AI in the corporate strategy
Three classes of AI model belong in a strategy — and only one of them can produce a lasting advantage. On Westjet, OpenAI and Tesla, and the single question worth asking: how do you gain a data advantage?
Posts
How we think about AI
These four essays were written in May and June 2023, during a two-month sabbatical spent at the major AI conferences in North America. They are published here unchanged, with their original dates. Much has happened since; the arguments about strategy, data advantage and control have held up better than the examples.
AI in the enterprise: tool or agent?
Does an artificial intelligence pursue goals of its own? What that distinction means for deployment, regulation and control — with a racing AI that learns to pirouette for a reward.
From software developer to prompt engineer
From Pascal’s calculator to large language models: how the work product changed from the program to the model to the prompt — and what that means for who you hire.
Artificial intelligence: now what?
The questions that started the journey. What belongs in an AI strategy, what capability to build now, and what can wait — written at the outset, from Santa Clara.
Want to discuss any of this?
That is usually a better use of an hour than another article.