Seriously? Why is this still happening?

Yes, this is a rant. But I have to get it off my chest:

I had a talk last week about how organizations can be more strategic in their AI journey.

But we ended talking about the same AI strategy trap as we have done so many times before.

So, I digested it on my last hike. Here we go:

1. AI strategy is NOT a shopping list.

In most conversations I have there is an agreement that strategy is more than tools.

But it is not a collection of AI use cases either. That is also just a shopping list.

Mapping use cases is not the same as setting direction. It's a test of what is possible.

Use cases proliferate, nothing connects them, and all you have is a long list of projects that keep you busy.

2. AI strategy is NOT model-performance evaluation.

This is a governance trap, thinking checklist evaluation of model performance and policy appliance is strategic. It's not.

Also governance does not equal model evaluation.

You fill your strategic questions with performance metrics instead of direction.

What do both have in common? They are tactical activities performing as if they were strategy.

And what happens is that the strategic question never gets asked.

AI strategy needs to address the following trade-offs:

🔹Speed vs. Explainability

🔹Local experimentation vs. Global standardization.

If there is one question AI strategy needs to answer, it's the following:

👉What kind of organization will we become through AI and who is accountable for the choices that follow?