Foundations of artificial intelligence and machine learning

Abstract

Artificial Intelligence (AI) is an interdisciplinary field of study that focuses on building machines that are able to think and, in particular, act in an intelligent manner. In this context, it is preferable to characterise intelligence as the ability to accomplish complex tasks, as opposed to anchoring this concept on the notion of human intelligence or thought. Machine Learning (ML) is a subfield of AI that encompasses software that improves with experience. A variety of teaching methods can be used to train ML algorithms, including supervised learning, unsupervised learning and reinforcement learning. In turn, Deep Learning (DL) is a subfield of ML that uses deep neural networks (DNN) to identify useful patterns in the input data. On the technical side, this chapter draws a distinction between artificial narrow intelligence (ANI), artificial general intelligence (AGI) and artificial superintelligence (ASI). All current instances of AI fall within the realm of ANI, but the increasing power of learning algorithms is gradually shifting the balance towards the AGI paradigm. We also explored the distinction between weak AI and strong AI based on consciousness, though concerns have been raised about the relevance and falsifiability of these concepts. AI is already playing an important role in the financial industry, including in the banking, investing and insurance sectors. However, the journey towards widespread adoption has been bumpy and halted by several roadblocks. An appreciation of key historical milestones helps to put the technology’s achievements into perspective and understand some of the future directions that AI research may take.