The future of automation

Many analysts have considered the likely impact of the future of automation on jobs.

For example, a report by McKinsey – Where Machines Could Replace Humans and Where they Can’t – suggested that 60% of all occupations could see 30% of their constituent activities automated with technologies available today.

Manual, repetitive, task-based jobs are clearly the most vulnerable to replacement. In the same report, McKinsey also noted that the roles least vulnerable to replacement include those in which physical activities have an unpredictable dimension, those which involve stakeholder engagement, and those which need expertise acquired through experience and people management.

However, as technologies further advance, automation may be able to encroach even into these areas where humans currently hold sway.

Another potentially troubling consequence of automation is the effect it is likely to have on the economic growth of developing countries. To date, they have depended on their low labour costs to provide the foundations on which they can build their manufacturing capabilities and drive economic growth.

Technological change

Notwithstanding these concerns, the technologies themselves continue to develop at speed. In our research, we distinguish between four different categories of automation: sensing, analysis, decision and actuation technologies. In each of these areas, we are seeing rapid technological change.

In sensing, for example, alongside developments in areas such as IoT, vision systems – both 2D and 3D – are becoming smarter, able to recognise people and objects and their key characteristics.

Perhaps the key trend in sensing is likely to be the move from sensing the visible to sensing at the micro and nanoscales, thereby dramatically increasing the volume and precision of available data.

In the field of analysis, we are seeing developments in automated pattern recognition, machine learning and natural language processing. Meanwhile, software developments, AI and decision theory are combining to support a new era in which decision-making can be reliably automated, even in conditions of disruption and change.

In actuation – the point at which things actually happen – the key emerging technologies are increasingly sophisticated robots and cobots with a far greater range of capabilities than anything on the market today.

Ultimately, the development of increasingly flexible robotics and AI-based decision-making is likely to lead to the emergence of more human-like automation – androids being just one manifestation of this.

Automation drivers

If that represents a broad-brush overview of some of the key trends in automation, what are the drivers for these developments in key sectors? Many of them are clearly industrial, to do with improved efficiency, productivity and accuracy.

But some of them are unrelated to market needs and address social, and environmental challenges such as ageing populations, natural disasters (precipitated by climate change) and increasing pollution.

For example, in healthcare, there is a looming crisis caused by the convergence of an older population with greater medical needs, more pressure on healthcare costs and the capability (if not the financial resources) to deliver highly personalised medical interventions based on individual symptoms and genetic make-up.

Automation could be one of the means by which we avert a healthcare catastrophe. It will proliferate both in hospitals and in the home.

Ultimately, we will see it influencing many of the processes and procedures currently carried out by medical staff, drugs and therapies – which may continue to evolve once administered – as well as implants and other body parts.

In manufacturing more generally, by 2050 we expect successful firms to look quite different. They will be capable of rapid adaptation to exploit new technologies and to be more responsive to changing markets and customer needs.

Manufacturing leaders will need to acquire a new skillset. It will no longer be sufficient for senior executives to have strategic commercial expertise – they will also need to have strong technical capabilities – or at least strong comprehension of the potential capabilities of technologies.

As the customisation and personalisation of products becomes the norm, increasing automation will be essential if manufacturing is to be cost-effective.

Tighter controls on energy materials and emissions will be another factor, driving the integration of sensed information from both inside and outside the factory.

The automated sharing of information along the product value chain will also increase to support more responsive customer service, faster deliveries and efficiencies beyond the factory.

Urban transformation

Cities represent another ‘grand challenge’. Today just over half of the world’s population lives in cities. By 2050, it is anticipated that this will have risen to two-thirds, placing enormous pressure on space, air quality, infrastructure and transport.

Over the next 30 years we will see a number of city and transport developments, ultimately leading to many daily activities being managed by automation systems such as deliveries, utility services, public transport and even policing.

At the same time, driverless and driver-assisted cars will be commonplace and congestion will need truly integrated, automated transportation environments where both airspace and ground space will be part of a controlled information network in which vehicle platooning will be essential on certain routes to manage vehicle volumes.

These are just some of the areas on which we see automation having a significant impact over the next 30 or so years. Many different technologies are needed to underpin these developments and some will develop faster than others.

However, the main barriers to future developments are arguably less technical and more societal, and as a society we need to recognise that we can influence the direction of automation developments through levers such as tax credits, R&D support and sectoral strategies.

So, with this in mind, and noting the potential impact automation will have on our lives, we should all be asking the question of not ‘can we automate’ but ‘should we automate’ and if so ‘where’?

---

Professor Duncan McFarlane

Head of the Institute for Manufacturing’s Distributed Information and Automation Laboratory (University of Cambridge).

He was also a leading member of the team that pioneered the Internet of Things.