What building board games taught me about AI

In my early days in publishing I was approached by an academic who had a brief from a government department to create some board games to introduce and explain how various businesses worked. (This may not seem to be about AI, but please bear with me.)

It was really interesting viewing a business as a game. I spent a lot of time working with him and explaining the concept and operation of book publishing. Together we created a game structure.

We covered editorial, production, marketing, advertising, sales, export, subsidiary rights, distribution…  in fact, all the facets of the publishing business. These we then turned into a working demo of our finished game.

However, there was one thing that was bugging him. He asked,  “How do we build in the special factor that makes one book a bestseller? – how does an editor know it will be a success?” 

The truth is that we could not quantify that component.

Of course, there are aspects that might mitigate levels of risk, such as the author’s track record, or the popularity of particular genres, but to predict which new title would be a certain bestseller, there was no measurable factor. In the real world, it’s down to the intuition of the editor, based upon their experience and instinct.

Advertising, creativity and thoughts about AI

Later, in my career, I found myself with the task of gamifying the world of advertising (a subject closer to my heart), and I hit the same stumbling block. How do you quantify the quality of creative work and predict its contribution to the success of a future campaign? It seemed demeaning to leave it to a throw of the dice.

Again, success is down to the genius, instinct and experience of the creative director. You can build all the structure, create analytic processes that mimic real life business operations – but you are still left wrestling with how to predict the outcome,  when it depends to a major extent on the slippery business of creative genius.

Reflecting on these experiences, I realise how close is the parallel with grasping the place of intuition within AI.

The truth is we are still some way from experiencing true AI. The majority of what we see is really automation and machine learning, rather than real artificial intelligence.

What is missing in AI today (as it was in my gamifying experience), is the ability to use that learning to reason and predict. And reasoning and prediction are key to decision making. That will be the earth-shaking, leap forward for artificial intelligence.

Still only around 30% of our decision making comes from information and data that we can process, and the remainder is based upon intuition and experience. But consider the rate and volume of data processing leapt over the last few years. If we are to truly harness that data monster to support decision making, the answer looks like being about AI. 

It’s still difficult to believe that artificial intelligence is close to matching the predictive intelligence of human ability – but watch this space. Let me know what you think.