Over the last few months I have attended a number of very interesting events, including the wonderful TEDx event in Vienna last year (https://www.tedxvienna.at/), the Big Data Forum in Berlin where I was chair (https://windpowerbigdata.com/), as well as the very interesting HTH event (https://hth-styria.com/). The latter, which took place here in Graz last week, was a fascinating experience, with a large number of Health Tech start-ups from Austria, Israel and around the world presenting their novel business ideas.
Whether the topic in question is energy, health or almost anything else in the technology domain, there is undoubtedly one common theme: Artificial Intelligence. We are well and truly at the epicentre of the hype on this one and I find this particularly thought provoking. Having spent my career focusing on the extraction of information from data, it is important that I properly understand what is going on here. Over the last 20 years I have worked with a range of statistical, physics-based and data-driven techniques. I have studied neural networks, clustering algorithms and other self-learning methods. But despite this, I must admit that the scope of A.I. is still not fully clear to me.
Don’t get me wrong, I am fascinated and excited about the potential of deep learning algorithms that can not only solve well-defined problems, but also recognise and understand the behaviour of complex systems. However, every time I hear a highly educated academic explain that all the most difficult challenges in their research were solved by A.I., I can’t help but feel a little disappointed. I fear that this “catch-all” explanation is beginning to replace proper and thorough discussion on the details, validity, limits and accuracy of our methodology.
I do believe that one day in the future, we may reach a point where massive computing power and extremely well-designed algorithms will allow us to pass even our most complex problems over to A.I. for almost immediate resolution. However, even as we approach such a utopia/crisis (?!) I believe that the human being will still have an important role to play for some time to come. Let’s consider the scenario where an A.I. algorithm performs a medical diagnosis, with the unpleasant outcome that the patient has a critical illness. Even if the algorithm is known to be 99.99% accurate, I am not convinced that the patient will feel comfortable signing up for an organ transplant without first seeking confirmation from a human professional. I believe the same applies, for example, in the analogous (but of course less emotionally significant) scenario of wind turbine maintenance. Will our colleagues in the service team exchange a gearbox based on the recommendation of a predictive algorithm, or would they rather rely on their own experience and intuition?
I very much like to play the game of chess. Indeed, I see this as a necessary claim, in order to confirm my status as a nerd. One of my favourite chess facts is that in 2005, Steven Cramton and Zackary Stephen, with world ranking 1685 and 1398 respectively, beat the grandmaster Vladimir Dobrov and his teammate in a freestyle chess tournament that allowed players to use computers to support decision making. Despite being heavily out-ranked, Cramton and Stephen were victorious due to their superior ability to use the computer as a tool within their overall strategy.
I believe and hope that the greatest contribution of computer science, at least in the near future, will be in the form of augmented intelligence. Computers and algorithms should be used to support our brains, relieve us of repetitive tasks and free our minds to do what we do best: imagine things that do not yet exist and then work out how to create them. Let’s not give up completely trying to understand how complex systems work. After all, that is what attracted most of us to the study of technology in the first place. Instead, let’s take the time to explain to one-another how our algorithms work, in order that we may learn together how to master this new digital world.