TwentyTwenty, Episode VI: Automation Acceleration

Check out the sixth episode of our podcast series, TwentyTwenty; Your Podcast for (Un)Precedented Time, here! Find us also on Spotify, Apple Podcasts and Google Podcasts.

Elizabeth Dykstra-McCarthy: Unprecedented times. Such has been the clarion call of this year. But just how unprecedented are these times? Whilst a global pandemic might not be an annual issue, many of the crises we are facing, from the erosion of civil liberties to debt crises, or from fears around technology to the shifts in geopolitics as major powers flex their muscles, these other products of trends and concerns we know only too well in which a state of global emergency has amplified. Even the pandemic itself was no surprise to those in the know. So what are these trends? Where have they come from? And what kind of path can we expect them to put us on?  

This is TwentyTwenty Vision, and I’m Elizabeth Dykstra-McCarthy with a podcast, brought to you by Foreign Brief in partnership with the Fletcher School, about how this year has accelerated global trends and the state of global crisis has made them that much more visible. 

EDM: 2020 has been a year when our relationship with technology became more of a glaring dependence. From the first days of social distancing and quarantine, our lives, professional and personal, moved online; as we move into the year’s twilight, this is hardly a novel observation. It has, however, transformed the workforce, and I don’t mean whether or not people are coming into work; but rather whether they need to come into work at all. 

Faced with restrictions on their workers’ movements, industries across the world have seen the long-term benefits of automation. Global supply chains, caught by obstacles around the world have needed automation to deliver food, manufacture masks and medical supplies, and reconfigure how to use AI to handle customer service. And for those industries that have been unable to capitalise on the advantages of robotics to remove workers from the equation, the question is more pressing than ever: could we go autonomous? Enormous projects, like from factories to open-pit mines, could employ a fraction of their current workforce and could have remained open throughout the pandemic, insulating them from systemic shocks stemming from human vulnerabilities. The motivation and the momentum are now stronger than ever. 

None of this would be possible without recent technological advances; many of the vaccines furthest along in their trials have been developed with machine learning tools to predict which parts of the virus’s structure will provoke an immune response and track COVID’s possible genetic mutations. Many pharmaceutical companies have been using AI technologies to monitor possible side effects of their COVID treatments and vaccines.  Some vaccines, such as the first-out-the-gate Pfizer model, rely on a stunningly hard to operate cold supply chain of -7 0 degree celsius. Whether or not the others require the same (the Moderna vaccine already doesn’t), it is an indicator of how much the world will rely on automation for the roll-out of any vaccine. 

Already, this may be setting off your alarm bells. If only a fraction of the workforce is needed, where would the rest of the workers go? 

Astrid Krenz: This industrial digital revolution is different from the tech revolutions that we’ve seen before. Because previous tech revolutions, many made workers more productive. And as a consequence, workers on higher wages and living standards was rising to a higher level. But this revolution is about new technologies and machines more and more replacing workers. And if more workers are replaced, there are less productivity gains left for human beings, as they are finally rather machines left at work

EDM: Dr. Astrid Krenz is a research fellow at Digital Futures at Work Research Centre at the University of Sussex, where she focuses on the economics of automation and labor.  She is an expert on the trend of companies moving production back to their home markets and what these moves do to the local labor markets. 

AK: What we see is a large rise in automation in terms of industrial robots of the past years. And figures from the International Federation of Robotics reveal that while automation was almost an unknown phenomenon in the 1970s with fewer than 3000 worldwide. The operational stock of industrial robots nowadays reached over 2.7 million and this trend has continued to increase. 

EDM: In the last 50 years robots have been used to increase production and reduce costs in industries as varied as the automobile industry, garment production and breweries. In Western economies, the number of robots in use increased fourfold between 1993 and 2003. In developed economies the impact hasn’t caused negative effects on employment – yet, according to a recent ILO study.  However, in developing economies, increased use of robots has led to a 14% drop in employment between 2005 and 2014, mostly because it allows firms to avoid offshoring to developing countries for their lower cost of labor. In other words; if it’s cheaper to use robots in developed countries, it directly takes away from jobs in developing countries where these would historically have been offshored to. In Bangladesh, automation had been slowly taking the place of garment workers since before 2018; by 2030, it is expected to gut 60% of all garment jobs by 2030.  Nor is this trend likely to stay narrow or slow down; before the pandemic hit, the World Bank estimated that 57% of all jobs could be automated in the next twenty years.

AK: It’s factories where we have low skilled labour working. And these are the workers that are most prone, are at the highest risk of being replaced by robots. So we already see this replacement taking place in various industries and workplaces worldwide. And low skilled workers will not benefit anymore. It’s rather capital, or those who own a capital that is firm owners, as well as high skilled workers that are complementary to production, like managers or engineers who benefit. 

EDM: And it isn’t just the manual laborers who lose their jobs, and the factory owners who earn more money. There are wider implications across industries and the global manufacturing sector as a whole. If we just noted that offshoring is reduced with more automation, the reverse is also true: automation has led to an increase in a practice known as re-shoring, where firms like Adidas move their manufacturing activities from areas with cheap labor costs like Southeast Asia back to Germany and the US.  Robots and AI eliminate the need for cheap labor abroad, especially in favour of their efficiency gains and the cost reductions of producing domestically.

AK: Re-shoring can be defined as the return of production processes back to the home country. So those production processes that had been previously transferred abroad or offshore. Firms are driven by the motivation to maximise profits and to produce with the cheapest input factor.

EDM: Nor is it just companies seeking to make the biggest bang for their buck; the idea of re-shoring appeals to recent trends in trade rhetoric of nationalising jobs; the ‘Made in Britain’ / Made in USA’ style policies. Politicians can appeal to their blue collar base, galvanising them with pledges of rejuvenating industries and job creation, restoring those jobs lost to offshoring. Governments might try to force firms to re-shore through tariffs on goods from competing countries; but this doesn’t pan out as neatly as expected. 

AK: What do we do in our research is to show that for three showing is influenced and reinforced by higher protectionism. So this means tariffs that are imposed on imports. This makes the transport of products or components from offshore destinations too expensive. The other thing is higher wages in foreign countries that were previous offshore destinations, and in particular about higher productivity of automation. So when the productivity of automation increases, which means that the machines robots or automation capture become cheaper over time, more firms were re-shore production processes and could use at home using cheaper robots instead of workers. 16:00 our model shows that protectionism increases the speed of re-shoring, but it does not improve the lot of low skilled workers.

EDM: In 2016, the shoe giant Adidas re-shored some of its manufacturing to Ansbach, Germany and Atlanta, Georgia.  The so-called “speedfactories”, were supposed to use automation at all steps of the process to manufacture shoes in 5 hours, as opposed to almost a month, with shipping included, at their Asian plants.  Re-shoring didn’t create new jobs in Germany or the US.

AK: For many years, added us, production of spot shoes had been offshored, mainly to China, Indonesia, and Vietnam. But then added us the new factories for trainers and Germany, in the USA. And now on these factories productions, many performed by automated processes, like best cutting robots, or 3d printers. Whereas previously in Asian factories 1000 more workers were employed. There are now only about 160 workers that were said to be employed in the new factories in the US or in Germany.

AK: The machines are able to conduct a lot of tasks that lower skilled workers would have done in traditional factories in developing countries, but they are very low skill type of tasks that are not yet be able to be replaced by machines, like putting laces into shoes, for example, this is a very complicated, very detailed task that the machines are not yet able to do. 

EDM: Garments also operate on shorter timescales, whereas fashions change from year to year, each task a machine is required to do would need to shift as well. At the moment, the machines in garment manufacturing aren’t yet able to adapt this quickly – or at least the cost of designing and producing such machines doesn’t compare to the cheap human labour. 

EDM: Adidas’s experience with re-shoring was not ultimately successful. In 2019, Adidas announced it would close its two speedfactories and would instead apply some of the technologies they used to their factories in Asia. The automation they used was only able to produce a small number of shoe models, failing to meet its ambitious goal of producing one million shoes per year. 

AK: The phasing out of low skilled workers due to automation can also be true for developing country. We already see increasing labour costs in Asian countries, like China, have increasing labour costs over the past couple of years. The trend of decreasing prices for automation Capita for the robots is also increasing over time.So what’s going to happen is, more and more robots will take place in those countries production processes as well, because we also have many Asian countries that have high developed technologies 

EDM: The question then becomes: where will these jobs go? What are the geopolitical implications of this movement of jobs/ industries? The answers here are more conjecture than fact but a certain direction of travel can be seen; the East Asian powerhouse is firing up as the automation engine. China is predicted to automate a whopping 12.5 million jobs by 2030 with one out of every three robots installed there.

And what does this mean for the world’s supply chains? Like no time before have global supply chains been under the scrutiny of our eyes: as pasta and toilet rolls were stockpiled, as medical supplies were shuttled from continent to continent and seized by manufacturing states; as the shocks from the Sino-American trade war sent ripples through tech supply chains, and most certainly as we try to imagine how a global vaccination programme might exist. And the logistical vulnerability of these supply chains is now clearer than ever. If globalisation is the heart of the modern economy, supply chains are its arteries, and they are susceptible to the slightest shock, shocks that by their very nature come when least desired. Before 2020, China had been producing approximately half the world’s face masks, but when COVID-19 swept across the country, they were unable to export necessary medical equipment like face masks in the quantities needed; and that’s before we get into the political wrangling of face masks in transit seized by various governments worldwide.

AK: So robots and machines, they are not vulnerable to a virus. We already see that digitalization has accelerated for COVID-19. And COVID-19, is likely to change production processes of the future implying an increased degree of automation and strong benefit of the machines as they are not vulnerable to a pandemic or virus. We see that people have to stay at home, they have to work in home office, if there are lockdowns in an economy, the people will not work in the factories. And if you have machines working instead, the duction can continue. This is something we saw, especially in the automobile industry. For the VW main automobile producer in Germany, was in lockdown for a couple of weeks. So if you have a highly automated production processing set the production would have been able to continue. Right. This is something that you learn from the pandemic. So obviously, in terms of the pandemic, it is beneficial to health machines during the production.

EDM: There go those alarm bells again. If VW factories in Germany or miners in Chilean copper mines aren’t needed, if their industries transition to require a select number of higher skilled automation operators and site managers and few of the hundreds or thousands of lower-skilled workers, will this lead to mass unemployment? Certainly this has been the fear. 

AK: As long as we properly train and educate workers such that they can cope with the new demands and tasks at the workplace, there will be no mass unemployment. However, we do have to begin early with education and training. We have to teach our children how to work with computers, how to write software programmes, how to understand these digital tech, knowledge, ease and how to use them.

EDM: But where might these jobs come from? It sounds like they’re disappearing but not reappearing.

AK: There will be new jobs that will be created, because there’s so many new technologies that emerge, and there are highly computerised technologies and jobs that demand knowledge in handling these technologies. I think for the short and medium run, no one should not be worried about mass unemployment. But this sort of Fourth Industrial Revolution, this specific feature of this revolution is that the technologies become more and more advanced that this replacement is taking more. 

EDM: The ‘Rise of the Robots’, a robot apocalypse which drives large swathes of workers into mass unemployment has floated around the peripheries of labour scholarship. This new technological revolution, the ‘Fourth Industrial Revolution, can sound alarmist; transforming the labour forces, social structures and the future of work as we know it. But many hearing the alarm bells have also often felt protected from the danger themselves. When these conversations began, we might only have envisioned automation as the roll-out of robots on factory floors, mass producing manufactured goods like automobiles: faster, safer, controlling for human error. We might even think of driverless cars bruising driving industries; But the automation acceleration won’t end there. As artificial intelligence, robotics and automation technologies become faster, more capable of tackling complex problems, and crucially, cheaper, these technologies will permeate into more and more industries making jobs, careers and entire workforces obsolete. Few sectors will be invulnerable to these changes and many, which might have seemed ‘unautomatable’ will fall prey to more efficient, computerised solutions. 

Think about travel agents. Twenty-five years ago, there were 34,000 travel agencies and 124,000 agents in the US alone. Each transaction would have cost around $30 and involved both travel agents and the agencies’ support staff. Now, the world of travel has been revolutionised. I can search for my own flights on my phone, compare prices on different platforms, search for hotels on Booking.com or Airbnb, compare prices, check reviews, see photos and book. Now its estimated there are 66,000 travel agents in the US, largely limited to esoteric destinations or luxury bookings. An industry decimated, some might say. Travel isn’t the first, nor will it be the last. Pharmaceuticals may well be the next; but more on that later. 

These jobs are those that we would traditionally consider protected from the onward march of the machines; white collar jobs. The truth is, as any automation expert would tell you, that any part of your job that is remotely repetitive can be automated. And rather than pay multiple humans to do a repetitive activity, where error might come into play; if there is enough demand, a machine can do it instead. Travel, pharmaceuticals, copy writing, mail delivery, traffic enforcement; legal writing; retail- the list goes on. And might include industries you’d least expect. 

Maya Markovich: Legal tech, it’s a nascent industry still, early, early in the disruptive cycle of the industry and of the legal industry and the legal tech industry as a subset of it. 

EDM: Maya Markovich serves as the Chief Growth Officer for NextLaw Lab, the legal technology catalyst for Dencons, a global law firm.  She leads their analysis of and collaboration with early-stage legal tech, partnering with startups that innovate new ways for law firms to use automation to improve services and client responsiveness, and has been recognized as one of the five influential women of legal tech in 2020 by ILTA.  

MM: The legal industry has been much slower to adopt and benefit from technology than many other industries. When nextlaw Labs was founded five years ago, as a legal tech focused innovation catalyst, many people didn’t really even know what we were talking about. Since then, things have progressed quickly. But the legal industry is still behind, the legal space is uniquely challenging, because in many ways, it really hasn’t changed since the time of the Magna Carta. Changing the way that things have always been done requires selling a 13th century guild on the benefits and mandates of the 21st century, and future proofing, you know, a tradition rich industry designed to stand on precedents, where the prevailing view is a lawyers work is always bespoke, can be very challenging.

EDM: This might sound familiar. The legal industry is far from the only one to be slow in adapting to and adopting new technologies, especially those which might endanger their jobs. But this is a train you either hop on or get hit by.

MM: There was a McKinsey study a few years back that found that across all industries, something like 60% of occupations have at least 30%, technically automatable activities, I’m sure that’s shifted a bit now, the World Economic Forum predicted automation specifically would create kind of a net increase of 58 million jobs by 2022, which is, of course, that’s beyond white collar, but that’s, you know, still a significant number, you know, tasks that professionals perform, that are high frequency but low risk, they can be found across most other industries, like insurance, accounting, financial services. And we see these types of trends across the globe with what Dencons clients worldwide are looking at the same types of issues. Also, of course, as automation capability gets better, jobs involving higher skills will probably be automated at increasing rates. I say this not as an endorsement of automation, necessarily, to make it clear, though, that it’s a very real and rapidly advancing phenomenon that’s going to impact all sectors and often faster than we think.

EDM: And workers, or those who benefit from their industries, have often resisted change. In 1589, William Lee invented a stocking frame knitting machine to relieve workers of hand-knitting. Upon viewing the machine, Queen Elizabeth I refused to grant him a patent, claiming that: “Thou aimest high, Master Lee. Consider thou what the invention could do to my poor subjects. It would assuredly bring to them ruin by depriving them of employment, thus making them beggars”.   

But it also might sound hard to imagine; what technologies could prove highly disruptive in the legal field? 

MM: So, multiple tech products can quickly identify key data to conduct analysis and draw inferences and flag terms for additional scrutiny. Some really plan to go even further by rendering recommendations and judgement calls, but then a lawyer interprets those findings, incorporates context and nuance and provides better informed legal service faster. The area of smart contracts. And that is a big deal. Because right now, there are many ways in which contracting and the practice of law that involves contracts, which is most law, in some form or another, it is very manual – wet signature, faxes back and forth ,it’s ripe for disruption, and the kind of the next level of, you know, beyond automating potential. Well, let’s keep the example to 2020 and the kind of problem which may plague the rollout of a COVID vaccine. 

MM: Something like if a pharmaceutical company has a requirement of shipping, is required to ship their products within a certain temperature range. And the contract specifies that if they don’t do  that, if it falls below or above, within the, within the course of the shipping, below or above that temperature, then it will engender either, you know, the contract is null and void, you don’t have to accept, you know, the, the delivery or a fine will be levied. 

EDM: A contract that is directly connected to the technology which monitors the criteria that the contract governs. Not only is it streamlined, but by cutting out multiple steps in the process, bypasses multiple opportunities for human input to become human error. That is typically the added value of tech: less time, greater reliability. 

MM: So what would happen in this very simplistic example is that the contract itself would connect directly to the API of the, you know, the nest or the thermostat within the shipping container. And if it goes above or below, it automatically would trigger the reaction within the contract, which, as you can imagine, eliminates just a huge number of back and forth, and keeps things moving as quickly as possible and ensures safety. Because we know how many examples there are of you know, shipments of these kinds of things getting somewhere and they’re not safe, or they’re not effective. And being able to have that backstop is going to be critical and, and a vast improvement over the current situation. Contract review can be extraordinarily time consuming, but it’s often written off of client bills. 

EDM: When something like contract review – repetitive, time-consuming – is written off of client bills, law firms can justify the poor efficiency by the high value of additional billable hours which it brings them. There is little incentive to change that system. Automation in the legal sector would seek to upend this self-perpetuating, wasteful cycle. 

MM: A client is going to think that if there’s one, person one does a task for their, on behalf of a client in 10 hours, and person two does a task, the same task in five hours, person two is going to be the favourite person of the client, and person one is going to be the favourite person of the their partner. And when you’re talking about how you excel within a career path, in law, traditionally, it’s always been defined as you know, high billable hours. And that is not what the clients want or need. And now they’re increasingly demanding a different model. And so now, as a result, tools that automate this kind of lower level work within law are going to necessitate lawyers to think differently about how they’re interacting and collaborating with their clients.

EDM: This would be a fundamental shift in the legal sector. But it echoes a wider trend around automation, one that might ring true for us more than ever in 2020. As more and more jobs are automated, it has highlighted how there are some jobs – teaching, nursing, caring, babysitting, doctoring, police work – where personal connection is something we are willing to pay a higher price for. Teaching can be automated, but we respond better to a human and, given a choice, want our children to be taught by and cared for by people, not machines. In a year where we spend increasingly less and less physical time with each other, the value of live human interaction has earned its premium.  

MM: I think we’re still yet to see the full impact of it. In the case of the legal industry, as it exists today, it’s essentially built for lawyers to run as a business, it’s not user or client-centric. So automation trends have been toward making lawyers more responsive to clients, I think there’s also now an opening for lawyers to do more experimenting with tools and processes with the psychological safety to do so. Whereas before the professions constraints really didn’t traditionally, didn’t make much room for that. So I’m optimistic that, at least for some, there will be a real shift in Outlook, that will make folks more open to that experimentation, and pave the way for the democratisation of legal services

EDM: Were such legal tech to really take off, and truly transform the legal industry, so that the work would evolve from time-intensive billable hours, to more client focused, the hope is that this might improve the democratisation of legal services; by reducing their costs, legal services could become more accessible to wider segments of the population. Legal poverty is a serious problem, last year 1.4 billion people had unmet civil or administrative justice needs, and automation could help free up and spread wider the much-needed services of lawyers. This seems an idealistic dream, with a touch of naive optimism perhaps, but certainly a potential outcome. There are a few quibbles with this utopia. First of all, reducing the emphasis on billable hours may well reduce how lucrative legal services are.

MM: When lawyers have more free time, because they’re freed up from this lower value of work – administrative, heavily administrative work – that they will be able to take on more clients. And if you add to that the fact that if a tool is doing much of the kind of the front end intake, for example, for a small family law firm or something like that, that the firm will be able to take on, hopefully more clients, and that will probably offset the fact that they probably won’t be able to charge quite as much. And so, but that will make the services more affordable.

EDM: Secondly, as with all technology, increasing automation comes with its own risks, and many of the pitfalls in technology have been ones we have been slow to see.

MM: automating a manual process that’s rife with implicit bias, which many processes, frankly, are simply codes in the bias. There are already countless examples of algorithmic bias causing serious issues and concerns, built as they often are, by homogenous groups of engineers. Things like facial recognition or predictive policing who gets a loan, there are so many examples. And I think like any emerging technology, if automation is done carelessly abused, or unregulated, or if it’s held in the hands of only a few, it can have strong negative repercussions. If whoever’s using the output from the automated process just takes the output without questioning where it came from, or how it arrived at certain conclusions. But you know, if the tech is open and transparent as to how the automated processes are storing and using the information that’s being collected, and it’s progressively democratised, it has incredible potential, it could be used as a way to monitor and enforce fairness and lending, quickly review decisions for bias

EDM: So for all its benefits there are some obvious flaws which are giving the automation industry, and its drivers, some pause for thought. But there is a wider trend at play. Yes, automation is supplanting many jobs. Yes, many more will be created. But the loss of one does not lead workers directly to gaining the other. For one thing, automation isn’t a one-stop shop, a zero sum game between humans and robots. In almost all examples of implementation, we still want the ultimate decisions to be made by people; the robots are the co-pilots, not the captains. In August, American’s Defence Advanced Research Projects Agency trialled AI algorithms in a series of simulated aerial dogfights, pitting AI against AI and AI against human pilots. The AI to come out on top beat an American air force pilot in 5 games out of five. So far the ambition isn’t to replace pilots, but to redistribute the work from within the cockpit, increasing capability rather than replacing roles. But the roles which support these human decisions are increasing by the day; eventually, a single human might orchestrate a fleet of pilot-free planes. This is but the first step on the road. 

MM: While law will never be the same, clients will need us more than ever before, not the old school kind of lawyer, but these trusted partners that can think creatively have deep knowledge of their business and have the EQ to understand their needs. And it’s hard really to say, what automation says about the value of work, ultimately, there will always be a need for those that can apply expertise to decision making, and strategic and creative tasks. I mean, this is just one person’s opinion, but I think the value is placed on high value work. But society is one thing, right? Business is another, and business always skews towards the better, faster, cheaper model, it’s nothing new. You know, from the days of the first assembly lines, automation became a forcing factor for labour to become organised. Societies and individuals and groups of workers in all professions need to reckon with what their work will mean in the future, and what it will need to mean, you know, at the global national and local level. It certainly is changing. I think automation is just one factor

EDM: At the most basic level, 2020 has taught firms a swift lesson is maximising productivity with fewer people present in person and, often, fewer people needed at all. Nor are those lessons likely to be unlearned. Automation acceleration isn’t one single fork in the road; one pivotal change, it is a journey and if previously we had opened the door and walked down the garden paths, 2020 has put us squarely on the road. 

The first step to automation is digitisation and 2020 has heralded an age of digitisation like never before. Changes in digital and technology adoption are taking place about 25 times faster than before the pandemic. We will soon see the consequences of business after business moving online; as more and more shortcuts and functionalities are found to aid and accessorise our online world. Next, these accessories will become increasingly necessary until they begin to supplant, not supplement, human jobs. The full range of automation impacts won’t be seen for several years yet, but if our interviewees had one piece of advice, it would be to take action now. Retrain the workers now for the world which will exist. 

Compounded with this trend are the economic consequences of the pandemic. Whilst automation is not a zero sum game with jobs, there are certainly some negative consequences. Whilst there will be more jobs, the kind of jobs will change; those who lost the jobs might not be the ones who gain the new jobs; these may well be the ‘left behinds’, the new lost generation. This isn’t good news, when unemployment rates worldwide are on the rise. And, since the 1980s,  in the US all jobs lost permanently to automation occurred during recessions. Another recent study found that over three recessions in the past 30 years a whopping 88% of job loss took place in “routine”, highly automatable occupations—suggesting that automation accounted for “essentially all” of the jobs lost in the crises.” 

Robots might not be taking your jobs yet; but don’t get too comfortable. Those in the passenger seat might soon be driving the car. Buckle up. 

This episode was researched and written by Dane Burrough, with Rachel Carp as the Studio Assistant. The Associate Producer for TwentyTwenty is Max Klaver, and is produced and presented by me, Elizabeth Dykstra-McCarthy. Many thanks again to our interviewees and, until next week, goodbye.