Originally published August 29, 2017

It’s absurd that we’re this worried about automation. If you can step outside of our economic norms, it should be the greatest miracle that we may be on the precipice of vastly reducing the amount of mostly menial work that needs to be done by people with finite amounts of energy, time, and life.

But in our current economic ideology, it is impossible to conceive of how a lessening in work could mean anything but a devastating effect on jobs and thus on pay and thus on life. The possibility of automation has revealed that we are more dependent than we realized on a system that requires almost all of us to work. If we devise technology that eliminates the necessity of human labor, we might become unnecessary. And in a system built on the efficiency of profit-making, we can become a remainder to be squeezed out.

That’s the nightmare that haunts the dream of automation: that life saving technology might end up destroying us. But this isn’t an article about expressing this anxiety, it’s an article about preparing for it.

Forget about the robots

Automation is often simplistically imagined as a robot humanoid stepping in to take your position, whether it be bagging groceries, filing taxes, or running a company. We can’t imagine it being done another way so we simply imagine metal hands handling fruits and veggies rather than human ones.

We have to remember that automation only causes as much anxiety and excitement as it does because it promises to be a paradigm shift. In this model, progress is not made linearly but occurs in fits and bursts, flowing along a plateau for a period of time and then bursting to a new level given some advancement in understanding. Though this model was traditionally applied to the history of science, it applies just as well to the history of technology.

The industrial age, for example, was a paradigm shift on a massive scale. Most of society and its relations were uprooted and remade in this wave of technology and its particular, political distributions. The advent of digital technology has involved another paradigm shift and because we are living through it, we can witness a number of industry and culture specific paradigms shifting in its wake.

Netflix, for instance, is a company that has expertly ridden the paradigm, going from mailing DVDs to creating a digital system that linked customers and streamed content to riding that paradigm as it buys and creates more quality content. From this, we can begin to see that in any particular industry, it will be less about robots driving to Blockbuster and picking up your VHS, and more about a hitherto difficult-to-imagine future that reinvents our imagination just as it reinvents the material way we do even simple tasks.

Automation is coming for your tasks

So, if not robots, what?

The key to understanding automation is that, even as we panic about the decay of the economy and the destruction of jobs, the process of change will occur in a more piecemeal fashion. Teaching a piece of software to do something is not an easy feat, and it does one a small dose of comfort to see leading AI companies wonder how even the operation of Microsoft Word, a feature of any entry-level worker’s resume, can possibly be done by a computer. Teaching an AI system, or more broadly, designing any mechanical replacement for a human task, will require deliberate education.

The truth of the immediate future will likely ride between the lowest doom of economic catastrophe and the highest hope of AI-augmented superhero-worker productivity. In retrospect, paradigm shifts are all-encompassing and vast, whereas in real time, the shift is often relatively slow but despite that speed, its change can be just as destructive.

Truckers will likely all lose their jobs once automated cars arrive. There are few fields more vulnerable to complete consumption by automation than trucking, largely, under this schema, because the job consists primarily of one task. If that one task can be automated, then it follows that the entire job will be.

Of course, trucking involves skills other than driving but automation is again, not robots stepping behind the wheel to do everything the trucker did but automation technologies consuming the world task by task. When we understand automation as a process of task-consumption, then we can chart the vulnerability of different jobs and industries based on how many tasks they require, the complexity and abstraction of those tasks, and the cost of each particular task. Trucking is by and large one task, that task is not very complex, and the cost is immense. That makes it a prominent target for automation.

Especially as most of this technology remains emergent, and likely expensive and difficult to implement, the cost of investment must be met by high savings. The most vulnerable tasks will be expensive ones that can be consumed most completely by an investment in technology. Automation’s consumption will increase most rapidly not when its power is established but when its costs fall enough to entice wary investors.

Though truckers are often brought up as the most vulnerable class, understanding automation as a process of task-consumption also shows that few jobs are invulnerable. When automation is understood as the replacement of simple work, it’s easy to assume blue collar workers are most vulnerable when in reality, with the help of AI, the most vulnerable work is repetitive–as tasks are repeated, AI is better able to learn and eventually replace.

So yes, your job might, on the whole, require a lot of seemingly abstract and complicated tasks but every time you find yourself doing a task again, applying the same skills again, running a mental or physical formula again–think about how that task could be replaced.

In this metric, seemingly safe, white collar office jobs begin to be vulnerable. What happens to Human Resources when chat bots can answer questions about healthcare coverage or social platforms can propagate job postings and analyze resumes? What happens to accounting when ledgers can be balanced, transactions processed, and finances audited faster and more accurately by a data-processing tool? What happens to legal when software can process all the laws on the books and discover relevant laws faster? Even the most prestigious, high-paying jobs can start to be eaten away, as machines prove better at identifying skin cancer than doctors and machines pick better stocks than consultants and investors.

Just as technology doesn’t change the world all at once, so will automation shift the paradigm of each job, each industry, and eventually society, task by task, piece by piece.

The hungry profit cycle

Even as jobs like trucking are threatened by automation, it’s essential to remember that they were already under the same underlying threat, even if that threat was not weaponized with such technology. To understand this incoming change, we have to approach it not only as technologists but as economists and political theorists.

Before automation, truckers were already under pressure to work longer hours across all manner of terrain, resulting in danger due to sleeplessness and exhaustion. This pressure is economic in that shipping companies want to strain as much value as possible from the worker and political, in that truckers rarely have protection from this regime.

At a base level, capitalism structures work and industry in the profit cycle. In this matrix, the owners and bosses use labor power to extract raw material and labor power to add value to it so that they can sell it in a market that will pay them enough to cover costs and to create profit. With profit as the end-goal, costs along the chain must, by necessity, be squeezed as much as possible. The two primary ways to save costs are to make the production process cheaper and more efficient or to recompense labor less for its work. While the former can be limited by available technology and resources, the latter can always be squeezed a little more, especially if benefits can be eliminated by hiring part-timers, the government is willing to step in to subsidize wages with welfare, and the workers themselves are effectively barred from uniting and demanding anything different.

Automation is disquieting because it promises to save costs on both ends. Automation will make the production process cheaper precisely because it can eliminate human labor, which is inefficient in comparison and, after the initial investment, more costly on a continual basis. Humans will always need a wage, even a paltry one, but other than maintenance costs, machines and software can just keep humming.

Are we creative enough for this destruction?

Some AI proponents and futurist technologists dismiss much of this, referring to the process of creative destruction as proof that humanity will outlive this crisis just as it has done others. The difference with this crisis, as I have sketched so far, is that unlike other economic transformations, this is happening more rapidly across a far greater range of industries because of automation’s ability to replace people task by task. Rather than one industry being destroyed and workers moving to another one in a newly needy value chain, automation will corrode almost all industries at varying levels such that labor will bleed everywhere.

When the CEO of Talla, for instance, a company offering enterprise chat technology that can partially automate HR, IT, and legal teams, is pressed in an interview to explain the effects selling this technology might have on the staff of the company it is sold to, his defense is that his product is meant to augment, not replace them. Even then, he acknowledges one of the sales points being the cost savings of hiring fewer people as the company grows. As, in his example, construction workers are augmented by technology that allows them to lift four times the weight they could before, companies will follow by hiring one-fourth as many construction workers as they once needed–and this pattern will befall every job and every task. As soon as any one employee is made more efficient, savings can be earned by cutting costs that have now become superfluous–including other employees.

AI and automation, in this pattern, may not be an outward ravaging but instead a rising tide, an increasing standard for skills and knowledge that drowns all but the most privileged. According to creative destruction, new industries will replace destroyed ones. This would likely still apply if we were talking about the wholesale destruction or displacement of an entire sector but instead, we are talking about almost all jobs undergoing a task-by-task optimization that makes the job as efficient as possible–until it is efficient enough to be eliminated all together.

The automation arms race

The winners of this automation arms race will be, broadly speaking, those that are best equipped to use this technology. This knowledge will require education of course and unless this particular ship is steered away from the course it has been hurtling on for decades, education will remain an immensely costly endeavor. With the capacity to learn and develop AI and automation technologies locked within the walls of expensive institutions, perhaps even favoring the ones in coastal tech hubs, entry level workers will not only be unable to learn those skills but unable to even enter the industries that could teach them. By then, the rote and repetitive tasks of the editorial assistants, the mail clerks, and the legal assistants may long be automated, leaving only the “higher level work” to be completed by higher level people.

While even the most optimistic technologists acknowledge the challenges that will result from automation, few articulate the language in which this problem can be solved. The question is ultimately not a technological one, because it goes back to the absurdity that started this post: why are we so scared that we might have to work less? That answer can only be answered politically.

Can we create enough accessible in-roads for young people to attach to the workforce and for older people to stay involved even as it changes beneath them? Must the price of education and skills continue to rise, even as the wages at the bottom fall out? Will the amount of jobs dwindle and concurrently, will the lower-skill jobs available continue to be precarious?

We are often tempted to shy away from the intentional creation of society, to treat it like an unforeseen consequence of technology and marketplaces. But despite its higher degree of abstraction and complexity, society remains a human creation. The more we see it as secondary, the more history will exceed our control, the more patterns of privilege and marginalization will continue. If technology is as progressive as its boldest opponents claim it to be, then it must also be used as a lens through which society too can innovate, as a tool which can inspire political transformation to match its technological transformation.

After all, the absurdity of our fear of automation belies the joy we could be feeling. Automation could be an incredible, almost utopian advancement if only we had the political imagination and will to match it. If we can design our society in hand with our technology, both grounded in principles of accessibility and democracy, then we could achieve something great.

The issue is ultimately not the technology but the distribution of its spoils, whether that be free time, money, or meaningful work. To profit heedlessly is to treat the distribution of what technology can create, and the society we then shape from this material, as merely an after thought. If technologists want to be leaders in the world beyond product development, and if political leaders want people to believe in their vision of the world and their ability to accomplish it, then they both need to understand how society is reproduced by what it does and the structure by which it does it. We do not have the privilege of acting in isolation.