Abstract

Applications of Artificial Intelligence (AI) are proliferating across domains. Now routinely discussed by the world’s media, AI is increasingly deployed in all sorts of everyday tasks – everything from securing financial transactions to booking a taxi. Yet despite the recent explosion of interest in AI, and a reasonable dose of hype, the concepts in machine learning which underpin many applications have a long history spanning decades. In this talk, we will illustrate the concepts in contemporary AI and look back to understand the reasons why we have reached a time when AI is starting to have a real impact. We will give an overview of how and when to use AI and show that drug discovery is not only borrowing proven AI technologies from other domains, but also stimulating the development of novel techniques, tailored to meet its challenges. We will talk about how to use AI successfully and discuss some of the many pitfalls, we will talk about the kinds problem where AI can really make a difference, some wider considerations when developing AI tools, and how to avoid the many pitfalls which can lead to unsuccessful AI projects. Looking to the future, we will discuss some of the practical challenges which remain, and the work taking place to overcome them.