Artificial intelligence and machine learning – what are they?
Over the next month, Dr Helen Beers, technical lead at the HSE’s foresight centre at the Health and Safety Laboratory, will explain how artificial intelligence and machine learning may influence occupational safety and health in the future and the challenges this might present at work.
In the first article, Helen introduces the concept of artificial intelligence and machine learning.
To help manage occupational safety and health (OSH) it’s important that we anticipate future possibilities and keep pace with change. We can then prepare for future challenges, plan and set OSH priorities.
To help us plan and prepare it’s important to consider ‘when’ something might happen. But, determining ‘when’ is challenging, as the chance of something happening is likely to change (and unexpected events can have a high impact and wide ranging implications for the future).
Having said this, the future is not a complete mystery. HSE’s Foresight Centre helps the UK government, organisations and businesses prepare themselves for the OSH of the future. By analysing information from a range of sources and applying ‘foresight’ techniques, we can provide insights on what the future world of work could look like.
These insights are based on what we know about the past, what is driving change, and our realistic assumptions about what is likely to happen at future points in time. Anticipating what might be possible or probable in the future involves synthesis of evidence combined with stakeholder engagement, as well as a degree of creativity and imagination.
We have applied this thinking to the topic of artificial intelligence (AI), and considered how this may influence OSH in the future. So what do we know about AI (in particular machine learning) and the challenges this might pose for the future world of work?
We know that the pace of technological change is already having an impact on society and the world of work; from computer programs that are exhibiting their drawings in galleries (e.g. AARON), to artificial intelligence taking over many routine tasks in the legal profession. In fact, AI has been listed as one of the top eight technologies that will be the most influential on businesses, and have the greatest impact on the ‘widest range’ of industries in the near future.
Could AI (sometimes referred to as machine intelligence) replace humans? It’s unlikely; unlikely that a computer can distinguish whether a picture is emotionally powerful (humans find it difficult enough to define creativity), and unlikely that computers will be able to acquire uniquely human skills, such as social intelligence.
So what is AI? It might be described as the ‘science of making machines smart’; involving giving machines traits such as reasoning, planning, learning, communication and perception. Research on AI is accelerating fast and includes a number of sub-fields, for example; robotics, voice recognition and machine learning.
Machine learning is based on algorithms (or mathematical recipes) that can take in data, recognise patterns (learn from data) and improve performance based on interaction with the data (and previous results). The algorithms automatically extract knowledge and ‘learn’ from the data without being explicitly programmed; i.e. there is no reliance on rules-based programming.
A recent, and newsworthy, example of AI learning from interaction with people featured Tay, a digital virtual assistant (or ‘chatbot’) created by Microsoft. Based on conversations with humans, Tay learned slang and obscenities, repeated them online and had to be removed within 24 hours (demonstrating that whilst AI can learn from interactions, if these interactions are inappropriate, the resulting performance may be unacceptable to humans).
Like any technology, if used irresponsibly AI could cause harm, and whilst there is work underway to ensure that AI is beneficial to society, some commentators are warning that AI is developing so fast that there could be unintended consequences.
In this series of articles, we’ll consider what some of these consequences might be; with respect to jobs, skills and OSH. We’ll begin by providing some further context in terms of what we know about current trends in AI, and how it might develop in the future, along with considering what is enabling (or hindering) the current trends.
©Crown Copyright 2016, Health and Safety Executive