despite each wave arriving with promises (or threats) that everything is about to change.
Some things change enormously. Most things don’t. And knowing the difference matters, and that as biased humans we’ll predict that wrong: matters more!
Sense-making is off the charts
When markets and ecosystems are disrupted, having a good strategic play can mean benefiting from the disruption rather than merely enduring it. So of course everyone is sense-making like mad.
What’s our AI strategy? What’s our play Where’s our advantage? What happens to our sector?
We know this innovation dance. What we’re rather less good at remembering is that most people don’t win the lottery.
LLMs are wickedly clever maths built on a fairly extraordinary discovery: given enough data and enough computation, patterns themselves have enormous value. And scale creates qualitatively new possibilities. We’ve seen that before too.
Written directions became route planners. Route planners became satnavs. TomTom became Google Maps. Maps combined with millions of people’s movements became Waze. At sufficient scale, patterns become powerful.
An LLM can predict the next few words of your email astonishingly well. One uncomfortable implication is that your email was probably rather predictable. We’ll happily trust those statistics.
Yet show us the statistics saying that we probably won’t be the organisation that exploits AI better than everybody else in our sector, and suddenly we’re less enthusiastic about statistical inference. We still think we ALL might have the winning ticket! Its part of our wiring…
Smart isn’t the same as wise
Imagine a tennis tournament. There are 128 entrants. How many matches need to be played to produce one winner? We could ask an LLM. Vibe code all night (if we wanted) It could write a program to calculate it. Generate test cases. Run the code. Inspect errors. Improve its approach. Ask another model to critique the result. Put a human in the loop. Produce documentation explaining the methodology. Very smart.
Or somebody who understands the problem could say: 127. Every match eliminates exactly one player. To get from 128 players to one winner, 127 players must be eliminated. For n entrants, there are n − 1 matches. That’s not particularly smart. It’s simply true.
And truth can be trusted in a way that cleverness cannot.
An LLM can inspect a million lines of code and infer patterns from them. Retrieval can supply it with more relevant context. Agents can generate code, run tests, inspect failures and iterate. These are remarkable capabilities. But capability isn’t wisdom.
What if we went full statistics?
We already have ways of thinking about uncertainty. Wardley Mapping, for example, doesn’t pretend that we can predict precisely what will happen. It distinguishes between the uncertain moves of individual actors and the more predictable climatic patterns acting upon everyone. Markets move. Technologies evolve. Components commoditise. Practices diffuse. Competition changes.
We can anticipate some of this and make better moves because of it. But perhaps we aren’t taking that thinking far enough.
We know we cannot reliably predict what AI will look like in five years. Yet organisations are making five-year bets on it.
Meanwhile, there are things we know with vastly greater confidence. People need other people. Trust reduces the cost of cooperation. Collaboration lets groups achieve things individuals cannot. People want meaningful lives. Communities have needs. Resources are finite. Organisations eventually change, merge, decline or disappear. Our careers are finite. Our lives are finite.
Oddly, we spend much less strategic energy on some of these near-certainties than on guessing which AI capability arrives next year. So what would happen if we went full statistics? What if we built organisations, funds, technology and even our own lives around the things with the highest probability of remaining true?
Not ignoring uncertainty. Not ignoring technological change. Not refusing to place bets. Just sizing our confidence appropriately.
Perhaps the bigger game is trust
This matters for governance, organisational design and digital transformation. Of course we should try to succeed. We should use AI where it genuinely improves things. We should experiment. Learn. Anticipate. Make good strategic plays.
But success shouldn’t require pretending we know things we don’t. For a charity, cooperative or social-purpose organisation, the best eventual outcome might be growth. It might also be collaboration. It might be acquisition. It might be merger. It might even be closure, because the need it existed to address has disappeared or somebody else can meet it better.
If our actual purpose matters more than perpetuating our particular organisation, all of those outcomes have to remain conceivable. That’s a different kind of confidence.
LLMs are impressive precisely because they extract useful signals from enormous quantities of uncertainty. We trust the mathematics enough to let a probabilistic machine finish our sentences. Perhaps we should become equally comfortable applying probabilistic thinking to ourselves.
Our predictions aren’t certainties. Our organisations aren’t permanent. Our strategies aren’t truths. Our cleverest ideas aren’t necessarily wisdom.
Technology will keep changing. Markets will keep moving. There will be winners and losers, bubbles and breakthroughs, mergers and failures. We’ll make bets because sometimes making a bet is exactly the right thing to do. But underneath all that movement are slower, deeper currents.
People will need one another. Trust will matter. Cooperation will matter. There will be people who need help. Time will remain finite. A forest will still be a good place to walk. And sometimes the most sophisticated strategic move isn’t finding the cleverest answer to an uncertain question.
It’s noticing what was already true.
I’m feeling very philosophical heading towards an event this week about Trust and AI. Made me reflect, and I think there’s so e strategies here that are untapped - and naturally I think I’ll find them, and the winning ticket! 🙄