Artificial intelligence: discussion and conclusions
In her final article on artificial intelligence and the impact on work and occupational health and safety, Helen Beers sums up the threats and opportunities.
In her previous articles Helen has looked at: what artificial intelligence is, the current trends and future developments, what’s enabling and hindering artificial intelligence, how artificial intelligence will impact on our jobs, the influence of artificial intelligence on future skills, how artificial intelligence will benefit the world of work and OSH.
Artificial intelligence (AI) is awakening fear and enthusiasm in equal measures. Some have likened the advances in AI to “summoning the devil” and there are concerns that AI threatens to end humanity. AI can scare people, perhaps due to the science fiction notion that machines will take all of our jobs; ‘wake up’ and do unintended things. However, where some see danger, others see opportunity!
This article pulls together information from a series of articles on AI and machine learning, its’ impact on the future world of work, and implications for occupational safety and health (OSH).
It’s likely that the upwards trend in capabilities of AI systems will continue; that systems will eventually become capable of solving a wide range of tasks (rather than a new system having to be built for each new problem), and that the adoption of AI within many industries will continue. Evidence suggests AI is currently unable to reproduce human behaviour or surpass human thinking; it’s likely to stay a complementary workforce tool for a very long time to come. However, steady gradual improvements in AI could reach a point where AI exceeds current expectations. The continued development of AI will depend on moral public opinion regarding the benefits and acceptability of it, on businesses continuing to gain competitive advantage from using it, and continued funding for research and development of it.
It is difficult to determine where this technology might create new jobs in the future, yet easier to see which tasks AI might take from humans. It’s likely that any routine, repetitive task will be automated. This shift to automation has happened for centuries, but what is different today is that it affects many more industries. It’s likely that we will adapt to technological changes by inventing entirely new types of work, and by taking advantage of our uniquely human capabilities.
Historically, automating a task has made it quicker and cheaper, which has increased demand for humans to carry out tasks around those which can’t be automated. In addition, rather than replacing jobs altogether, technology has changed the nature of some jobs, along with the skills required to do them. As the workplace, jobs and tasks change, knowledge will need to be updated, and skills will need to adapt. ‘Soft’ skills, such as collaboration, flexibility and resilience, will become increasingly important. The challenge will be to develop our skills as quickly as the technological advancements are being made. Therefore, we may need to ask ourselves, what the health and safety risks might be if the technology advances faster than skills required for working with it?
In the future, if over-reliance is placed on technology people could become disconnected from the process. They may cease to understand how things work (become de-skilled) or fail to appreciate how bad things are when they go wrong. Whilst an AI system can present data and recommendations, the decisions on what action to take is one for humans. However, if humans blindly follow automated instructions, without knowing how to question them, this could have negative implications for OSH.
Greater numbers of workers will be ‘new’ to their roles and tasks (with resulting implications for risk management). Therefore ongoing workforce training and re-learning will be increasingly important in the future.
In a future where benefits and risks are ‘incalculable’, it will be how humans choose to use the technology that decides whether it’s good or bad. To harness the power and benefits of machine learning we need to decide what we want machines to ‘learn’ and/or do, and what questions we want them to answer. It is clearly important that controls and goals for AI are set, and that a lot more empirical work needs to be done to gain a better understanding of how goal systems (in AI) should be built, and what values the machines should have. Once this is done, it will provide an idea of what sort of things should be put in a regulatory framework, or whether existing regulatory frameworks are robust enough.
If AI is seen to contribute to business success via enabling a better understanding of customers, along with a more rapid response to their needs, then its uptake within the world of work is likely to continue. In the future, many tasks will have the opportunity of input from AI. However, rather than replacing humans, it is the combination of AI and humans that is likely to bring the greatest benefits to the working world. Therefore, we might conclude that it will be how AI ‘interacts’ with humans that will influence its role in the future world of work. If human values are carefully articulated and embedded into AI systems then socially unacceptable outcomes might be prevented.
So, does AI present opportunity or danger? Will machines take all the jobs or create more than they destroy? Opinions on this are divided, and the reality is likely to be somewhere in between the two extremes. AI will continue to change the world of work, and workers will need to engage in life-long learning, developing their skills and changing jobs more often than they did in the past.
In the future, as humans increasingly work together with AI, the challenge for us in HSE’s Foresight Centre is to ensure that we anticipate any negative health and safety consequences, assess the risks, and share this knowledge to benefit the future working world.
©Crown Copyright 2016, Health and Safety Executive