One of the most persistent technology predictions to have emerged in the past decade is that artificial intelligence (AI) will be transformative in how we go about our daily lives. If the screed of near-horizon futurology is to be believed, one of AI's most significant impacts will be on the way we work. Opinions are divided between, at one end of the spectrum, robots taking over from the human workforce and fragmenting the labour market while, at the other, the forecast that we can leave iterative grunt work to computers while we get on with tasks that are more creative and rewarding, while perhaps even fitting in an extra round or two of golf.
When pictures are painted in such extremes of light and shade it's time to call in the experts, and in 'Working with AI: Real Stories of Human-Machine Collaboration' (The MIT Press, £28, ISBN 9780262047241) we benefit not just from the hard-won wisdom of two leaders in the field – Thomas H Davenport and Steven M Miller – but also of the people involved in the dozens of case studies presented that detail real-world applications of AI in the commercial, research and administrative space. What the authors call 'real stories of human-machine collaboration' come together over nearly 300 pages of analysis and insight to produce one of the most balanced narratives of AI in the workplace produced to date.
It's clear from the book that the integration of AI into the workplace is more subtle than the Armageddon-Utopia fault-line traditionally served to mass media consumers. As we make our way through the case studies – that range from digital systems for insurance underwriting to chat-based telemedicine platforms to the analysis of diesel fuel samples – what emerges is that the relative strengths and weaknesses associated with improving productivity, increasing accuracy and developing better workplace practices are context-dependent: some types of work are better suited to AI-augmentation.
With two-thirds of 'Working with AI' devoted to confirming by example that AI in the workplace has now moved beyond futuristic speculation, what we have is a ballast of evidence that the technology is "happening now to many companies and workers". But where the book really gets interesting is in their 'insights' section, that addresses the issue of change. What is the case for intelligent management systems? What are the platforms that make AI work? What personnel do we need to make it happen? How does it affect the trend for remote work or hybrid working? Most importantly: what can't AI do?
The conclusions drawn by 'Working with AI' are perhaps not surprising to the informed technologist. AI works well, and we should adjust to the idea of it becoming more familiar to more of us, say Davenport and Miller. But it's expensive, and it may be a long time before we're paid to play golf in a fully automated world of work.
The sustainable approach that will help avoid a third ‘AI winter’
Energy-hungry artificial intelligence systems are heading for a dead end. The solution is to design systems that work like the human brain, not an oversimplified version of it.