Forecasts eliminate complexity. The basic approach is linear projection of trends into the future based on assumptions about a handful of variables. Everything else is fixed by omission.
Complexity is reality. Deciding the next right thing for your business needs more. Business plans are good at the statics of context: if we change this one thing, what is the impact? This week, 4 articles that illuminate the dynamics: what else may be changing around you?
Why Change Fails Chapter 5 is the most coherent explanation I have read of the inherent flaws that undermine all of business theory. Mark Eddleston captures my view of methodologies, best practices and benchmarks.
“Complex systems can’t be fully understood through analysis because their behaviour emerges from interactions that can’t be captured by examining parts in isolation. Problems in complex systems often don’t have single causes. They arise from feedback loops, historical patterns, and contextual factors that resist simple diagnosis.”
Everything works. Nothing works unless it’s in context. In practice that means nothing is repeatable. Big enterprises could always afford to fit systems to their context. AI puts that within reach of smaller businesses too.
Follow this up with David Peterson’s The knowledge problem. The trouble with business school and consulting ideas is that everyone tries to invent their own theory. Michael Porter’s five forces is just Adam Smith reduced to a jumble of boxes and buzzwords.
Peterson takes economic theory and applies it to AI. Specifically, he references Hayek’s classic essay The Use of Knowledge in Society.
“His point wasn’t that central planners lacked intelligence. It was that the knowledge they needed wasn’t the kind that could be aggregated at the center, regardless of how sophisticated the aggregator became.”
AI extends the knowledge that is accessible and legible, adding untold volumes of words and images that we can compute over.
That still leaves an enormous amount of data that cannot be accessed. Some is simply not recorded. Much more is embedded in the accumulated knowledge of generations. Most elusive of all is the fleeting and time-dependent kind - valuable at a specific point in time, then gone.
AI gives us an incredible new set of tools to work with that knowledge. Those tools need to be distributed across people and organisations. Hayek tells us this cannot be centralised.
That captures why I am optimistic about AI. The idea of a central machine that knows everything and controls everything is fantasy.
Séb Krier picks up on the idea in this thread on X.
“I think that over the next five years we are likely to see both substantial progress toward something like ‘weak AGI’, i.e. systems that can do most cognitive tasks humans can do, and growing diminishing returns to raw frontier model improvement in the economic sense.”
So the tech keeps improving but with diminishing economic returns. He follows this by arguing that the tools we use to deploy AI - the harness in the bizarre current jargon - will become more important, not less.
A “harness” is just a product or service that lets a user apply AI to a particular task. In plain English, that means AI will generate a flood of opportunity for startups and new businesses. Quite a different view from the common narrative that all the value will flow to a handful of tech giants.
The idea of all the value from AI accruing to big tech is underpinned by more than AGI. The other leg of the argument is financial. AI changes the economics of tech because running models is expensive, too costly for the old VC-funded SaaS model. Or so they say.
What if those numbers are also defined by outdated assumptions? Devansh believes so and sets out why in How the Next Generation of AI Models are Going to Completely Change AI Inference.
You don’t need to understand all the technical specifics. This article alone is clear enough to see the impact. AI models will keep getting better and much cheaper to run, at least for many applications.
That will create new points of scarcity and expand the field of business applications. If AI margins improved sharply, what new opportunities would open up for your business?
Thank you for sharing my work.