Would AI have chosen Daniel Craig?

For most of human history, decision making relied heavily on human intuition. Experience and gut feeling guided even the most consequential choices. Any data used was typically selected and shaped to justify views and decisions that had already been made. This was obviously less than ideal.

The pendulum then swung dramatically toward data in the post World War 2 era. As computational power grew in the 80's and 90's, organisations began trusting algorithms to extract insights from datasets that were growing rapidly. This brought its own limitations, particularly as we entered the era of Big Data in the 00's. This was the catalyst for moving more and more decisions out of the hands of people. Big data is too big for people to compete with algorithms. There was a belief in the 2010's that we were heading towards pure data-driven decisions. And this has happened in some areas of society, but it misses an important point - both data-driven AI decision-making and human-driven decision making have their own strengths and weaknesses. We need to think about how to combine them effectively.

Bond. James Bond...

In 2005, Barbara Broccoli made a decision that would reshape the James Bond franchise. Against considerable opposition, she cast Daniel Craig as 007. This choice drew intense criticism. For one thing, Craig didn't match the traditional Bond image. Yet Broccoli's instinct proved right. Casino Royale was a huge commercial success, and probably my favourite of the series.

Would AI have chosen Craig? AI relies on previous patterns in Data. Broccoli's choice succeeded precisely because it broke from convention. Amazon and Netflix etc. follow an algorithm-driven approach. Data models typically favour safer choices that match historical success patterns. I can't imagine they would have even considered Craig. Would TV shows like Friends and Seinfeld have survived today beyond their first season?

The streaming era illustrates the opposite extreme of the gut instinct decision-making common several generations ago. Algorithms aren't completely in control, but they do guide content decisions almost entirely. Viewer data is used to predict success, which is fine in the short term. But what about innovation? Would an algorithm have approved casting Bryan Cranston as Walter White in Breaking Bad? Some of entertainment's most celebrated decisions completely defy historical patterns.

And yet, I would say that neither extreme represents the optimal path forward. We need to think long and hard about how to combine human creativity with AI's analytical capabilities. We need to use AI as a co-pilot and trust human judgment. How? Well, it depends on the use case, but a good principle is to always build AI systems that are understandable to people. Let us provide the contextual understanding and final decision.

If we can do this, we will make better decisions than either humans or AI could achieve alone.