Myths about the effect of AI on jobs are not helping us prepare for the future. James Mclaurin breaks down eight of the most widely-accepted ones. 

Estimates of how many jobs will disappear as employers ramp up their use of artificial intelligence range from less than 5 percent to nearly 50 percent. Why so much uncertainty?

Of course, predicting the future is a dark art at the best of times, but predicting the effect of AI on jobs and work is particularly challenging. Widely-accepted myths impede both individuals and the country as a whole as we work out how to make AI work well for us. 

Myth #1 AI will eliminate many types of jobs

Techno-optimists respond that AI is going to create more jobs than it destroys, but the truth is that talk of ‘disappearing jobs’ is a bit of a red herring. Usually, AI can only do parts of jobs that exist now so it will change many more jobs that it eliminates. (Different ways of accounting for this fact partly explain our uncertainty surrounding AI’s impact on jobs). So, young people entering the workforce now are less likely to be affected by jobs winking in and out of existence than they are by changes in the nature of jobs, job satisfaction, job security, pay etc. 

Myth #2 Use of AI and robotics will free up time so workers can tackle more interesting, high-value tasks

AI certainly will ‘free up time’, but we should question blanket claims that this will generally make existing jobs better. Sometimes AI will make jobs harder when, for example, it can handle simple cases but leaves the hard ones for the humans. It will sometimes automate interesting, valuable tasks just as Amazon and Spotify have automated the enjoyable task of interacting with customers about the books and music they like.  

Myth #3 AI will automate drudgery; high-value jobs requiring lengthy training won’t be disrupted or replaced

AI is good at some things people find difficult, such as memorising and manipulating large amounts of information, or spotting weak patterns in noisy contexts. Conversely, it is often not good at performing tasks humans find easy. AI is already successful at detecting various types of cancer (a difficult high-value task) but there are still no AIs that can tie your shoes. 

Myth #4 Human-centred jobs won’t be disrupted because people dont like talking to AIs 

Complex verbal and interpretative tasks will be the domain of humans for some time to come. But there are many contexts in which people prefer to deal with an AI. This is because, for example, it is embarrassing to talk to a doctor, or annoying to talk to a pushy salesperson, or because the rest home’s robotic pet is endlessly patient or because consumers think TripAdvisor really knows more than their travel agent. 

Myth #5 Successful AI inevitably replaces human labour

Good AI need not replace humans as it might meet unmet needs. Citizen AI provides simple legal advice for those unable to afford a lawyer. Similarly, New Zealand’s newly-developed Covid-19 chatbot ensures that people can get accurate information without wasting the time of busy health professionals. These systems provide new services. They don’t replace existing workers. 

Myth #6 We can design new technologies to make people more productive rather than replace them

It’s certainly true that AI often increases the productivity of workers rather than replacing them. But whether a company decides to use AI to make more product with the same number of staff or make cheaper product by cutting the wage bill, is a commercial decision, not a fact about how we design the AI. 

Myth #7 If AI leaves New Zealanders working fewer hours a week, we will be less happy and less productive

Doubt is already being cast on this idea by the various companies experimenting successfully with four-day work weeks. Compared to workers in other OECD countries, New Zealanders work long weeks. Interestingly, people in OECD countries with high GDP typically work fewer hours on average per week. Used well, AI could increase our productivity and decrease our work week at the same time, enhancing our wellbeing and enabling greater participation in cultural and community life.

Myth #8 AI will inevitably push New Zealanders into low-paid miserable jobs

All industrial revolutions have tended to increase inequality. Today’s tech titans are an echo of the industrial moguls of yesteryear. In the first industrial revolution, ordinary Britons got steadily poorer for nearly 100 years. Newly super-productive British men decreased in height by an average of 1.6 cm per decade. But increasing inequality is not an inevitable result of automation. In the long run, the first industrial revolution has been a good thing, enhancing the wellbeing and longevity of us all. This was ultimately achieved by social and political changes, running alongside technological advances. After prolonged protest that at times bordered on insurrection, governments were forced to pass new labour laws, mandating standard work weeks, weekends, and holidays, prohibiting child labour, and allowing effective bargaining about wages and work conditions so that the workers could finally afford to purchase the goods they made. 

* Professor James Maclaurin is an investigator on the AI and Law in New Zealand Project funded by the New Zealand Law Foundation. Thanks to the other investigators, Professor Colin Gavaghan, Associate Professor Ali Knott and Joy Liddicoat for comments and suggestions on this article. 

Professor James Maclaurin is a philosopher and co-director of the University of Otago' Centre for Artificial Intelligence and Public Policy.

Leave a comment