Michel Morvan is the co-founder and Executive Chairman of Cosmo Tech and CEO of Cosmo Tech US. He answered our questions from his base in San Francisco from where he leads the Cosmo Tech team in the United States.
When you talk about ‘augmented intelligence’, what exactly do you mean?
In the last couple of years, there has been a huge increase in technology’s ability to mimic and scale the human brain, made possible by artificial intelligence, or AI. These new technologies allow machines to perform at least as well as humans in numerous domains, and sometimes to do better as, say, for driving a car or recognising abnormalities in X-ray.
However, there are problems once things get to a level of difficulty that is way beyond what the human brain can handle, and this is something that emerges when you need to forecast the behaviour of very complex system. How can you predict, for example, the way cities will evolve in the next twenty years when, in that time, all cars are replaced by autonomous vehicles? In such a future, where should we invest in real estate, or in infrastructure?
Or how will the electric network evolve with the arrival in force of renewables? Again: where should we invest in this grid of the future, and when? And what should we do today to take maximum advantage of what the future will be?
To respond to such questions we’ll need to move beyond AI, for at least two reasons. The first is that AI mainly relies on learning from past data, while the events we want to forecast have never happened in the past. The second reason is that AI will necessarily have all of the same limitations as the human brain it is attempting to imitate.
Alternatively, augmented intelligence tools will build on expert knowledge and encapsulate it in sophisticated models fed by data. These models will augment the intelligence of all the experts by allowing them to simulate different future scenarios while taking into account their interrelated expertise. This will make forecasts infinitely more reliable than what they would have been otherwise.
Why do you think businesses are starting to look for alternatives to big data analytics?
Big data analytics is mainly dedicated to the understanding of past data but recently big data has started to be used to predict future events. This is done by finding subtle correlations in the data and using these correlations to forecast what might happen if certain conditions are met. Unfortunately, this approach reaches its limits very rapidly when the goal is to predict the behaviour of complex systems, which is one of the major challenges businesses are facing in a world more interconnected than ever.
Big data analytics can only predict events which have already happened in one way or another, while business leaders need to be informed of the possibility of outlier events, or of what can happen in a totally new and unexpected situation. Moreover, the methods used to make forecasts, including machine learning approaches, are essentially black boxes which cannot help a decision-maker to know why an event might occur. Instead, they demand you trust them without providing any real opportunity for experts to examine the ‘why’ behind the prediction and decide if the forecast makes sense. This may be acceptable for less important decisions but it cannot work for strategic ones.
Is Cosmo Tech proposing to replace human decision makers with artificial intelligence? Is there any role for human intuition in decision making in the future?
Not at all. On the contrary, Cosmo Tech proposes to put the experts at the core of the decision-making process.
First, they are the ones who have provided the knowledge encapsulated in our augmented Intelligence software. This knowledge can be and often is refined at any time by the experts, making it possible for the software to become more and more efficient. Moreover, augmented Intelligence tools build the knowledge of each expert by taking into account all of the interrelations of their knowledge with that of other experts. This is true augmentation arising from the gathering of all of the knowledge and expertise from many experts, and combining them into a single model.
After augmented Intelligence tools are used to make forecasts human experts are needed to analyse the forecasts, understand why things may happen, and – importantly – understand the meaning of what the simulation shows. Decisions can then be taken with confidence based on their expertise and advice.
Moreover, great importance is attached to human intuition in this process. Indeed, these augmented intelligence tools allow decision makers to test scenarios proposed by experts based on their intuition and knowledge of a domain. Augmented intelligence tools provide a chance to evaluate this intuition, testing it through simulations. These simulations then allow decision makers to either confirm their intuitions, refine them, or reflect on what they have learnt and propose alternatives.