Last Monday, we had the worst traffic jam in Belgium since years, with a length totaling above 1200 km. The reason? Snowfall that started at 8:00 am sharp all over Flanders and Brussels. Driving back home very slowly in the late evening, I listened to a small radio interview on national radio with the responsible forecaster from the national weather center, who predicted the day before that road conditions would be code orange (for severe weather conditions), and even put the time down of 8:00 AM when the heavy snow would start. Question from the reporter: how were you able to do this and predict so accurate? Answer: Intuition.

Computing power vs human experience

The well-known weather center of Belgium has more than 100 years of experience in predicting the weather – and warning government for potential dangerous weather conditions. They have access to different weather simulation models, calculated by the most advanced computers in datacenters all around the world. And they probably pay a lot of money to have access to those models. These models can predict a lot, but the fact remains that Sunday evening one weatherman had the very difficult task to call out when and how much snow would fall in this little country called Belgium. And it was for a large part up to his knowledge and intuition to make that final call.

Despite all the data we have available, in control rooms, in data centers, in business, in marketing actions, there are still humans making the final calls. Sometimes data overload and simulations can conflict each other, especially in such a complex system as weather. Luckily for us, there are weathermen making the decisions.

AI in business

Even in consumer behavior, or oil and gas process plants, there are still humans doing an overall sanity check of the actions to be taken. Why? Because the digital world not always has a full context. And on top of that, we do not always give that full context to artificial intelligence so it learns from it.

More and more research is done in so-called ethical artificial intelligence and computer learning, or learning out of big data and move to thick data. Researchers like Tricia Wang for example, are helping out by enriching big data with human context, to learn beyond the numbers, showing that there are multiple perspectives into big data. Another important researcher in this field is professor Cesar Hidalgo from MIT, who also looks beyond big data and understands that you have to come down to a single unit sometimes, to understand the data better from a single unique perspective.

Lessons to learn

There are for me 2 key takeaways out of this weatherman and data crunching machines. First of all, all data needs context and a higher complexity of synthesizing, to reach a certain outcome. In the context of the weather prediction, it is the combination of the different simulation maps and the intuition of an experienced weatherman that leads to a better forecast. Slight variations of big data and computer learning can have different outcomes. Secondly, let us not travel blindly through big data analytics without digging deeper into specific data points, and learn the every changing context of some of the data points. This is especially important when the subject is human behavior. As such, AI and Humans will work together for a better future and this can bring us back to the initial laws of robotics written by Asimov in 1942.