25 Nov 2024
Is Ethical Automation Possible? (c’td)
In our last post, we took a deep dive on the role of ethics in automation at the workplace. We even made some recommendations for employing automation in an ethical and future proof way, emphasizing that it’s not necessarily a no-brainer that automation has to mean job cuts.
On the contrary, cutting jobs during a period of economic expansion is often a big mistake, leaving firms worse off than they would have been if they’d avoided cuts, and implemented smarter automation strategies that improved worker productivity instead, so that when times of economic contraction occur, they are more competitive and able to respond with an adaptable workforce.
Today we’ll focus more on the ethics of automation as it concerns technology itself. That means talking robots and Ai, of course, but also about other forms of digital automation, such as RPA, or Robotic Process Automation.
Automation: Past and Future
Most Automation is in the Past
We said we’d talk mainly about automation technology here. A picture of the future is often easier to paint with an eye on what’s past. So let’s look at automation to date, and try and extrapolate.
We said in the previous post that the majority of automation’s effects on our lives were already present. We talked mainly about the existence of home automation, like dishwashers and laundry machines, as well as running water and electricity.
But the effects of automation were even more profound than this in our working lives. Most of the work that we would have been employed to do 150 years ago, is now done automatically, by machines or computers.
If you need a little convincing, consider the following table from the landmark book The Rise and Fall of American Growth, by economist Robert J. Gordon.
Circled is the decline in farm laborers as a percentage of the American population between 1870 and 2009 (the most recent census data when the book was published in 2016). Farming alone went from the leading employer of people in America, one of the world’s most agrarian societies at the time, to employing statistically almost nobody. Meanwhile crop yields and productivity exploded, despite about 98% of agricultural jobs disappearing.
Other careers nearly disappeared over the same period: particularly domestic servants and laborers, who now make up a tiny fraction of workers, whereas the ranks of managers, and other professionals have grown enormously, more than doubling in that time. Not only that, but life expectancy, household income, home ownership, and a scad of other statistics show that life improved dramatically throughout that period.
How society adapted to major disruptions in the source of work in the past should tell us a lot about how we will adapt in the future too.
Automation And Food
As agrarian societies transformed into industrial ones, the quality and availability of food increased. That’s a really important point.
As fewer workers were employed growing food, food quality and quantity increased, while prices dropped. This seems counterintuitive at first glance, but it makes sense in the context of automation.
One of the reasons for this was that as the efficiency of agriculture increased, new production processes and economies of scale allowed for a more diverse market for new products. Canning, freezing, and processing of food allowed more of the total production to reach consumers, and that allowed the food industry to market a wider variety of foods, thanks to this bigger and more adventurous customer base.
As food production was automated, the typical diet in the developed world changed to reflect the easy availability of different foods. Here is a useful table from Gordon’s book:
As you can see, diets changed such that eggs became a much larger part of the typical diet, as did fruits. Farm animals declined in importance – particularly beef and pork, which were more expensive to automate than chicken and egg production.
The Downsides of Automation
This is not all good. Americans in particular came to rely too much on dairy products, fruits, and particularly sugar. This was not yet such a huge public health problem in 1940, but it would soon become one. World War II stimulus programs for the sugar, wheat, and dairy industries became permanent fixtures of the food landscape in America.
The same processes that brought a wider variety of goods also brought more unhealthy habits and lifestyles. Freezing and preservation allowed fast food and convenience foods to reach consumers, who were able to access it thanks to a growing individual car culture.
Thus today, the west is facing an epidemic of obesity, diabetes, and other illnesses caused by poor diet. Not by a lack of calories, but rather by an embarrassment of riches in wheat, sugar, corn, and dairy.
Automation may have freed people from farms, but it also enslaved them to what writer Eric Schlosser called “a toxic food environment,” in his hit history of the fast food industry: Fast Food Nation.
This is not even to mention the ecological effects of industrial farming, which are serious and ongoing.
Thus the problem of the developed world became rather different than one might expect: automation did not make jobs disappear over the long run, but rather produced a surplus of food goods, and that surplus began to cause serious concern to public health officials and the population. Those concerns are very much still relevant today and in need of addressing.
Past is Prologue
This is all very important, because the effects of automation, while different for each generation, can be expected to cause many of the same types of problems.
Today, automation is allowing for the ever-cheaper production of environmentally harmful products, including petroleum products like plastics, but also cars, computer chips, and paper goods, among others.
As global climate change worsens, we are developing automated driving technologies, so that people can continue to rely on individual transport rather than mass public transportation. We are even innovating ways to bring yet more plastics and disposable goods into our homes with 3D printing, same-day Amazon deliveries, and food delivery.
If the past is prologue, then we should be concerned that as manual work like driving and food preparation are automated, this reliance on large transportation networks to deliver daily conveniences will only increase, and the massive ecological footprint of those operations will continue to grow.
Google Drones Can Already Deliver You Coffee In Australia
In Australia, Amazon has piloted services that deliver fresh coffee to the doorsteps of consumers via drone. As much as the company may defend itself with promises that such services will be “carbon neutral” or will eliminate other forms of pollution, it’s not hard to see how the use of a global logistics system to deliver hot coffee to the doorsteps of 8 billion people is problematic.
Just as the Industrial Revolution turned agriculture into a major source of pollution and over-consumption-related illness, the mass logistics age is having its effect. What kind of a world will we live in by 2050? Will it be a world where most consumption is now being done via a mass logistics system that brings every daily convenience to our doorsteps?
Automation and Labor Politics
Globalization has certainly changed our consumption habits, but automation has never stopped having its effect.
As automation began to play a much larger role in manufacturing in the middle of the 20th century, the power of manufacturing trade unions in the West began to slowly diminish.
Eventually, their negotiating power either precipitated or wasn’t enough to stop the flight of manufacturing jobs away from the developed world, and into the developing one, with Japan and later China becoming major players in the global economy for the first time. Today India, Pakistan, Mexico, Brazil, Vietnam, Bangladesh, Taiwan, and Indonesia, as well as parts of Africa, are playing the same trick, and the negotiating power of Chinese and Japanese workers is under pressure.
And yet, this is not a perpetual cycle. Eventually, there will be fewer workers in the developing world than in the developed one, and as economies across the world become more mature and prosperous, birthrates will begin to decline, and the cost of labor in those places will go up, returning a great deal of bargaining power to the workers, and prompting some manufacturers to “inshore” or bring jobs back into the country where companies market their products.
This is not a thing of the future: it’s a thing of the present. From manufacturing to sales and marketing, inshoring, or relocating more of a company’s productive labor in its territory, is on the rise. Not only that, but the type of inshoring is also changing, with some evidence suggesting that companies are increasingly choosing to locate business units in their territory, but specifically in smaller and more affordable cities, rather than large economic capitals, according to one paper in the Journal of Urban Economics.
Automation and Energy
Energy is also becoming a driving concern that is causing more companies to inshore their operations, realizing that reliance on overseas production has serious geopolitical implications for a more energy-hungry economy.
Transformer-based neural networks like Chat-GPT require immense investments not only in the physical infrastructure of processor clusters and data centers to run these models, but also in the electricity that is used to train them.
These computer clusters could be analogized to “digital brains,” while the transformer based models are like “digital minds.” They are the software that the digital brain uses to answer questions in Large Language Models, or to understand physics or mathematical questions in other transformer-based systems.
The future of automation is largely connected with our ability to access these transformer systems and integrate them into user-facing applications. That way when an application needs to understand text or produce text, it can call on an LLM. When it needs to understand a physics problem or a problem specific to another area, like visual information or even music, the appropriate models can be consulted, and the outputs can be appropriately sent where they need to go.
Future consumer and business facing applications will need to draw on multiple such systems at the same time, many times throughout the day, or minute by minute, requiring a huge physical and energy infrastructure to maintain them.
That’s why companies like Microsoft are now making investments in energy production, with their plan to recommission the Three Mile Island nuclear facility, in hopes of supplying the energy the company will need to train its future generations of deep learning applications.
To a great extent, the commercialization of Ai will depend on plentiful cheap energy, and that’s a problem for national power grids and for the environment.
Old Energy Crunch
Again taking past as prologue, how will the society of tomorrow adapt to the rapid rise in energy demand from these systems?
We are facing an energy crunch in the next decade, as transportation and productivity become more dependent on electricity, and green energy becomes all the more important.
But we have been here before. 65 years ago, we faced a similar energy crunch, although the energy in question was food, instead of kilowatts. The world faced a population rise without enough food readily available to feed the expected growth.
This realization led to the “green revolution” of the 1960s, a generalized effort to increase the availability of nutritious food to the developing world. Many predicted collapse in the face of ballooning population growth. But collapse did not occur.
The above United Nations data tracks the progress of the Green Revolution since 1961, showing that as the world population continued to grow by about 50% every generation, the production of cereal grains kept pace, and exceeded this rise, while the overall use of land for cereal production remained almost unchanged.
The solutions that are now familiar to everyone, such as genetically modified crops, new fertilizers, and new food processing technologies were responsible for the survival of the developing world over the last 65 years. There were 3 billion people in the world in 1960, of which as many as half were food-insecure. Today, there are over 8 billion, with less than a third of the world being food insecure.
There is no question that food insecurity still exists, however, that is almost entirely due to political challenges, and not because of an inability for the world to grow enough food.
New Energy Crunch
The uncertainty of where we will get electrical power to satisfy the growing demand is already approaching a point of crisis. But it is often during such a crisis that large leaps forward are made.
Whereas the green revolution was about fundamentally redefining how we produce the food energy we need to survive, requiring us to deeply analyze and manipulate the genetics of plants and animals, the green energy revolution will have to do the same thing again, fundamentally changing how our energy economy works in order to end our reliance on stopgap methods of energy production, like fossil fuels.
A graph of the future production of energy, to avoid the most catastrophic effects of global climate change, will need to look similar to the graph of food production since 1961; we’ll need to produce energy even faster than our population grows, and we’ll have to do it while using as many, or fewer resources than are now being exploited.
Advances are being made in hydrogen fuel cell technology, hydroelectric, nuclear fission, and even nuclear fusion technologies. All will be needed, as the power grid of the future will need ways to produce, move, and store usable energy for applications ranging from transportation to food production and the operation of supercomputer clusters that will power Ai products.
Fusion technology in particular, though it’s often criticized as being “always 30 years away,” suffers chiefly from a lack of public investment in the massive infrastructure it requires to become workable. Yet the promise of limitless energy available from renewable sources may, in another generation, prove too tempting to continue to delay any longer.
Whatever the case, if history is a guide, the sheer economic pressure of the world’s energy needs may provide a catalyst for real leaps forward in the future.
The New Benzine: Ai and Robotics
We are often told that the free market almost magically determined in around 1860, that the availability of whale oil was waning, at precisely the moment when cities were growing geometrically in the new industrial centers of the world. The whale oil that was lighting cities could not sustain the demand.
The erstwhile transition to an oil-based economy brought forward effects that no one anticipated. Though driven early on by a desire to light our cities, oil refining created a toxic and volatile byproduct, which for many years was seen as a dangerous waste product: benzine. Yet 50 years after the transition to kerosene lighting began, benzine became the driver of a new tech revolution.
Again, it’s worth taking history as a guide. If, as it has been popular to suggest, data is the “new oil” of our hyperconnected digital world, and if we are now investing in huge energy projects that will help us to harness and use that data at scale, then what is the long-tail effect that will come from transitioning to a data-based economy?
One of the biggest problems we’ll face in the next 50 years is population decline and its effect on our global economic system. Despite the world population being larger than ever before, the developed world is set for massive declines in population growth, and the subsequent shortage of workers needed to keep the global system going.
This stark reality is made plain in Peter Zeihan’s recent bestseller: The End of the World is Just the Beginning: Mapping the Collapse of Globalization. In it, Zeihan argues that the twin crises of population bust and energy crunch will inevitably lead to a collapse in the standard of living globally.
But is that inevitable? One can suppose that, just as with the transition from whale oil to crude oil, our growing data-based economy could produce a natural solution to our population woes: robotics and Ai taking over for our complex, high-skilled workforce.
Externalities of an Automation Revolution
The need for electrical power that transformer-based Ai and robotics-based manufacturing will create, is going to see that transition pick up pace. As populations grey, and flatten their growth, we will need these systems to take over for a shortage of manpower. In our hyper-specialized, globalized world, there will be a simple choice between deglobalization and de-industrialization, or the adoption of Ai and robotics as the answer to maintaining our high level of development and integration.
Just as the oil boom created a toxic environment that took generations to become apparent as a problem, our transition to an automated global economy will do the same. Assuming our adoption of data based solutions is as rapid and far-reaching as our latter adoption of oil, the consequences to public health and safety are hard to fathom today.
The whispers of this kind of concern can already be felt in the presence of RPA and GenAi technologies in the workplace, where, of course, workers fear that they will be displaced. Yet it seems far more likely that they will simply be used to produce and to sell more products, whether or not those products are wanted or needed, and indeed, whether those products present a danger to society.
Today’s disinformation “shitposting” bot on a social media platform may be tomorrow’s Ai sales agent for a dangerous or inferior product. And one should never underestimate America in particular’s appetite for low-cost, disposable lifestyle innovations, no matter how useless or damaging.
If the promise is that the successors of Chat-GPT will be writing Hollywood screenplays and formulating brand strategies, then look out for a sea of low-quality, or poor value-add products that more readily find an audience in a hyper-targeted, globalized marketplace where instant convenience has become the norm. It won’t be that Ai does any of these things “better” than a human does. It will simply be a question of whether the things Ai makes can earn money. If they can, people will try.
We are not writing a troglodyte tract here. DoFlo is an automation company. But we are being very open and very honest about the fact that what we and our industry do will have externalities that will need to be addressed.
Whatever those externalities are, the technology isn’t going to go away. Like Pandora’s Box, automation will bring forth many demons, but it can also bring hope.
The Politics of an Automation Revolution
In the 19th century, nationalist and internationalist movements began partly as a response to the disappearance of agrarian jobs. As workers left farms to find jobs in urban centers in the booming manufacturing, packing, and shipping industries, working conditions often deteriorated. Political movements arose that preached economic protectionism, worker’s rights, and trade unionism, and governments began to take a much more active role in the economy as a result.
Innovations such as the “New Deal” of the 1930s or the Marshal Plan for Europe in the 1940s, were huge government stimulus and investment programs, largely in response to the plight of out-of-work soldiers, factory workers, and others displaced by a changing economy and a series of devastating global wars.
Those programs were highly successful, leading to what we now call the “developed world,” a place where unemployment has nearly disappeared, and economic and violent political turmoil is relatively rare.
These economic miracles did not come free of charge. They took enormous investments and many decades to develop.
Today, we face a similar wave of displacement, and we are going to face the political consequences of that displacement. Even though automation will be critical to maintaining the global system as populations wane, the transition will not be even or instantaneous. This will mean, increasingly, the appeal of authoritarian and protectionist politics, which will promise to reverse or fix the damage that labor displacement is causing.
What these political changes are more likely to do, of course, is to destabilize the global economy yet further; yet they can’t be avoided. As long as we are not yet willing to make the huge investments necessary to address the effects of a growing service economy and growing automation, we can’t do much of anything.
Automation And Workers
While the professional class has grown in the developed world, the service economy continues to rely on low-wage, low-training positions that are difficult or expensive to fully automate. This so-called “gig-economy” has largely innovated a race to the bottom in the standards as well as the pay and worker’s rights of the average temp worker, who is less likely to receive health or retirement benefits, ongoing education, worker’s compensation, or collective bargaining rights.
This should clue us in to the fact that in the short term, unchecked automation will undermine the role of workers in the overall economy, making it harder for us to maintain our standard of living, as companies innovate in order to take a larger piece of a smaller pie all for themselves.
As competition amongst logistics companies and convenience services increases, workers will continue to pay the cost of increased risk and insecurity in their employment, unless we take steps to protect and support workers in this new economy.
That is one reason that steps such as student loan forgiveness, free university education, and direct transfer payment systems such as Universal Basic Income are becoming more attractive solutions to growing problems. Developed nations do need the workers of today to be economically secure and prosperous, for those workers to be able to afford to participate in the economy of tomorrow.
Those seas of supercomputers and highly automated inshore factories won’t find customers if there is nobody with any money to buy what they make. Investing in workers means investing in a market for your products.
Some progress is being made, but the pressure from automation in every part of our working lives is not going to lessen in the near future. More must be done.
Managing the Effects of Automation
So we’ve tied many strings together: how the transition from agrarianism to industrialism, from pre-oil to post-oil, from labor-intensive to highly automated, and from local to global economics has shaped our present and will shape our future.
We’ve seen how the need to solve the problems of today, be it creating more food, or more energy, or more work, shapes the problems of tomorrow. What does this tell us about what we need to do right now? How can we better build towards a future we want to live in?
Here are a few ways we believe that companies and governments can contribute to a more stable and prosperous future in an age of automation. Some of these steps are socio-political, and some have to do with technology itself, but all will be necessary to some degree to avoid the worst effects of instability and economic insecurity that automation is necessarily going to bring.
We cannot avoid disruption. That is a fact of life, but we can decide now to build a better future for our kids and grandkids, if we really want it:
Support the Transition from a Labor-Based Economic System
As populations stop growing, and as more work is automated, the decline in the share of economic participation from workers will present big problems for the global economy. We have to begin now to think about a world where our jobs, our work benefits, and our salaries are not the driving forces in our lives.
Right now, declines in worker pay during economic contractions can contribute to outsized vectors of instability for governments and private companies. In California for example, with its emphasis on income and corporate tax as a key source of government revenue, fiscal policy guarantees that when employment dries up, the means of the state to intervene in the economy by stimulus spending is impaired. That contributes needlessly to even worse economic conditions, which further drive contraction.
In effect, the government guarantees that it will have lower revenues during times when it will be facing higher costs from programs such as unemployment and disability programs. The systems put in place to reduce the economic effects of a recession lead to political crises, and lower growth during times when the government is not interceding but is collecting more revenue than it needs.
The perennial conservative approach to this problem is to cut taxes, and to engage in austerity policies, in hopes that the instability introduced by government’s role in the economy will be lessened if the government simply doesn’t participate at all. But such policies are deeply unpopular and politically hard to sustain.
Systems such as Universal Basic Income, which we plan to talk about in a future post, are ways that we can disengage the government from its role in the employment economy, and still create a stable basis for it to contribute to the consumer economy when needed.
Support and Invest in New Green Energy
It seems obvious from what we’ve discussed so far, that energy is a key piece of a prosperous future. Any automation company that isn’t promoting a transition to green and renewable energy sources is probably not engaged enough in the problems that their technology are going to face and contribute to. Automation will require a cleaner and more efficient energy grid.
Without big advances in energy technology, we won’t be able to survive the worst effects of global climate change. Without a stable climate, we won’t be able to sustain our population. A population crash would only worsen our reliance on automation and our need for energy, so clean energy is an absolute requirement for a stable future.
Pursue Human-Centered Innovations
We like to say that we don’t mind robots taking over our jobs: as long as we, the workers, own the robots.
The recent benefits of automation have been largely enjoyed not by workers, but by the corporations that employ them. Today, automation is rightly seen by workers as a threat to their livelihoods and not the promise of greater prosperity in the future that can be shared by everyone. This leads to so-called “enshitifcaiton,” and the hollowing out of the middle class that comes with it.
It’s not enough to talk about the benefits of automation and Ai for businesses and governments, without talking about how society will benefit. If the object of automation and Ai implementation is to reduce headcount at companies, the benefits of efficiency gains will come at an unsustainable cost to the human beings at the center of the global economy. Corporations will continue to be enriched, but those riches will be ever larger pieces of a shrinking pie.
We need to focus on technology innovations that make individuals and companies more effective and competitive: not just leaner and cheaper. Generations of austerity policies in government and business have led us to believe reflexively that the only way to apply new technologies is to reduce spending and increase throughput, with the cost savings almost always coming from reducing employee numbers.
That must change. If we want any hope of maintaining the conditions of a functioning global economy, there must be jobs, and well-paying jobs, for everyone. Workers must make enough not only to survive but to prosper and to raise families so that our economy can continue to find workers and consumers to continue on after we retire.
Be More Human
Perhaps its a bit “woo-woo” to say that the answer to our present challenges is to “be more human,” but that’s also what we truly believe.
Automation is a necessary change in the way that human beings interface with the economy, as our populations peak and subside in the next century. We need automation to replace -not workers- but work itself. As we pursue that future, we need to consider the role of human capital very carefully. It is we humans who make decisions, buy things, build our lives, and carry on the light of consciousness into the future.
Removing human beings from our economy is the last thing we need to have a shot at surviving as a species. We have to place the needs of people at the forefront of everything: the way we think about everything we do, as companies, as individuals, and as cultures. Automation should mean more human opportunities; more dignified work, more freedom of choice, more safety, and more opportunity. If we are not doing that, we are not succeeding at building a future we want to live in.