Main content

AI - wealth generator for some, systemic inequality enforcer for others?

Cath Everett Profile picture for user catheverett February 5, 2021
While AI may offer the “biggest wealth opportunity of our lifetime”, the big question is will that wealth be limited to the technically-savvy and well-educated?

rich and poor

Rising levels of systemic inequality are becoming an “insidious side-effect of our increasingly AI-powered society” due to changes in the nature of work, according to a recent article in the Harvard Business Review (HBR).

The piece entitled ‘Algorithms are making economic inequality worse’ puts forward a number of key arguments to make its case. It posits that a new digital divide is widening the career, and therefore the wage, gap between workers who have access to higher education, leadership mentoring and job experience, and those who don’t. 

This situation is leading to the creation of a “global, low-paid, algorithmic workforce”, in which AI-powered organisations are run by “a small cohort of highly paid employees, supported by sophisticated automation and potentially millions of algorithmically managed, low-paid freelancers at the periphery”.

On the one hand, this “new AI underclass” will have few opportunities for promotion or development. On the other, it could well find itself increasingly “penalised by automated systems that determine access to welfare, lending, insurance, or healthcare, or that set custodial sentences”.

This troublesome situation has not only been accelerated by the pandemic, but could also be made worse than the other jobless economic recoveries of the last 30 years by a speculated K-shaped recovery from this latest recession, where the “prospects of those at the top soar and everyone else sees their fortunes dive”.

So what is going on here? Are things really as bleak as they appear and, if so, what can be done about it? In the opinion of Katya Almasque, General Partner at early stage venture capital firm OpenOcean UK, the dystopian viewpoint put forward by the HBR article is overly negative. 

Her belief is that it is important to assume a balanced position as technology in general, but AI in particular, is capable of generating positive change, for example by reducing unconscious bias:

We’re experiencing a shift towards a new digital economy and every time change happens, new jobs and paradigms of employment evolve. We might see systemic change, but it won’t necessarily lead to more systemic inequality…In fact, AI technology could help us build a fairer society if we all take a balanced view and work to overcome the challenges.

The impact of AI’s scale and persistence

But Applied Futurist Tom Cheesewright is not so sure. His take is that the introduction of new technologies has historically “exacerbated” systemic inequality, in the short-term at least, by enabling organizations to become more productive with fewer people, which leads to a shift in wealth distribution towards company owners. But he says:

What’s different this time is that historically these were short-term effects, with the longer term ones being notional tide-lifting, which means more economic growth, more and different jobs etc. But many people, myself included, don’t believe AI will have that effect because it’s such a general-purpose set of technologies. So will it exacerbate systemic inequality? Yes. The issue is the scale and persistence of AI. 

Another key challenge, believes Johannes Moenius, Director of the Institute for Spatial Economic Analysis at the University of Redlands in California, is that disruption was previously brought about by individual technologies, such as the steam train. This time, however, society is having to cope with “a slew of them”, ranging from blockchain and 5G to the cloud, which are all arriving together.

As a result, despite the huge opportunities such technology could bring, Moenius is concerned about the impact on society over the next 10 to 15 years - although he believes the situation should rectify itself within 25 years. He explains:

The issue isn’t that we’ll have new arrangements in future. It’s that the rollout of technology now looks like it’ll happen much faster than the creation of new labour-intensive industries. And if you roll out labour-saving technologies way faster than you can create new types of work, it leaves a gap – a skills gap, a creative destruction gap, that can destroy jobs quickly in the near future without us being able to rebuild societal structures quickly enough. It’s not an entirely new problem, but it’ll be at a different scale.

Cheesewright agrees, pointing out that the situation is being accelerated and made worse by the current pandemic-related economic crisis:

I’d love to be proven wrong, but it’s hard to see what classes of jobs are going to be created that could employ people at the scale of those already being destroyed in areas, such as retail. Broader digital technology, not just AI, is likely to eliminate two out of three jobs over the course of the next decade, but what’s going to replace them?

Low-paid service and caring roles, which focus on tackling the day-to-day admin and chores of wealthier workers, will see steadily rising demand. Creative areas, such as handmade goods and crafts, are also likely to grow in stature, while content creation skills in fields, such as gaming, mixed reality devices, and retail will likewise be highly sought after. But Cheesewright doubts any of these options either alone or in combination will be enough to make up the shortfall.

The repercussions on social inequality

And such a scenario inevitably has an impact on social and economic inequality. As Moenius points out, while “Covid has already pulled us five years into the future”, disproportionately hitting those on low incomes in the process, this situation will only “continue to decline as automation really hits”, again affecting disadvantaged communities more than anyone else. He explains:

Radical technological developments lead to a fanning out – it’s almost an economic rule. So people with higher educational levels, who can benefit from technology, will see their wages continue to increase, while those with lower educational levels will make less in real terms than the people before them. That’s mechanisation in a nutshell.

But Moenius is dubious about HBR’s contention of AI-powered organizations creating a “code ceiling” that will prevent low-income workers from climbing the career ladder because they are managed by algorithms and have little interaction with other humans, whether managers or colleagues: 

It’s a technocratic view of things and it’s too dystopian. As an economist, I say if you’re losing, or not gaining enough access to, talent, the cost is higher than the efficiencies provided by an algorithm. In fact, using AI should mean it becomes cheaper and easier to discover your hidden talent, which actually gives people an opportunity to progress. And why would you as an employer want to miss out on the opportunity to find someone smart?

Cheesewright similarly believes that the key challenge facing society does not lie in a widening gap between the educated and non-educated:

It’s more factionalized than that. It’s about the 0.1% pulling away from the 1%, the 1% pulling away from the 10%, and the 10% pulling away from everyone else. We already know that many graduates are under- or unemployed as there are fewer job opportunities around today, but in 20 years time in an AI-powered economy, there’ll be many more in that situation because, despite their advantages, job opportunities will simply have diminished. This means work is also likely to be more precarious.

In fact, in his opinion, the 2030s will be the time of the “big destruction”, although the 2020s will not get off scot-free either. Cheesewright explains:

The groundwork is being laid during this recession. As companies tighten their belts, they’re being forced to systematize, which is the first step towards automation. It’s why this period is quite dangerous. When people work remotely, the hidden inefficiencies that represent quite a big part of their day come to light, and that makes it much easier to take human work and turn it into AI work.

Substantial risks ahead

As a result, the repercussions of such a scenario could be serious. Cheesewright continues:

[Ex-President] Trump’s base felt economically disenfranchised by the changing nature of the economy as it’s become increasingly digital. But there’ll be more social unrest as more people are displaced and disenfranchised over the next 20 years, although the pendulum will swing back and forth.

Moenius agrees that “substantial risks” lie ahead:

Hopefully they can be resolved in the long-term, but they’re a major worry in the medium term. Many economic models are looking at equilibrium, but as [economist John Maynard] Keynes said: ‘In the long run, we are all dead.’ The problem is that our adjustment speeds are not in lock-step with technological developments and so my main concern is things are not really being dealt with. A potential consequence of this situation is social unrest, but it depends on how we handle the situation and how quickly politicians, employers and workers react. 

To date, Moenius says he has seen little serious discussion of the matter at a policy level, with the focus of most countries being on “winning the automation race”, although individual councils and cities have expressed concern. Moreover, he adds:

Now with COVID and the recession being front and centre, automation is being considered more of a long-run problem, although I’m convinced the timescales are now much shorter than a year ago. So it’s not an issue that’s at the forefront now, but it probably will be in a year or so.

While not a fan of either universal basic income – too expensive and lots of people define themselves by their work – or robot taxes – difficult to implement and could inhibit innovation – Moenius does believe that the creation of a negative income tax could help to “keep people’s incentive to work without letting social inequality get out of hand”.

This approach would involve the creation of a privately-managed “wealth participation fund”, into which employers would be required to contribute between one to three percent of every dollar they earn. This fund would then be used to provide workers with credits if their income falls below a certain agreed level.

As Moenius concludes:

Most economists agree we’re at the start of the biggest wealth opportunity of our lifetimes, so now would be the time to set a fund like this up to ensure that potential doesn’t turn negative.

My take

Government action and a balanced, responsible approach by employers to digitally up-skilling and re-skilling would seem imperative sooner rather than later to ensure society remains cohesive and doesn’t buckle under the strain of technological change that is too fast for existing structures to cope with.

A grey colored placeholder image