The chances that you have heard of Artificial Intelligence, more commonly known as AI, are high. Whether you have been closely following the development of AI through news articles or heard the words floating around your work office, you have most likely heard about it. What is interesting is that this powerful piece of technology is probably the most misunderstood area of Scientific and Technological research. There is a huge disconnect between what AI is representing and its actual definition, meaning that AI has been given both inflated expectations and a glorified position in the market. With more companies implementing artificial intelligence within their processes, we must consider what we really mean when we talk about AI because in truth, do we really know?
It would be understandable to be confused about what “artificial intelligence” actually involves since we are not always operating from the same definition and consequently, AI could be describing anything from virtual robots to deep learning algorithms. The Oxford Dictionary defines AI as “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and transaction between languages.” So, AI is essentially an attempt to make computers demonstrate aspects of human intelligence.
Let’s consider an unquestionable piece of artificial intelligence: the robot ‘Sophia’. Sophia was created by AI developer David Hanson who was dedicated to creating socially intelligent machines. Hanson utilised cutting edge machine perception to allow his creation to recognise human faces, see emotional expressions and recognise various hand gestures. Since the AI components have been combined in various ways, Sophia consistently learns from experience resulting in her responses being unique to any given situation, or to borrow from the Oxford Dictionary definition – “to perform tasks normally requiring human intelligence”. Many financial firms are claiming that they already implement AI into their process, and so does this mean that they are also modelling this advanced human intelligence?
Let’s look at two recent “AI” examples in the financial industry: Invesco has recently highlighted their implementation of the AI “chatbot” which enables participants to hold online conversations with a support person who is really an AI-aided computer. Morgan Stanley has also claimed that they use AI to study analyst reports in order to capture the companies that are going to appreciate over the coming months. Both of these examples use programming and algorithms to systematically extract information and identify associations, indicating that they are unable to learn or adapt. Even though ETF’s and chatbots are often marketed as AI, in reality there is no human intellect and so, can these really be labelled artificial intelligence? Subsequently, this raises the question…what piece of technology are these firms actually using? Is this just a marketing exercise to associate themselves with a technological buzzword and improve perception of their product?
There are two main components of Artificial intelligence: Machine Learning and Advanced Analytics, which could be described as the more precise terms for what companies are implementing into their processes. Advanced analytics includes both predictive and augmented analytics which are tools that analyse current data in order to make predications or prepare data in an efficient way. Some widespread examples of this include: fraud detection and data collection. Unfortunately, this analytical tool does not get more intelligent over time nor does it learn from the data it is given and therefore should it be branded under ‘artificial intelligence’?
Machine Learning, also referred to as ML, is another useful tool which is defined as the study of algorithms that allow computers to learn from datasets and make decisions accordingly. Machine learning is a subset for AI and is a method of training algorithms to become self-learning and adaptive. Many companies are still only in the training stage and have not yet fully developed these self-learning algorithms so again, can this always be regarded as ‘artificial intelligence’?
If companies were more accurate in what they recognised as AI, it might allow us to better understand the technologies we have around us and what actual value it might add. Artificial intelligence is not just a compilation of algorithms or a chatbot created to replicate human actions, it is a system which becomes increasingly capable with experience. To label something ‘Artificial Intelligence’, it should show high levels of human intellect which is created through the implementation of the following characteristics: pattern recognition, learning from experience, deep learning, machine learning, advanced analytics and big data.
Even though some companies are implementing some aspects of AI into their processes, many have not adopted all attributes and so have not yet fully achieved artificial intelligence nor a human intelligence quotient. We are increasingly surrounded by over-attributed inventions such as smart beta, smart motorways and even smart water. Therefore, it might be an appropriate time to sit back and ask yourself: are these actually smart creations or simply smart marketing? And similarly, is AI really AI or simply smart marketing?