It would be natural to expect high technology to flourish in the wake of Covid-19. After all, what better way to socially distance factories, schools, shops and hospitals than to fill them with robots and other AI technology. But in other respects, the pandemic could prove a brake on replacing people with machines.
To be sure, the pandemic will add pressure on businesses to cut costs through automation, particularly if increasingly cash-strapped consumers are pushed towards cheaper goods and services, he notes. Demand for industrial robots was of the only import sectors to show growth in the US during the first eight months of the year, only just lagging demand for pharmaceuticals.
“The dismantling of global supply chains, due to escalating US-China tensions, will drive automation in manufacturing as production moves back to countries where labour costs are relatively high. At the same time, we see an acceleration in the use of robots in hospitals and in other places to allow for more social distancing,” says Carl Frey, economist at the Oxford University, authority on the future of work.
In the health sector, for example, there will be increased roles for technologies like telemedicine. “You can do surprisingly much over video conferencing tools and it protects doctors and hospitals, which are hubs for spreading diseases,” Frey adds.
And warehouses, which were bottlenecks during the crisis, are also ripe for further automation.
“When you have a relatively structured environment it is more feasible to automate order picking. It is becoming increasingly feasible to automate with more dexterous robotic hands and better object tracker recognition software, as OpenAI has demonstrated. So, I think that's a domain we're going to see automation accelerate.”
Workers of the world...need jobs
Use of drones and autonomous forklift trucks in warehouses by the likes of Amazon and Walmart have risen sharply during the pandemic.
But other factors will prove a drag to the adoption of automation, not least the problems of joblessness.
“There are a lot of (income/wage) subsidies right now supporting people who have lost their jobs or trying to keep people on a payroll. The subsidies are eventually going to expire,” says Frey. “If (the pandemic) persists and people start defaulting on their mortgages, what is that going to mean for the health of our financial system and in turn, our labour market?”
Potentially making the labour market weaker still is the threat of offshoring of not just lower paid clerical positions, but higher-end jobs too.
“We are finally shifting forcibly into remote work,” Frey says. “I think certain tasks will continue to be done remotely and if things can be done remotely, well, they can be done from anywhere, right? So, if I can teach my students from my home, then Oxford University could find somebody in Delhi to do that job at a fraction of the cost and potentially even better.”
A prolonged period of slack in the labour market is likely to be compounded by the crisis on the high street as these shops lose out to online retailers, which, even though they’ll expand their own workforces, won’t take up enough staff to make up for the losses. This will keep downward pressure on wages for some time and prove a drag on the move towards automation.
Innovation needs migration
Meanwhile, the pandemic’s effects on flows of labour around the world could prove another stumbling block.
“Migration is a key driver of innovation and there are a lot of very consistent studies showing that this is true both for sending and receiving countries. The IT cluster in Bangalore for example wouldn't exist without the Indian diaspora,” Frey says. “But now, we are seeing a lot less of immigration and this is likely to continue in the future as a consequence of this situation and the dismantling of global supply chains, which have acted as a source of knowledge transmission. Whether for AI or synthetic biology, this is bad for innovation.”
There will also be less tech transfer as long as Americans remain wary of China.
“I think any American administration now want to have some ties with China but probably they are not going to want Chinese students to study machine learning at MIT and they don’t want American companies to just send technological know-how to China in return for market access. China access to Western technology is going to be significantly diminished as a result of that.”
To be sure, China is at the cutting edge of some technologies. But its surveillance state and social credit system is likely to lead to more conformity at the cost of radical innovation.
“Studies show that people in East Asia, including China, are more collectivist than their American and European counterparts. This gives them an advantage in solving coordination problems like building infrastructure or prompting a coordinated response to the pandemic but it also means less innovation if you are afraid to stand out and society does not reward you for standing out. There is plenty of evidence on this,” he says. “China excels in exploiting and commercialising new technologies, as it is already doing in e-commerce, for example, but few genuinely radical innovations have come out of modern China.”
And despite the volumes of press, “very few companies have actually adopted AI technologies,” according to Frey, and where it is used, it tends to be by superstar firms, like Amazon.
One problem is that AI is currently “extremely data-inefficient and the field is becoming increasingly narrowly focused on deep learning which is the most data-intensive branch of AI.”
“It is mainly confined to video games, or cat and dog classification, so areas where we have an abundance of data. If you want to teach a robot to empty your dishwasher, you can't do that through millions and millions of trials, right? It will just end up destroying a lot of nice porcelain,” Frey argues.
Deep learning is only suitable for a limited number of tasks until it becomes more data-efficient. “We are at the very beginning of artificial intelligence but potentially at the end of what deep learning can do alone, at least to a degree where it makes sense.”
