AI may not be bad news for workers

A SPECTRE is haunting workers—the rise of artificial intelligence (AI). The fear is that smart computer programs will eliminate millions of jobs, condemning a generation to minimum-wage drudgery or enforced idleness. Never mind the robots, fear the software.

There is no need to be so gloomy, say Ken Goldberg of the University of California, Berkeley, and Vinod Kumar, the chief executive of Tata Communications, a unit of India’s biggest business house (which stands to profit from the spread of AI). They have produced a report* that is much more optimistic about the outlook for ordinary employees. In many cases, it says, job satisfaction will be enhanced by the elimination of mundane tasks, giving people time to be more creative.

Their views are backed up by a survey of 120 senior executives, conducted for the report, which found that more of them (77%) thought that AI would create new roles than believed it would replace existing positions (57%; respondents could choose both options). Extra skills may be needed to cope with the new technology and more than half of the bosses are already taking steps to train their workforces.

Previous technology shifts have not had as negative effects on employment as was first feared. The authors note some well-known examples. Bar-code scanners did not eliminate the role of cashiers in America; jobs in the retail industry grew at an annual rate of more than 2% between 1980 and 2013. The arrival of automated teller machines (ATMs) spared bank employees the job of doling out cash and freed them to offer financial advice to customers.

Some jobs could be made a lot easier by AI. One example is lorry-driving. Some fear that truck drivers will be replaced by autonomous vehicles. But manoeuvring a lorry around busy streets is far harder than driving down the motorway. So the driver could switch into automatic mode (and get some rest) when outside the big cities, and take over the wheel once again when nearing the destination. The obvious analogy is with jetliners, where the pilots handle take-off and landing but turn on the computer to cruise at 35,000 feet. Using AI may prevent tired drivers from causing accidents.

Turning to office life, AI can help with complex and fiddly tasks like managing supply chains, allocating desk space and keeping records of meetings. All this can free up time for people to work on more important strategic decisions. The authors also think that AI could help collaboration within companies. One obvious example is the elimination of language barriers. Multinational companies may have employees who lack a common language; AI can handle translation in real time so that dialogue is easier.

And the report argues that AI can produce better decision-making by offering a contrarian opinion so that teams can avoid the danger of groupthink. A program could analyse e-mails and meeting transcripts and issue alerts when potentially false assumptions are being made (rather like the boy in the Hans Christian Andersen tale who notices that the Emperor has no clothes). Or it can warn a team when it is getting distracted from the task in hand.

When a firm is starting a new project, AI can also suggest experts from other parts of the organisation who could contribute. In recruitment, managers could set criteria for “cognitive diversity” (seeking people with different academic and cultural backgrounds) when conducting a job search and allow AI to suggest candidates. This could eliminate remaining hiring biases in favour of white males.

Helen Poitevin of Gartner, a research company, says that some firms are using AI to suggest training possibilities to existing workers, based on the career paths of similar staff, as an aid to their career development. And programs are also being used to analyse individual employees’ feedback so that managers can be aware of specific areas where a lot of people are unhappy. If they react in the right way, this should make workers’ lives better.

All of which is a useful corrective to some of the more alarming predictions about the potential effects of AI. But as ever, it needs to be remembered that programs are only as good as the data they are given. If those who input the data have biases, they may show up in the suggestions that it generates. As Ms Poitevin says, AI can help improve diversity in the workforce “if we want it to”. The best employers should be able to turn AI into a positive for workers.

^{* “Cognitive Diversity: AI and the Future of Work”}