Why fears of automation driving unemployment are overblown

It seems that a $5.3 billion profit announcement is not enough to boost your share price these days. You have cut your workforce as well. In the banking sector that is made possible by accelerating automation of routine transactional tasks.

Thousands of jobs are set to be obliterated by online banking with the National Australia Bank the latest bank to announce it will cut deeply into its workforce over the next few years. Around 6000 of its jobs are set to go, a net loss of 4000 after it creates 2000 new digitally-focused positions.

Stories like this come on the back of claims that automation could potentially eliminate up to half of all jobs. But is this really plausible? The latest research argues it isn’t and that may be a relief to many of you.

When Oxford academics Carl Frey and Michael Osborne released their initial analysis on the potential impact of automation and artificial intelligence on jobs in 2013, they fed an age-old obsession we have about robots taking our jobs. In this dystopian future humans are stripped of meaning, becoming victims of our robot creations.

Of course, Frey and Osborne would not have had this in mind when they prepared their report. They were trying to shed light on a question that many of us thought required reexamination in light of the steep growth trajectory of robotics and automation.

Not surprisingly, alarm bells rang around the world when they reported that 47 per cent of US jobs could be potentially automated given existing technologies. The temptation to interpret their findings as the ‘end of work’ was irresistible for many, particularly tabloid journalists.

We have come a long way since the early work of Frey and Osborne in developing an understanding of what actually might be the impact of automation and artificial intelligence on employment. There is no straightforward answer to this but one thing is very clear — we are not on the threshold of a workless future. In fact, there is no shortage of valuable work to be done when you take account of unpaid work — work that could be paid for under different circumstances.

Irrespective of the importance of this, we are not about to see the collapse of paid employment anytime soon, a reality that the latest research by Carl Frey and others is now coming to terms with.

Initially, the research on automation and artificial intelligence used detailed occupational data, expert insights on the impact of various technologies and mathematical modelling to derive estimates of the job displacement potential of existing technologies. The focus on occupations rather than tasks within occupations had serious methodological limitations, the main one being overstatement of possible impacts. The adoption of a task-based approach by some researchers suggested that less than 10 per cent of occupations were vulnerable.

When Carl Frey and others applied the new approach, they found that one in five occupations might be vulnerable. Twenty per cent is very different from 50 per cent. This needs be more widely known to counter the sensationalist narrative that dominates the debate on the impact of automation on jobs.

The research on automation and jobs had been informed by judgements about the potential to automate different occupations. This is not a perfect science as the researchers acknowledge. There is considerable uncertainty about this and whether conditions exist in the real world to systematically apply automation technologies in practice. For example, firms need to have a high level of technological awareness and the capability and resources necessary to adopt automation technologies. Many small- and medium-sized enterprises are not well positioned in this respect, limiting the pace of technological adoption and diffusion.

Adding further complexity to all of this is the incredibly influential role that global technology providers are playing. They are aggressively pushing automation and robotics, at a pace that few companies can keep up with. In the race are the German, US and Japanese automation giants with China closing in. This tech-push is generating a sort of automation ‘arms race’ between some large companies and nations seeking digital dominance in the global economy. While a faster pace of automation can be expected to flow from this over time, the barriers to entry should not be underestimated.

The apocalyptic vision of a world without jobs needs to be replaced by a more grounded perspective on the impact of automation and artificial intelligence in our workplaces. Jobs are being lost, roles are being changed, and new jobs are being created by automation and artificial intelligence. It is a complicated landscape.

John Spoehr is Director of the Australian Industrial Transformation Institute at Flinders University

@JohnSpoehr