Mind Matters Natural and Artificial Intelligence News and Analysis
Pipe line oil and gas valves at gas plant pressure safety valve selective
Image licensed via Adobe Stock

“Ghost Work” and the Enduring Necessity of Human Labor

Contrary to popular assumptions, the greater the automation, the greater the need for human labor.

In the famous sci-fi classic Dune, there are no computers. The only computing beings are humans with drug-accelerated reasoning abilities. Strange for the sci-fi genre, where computers are often front and center.

The reason there are no computers is because they have been banned. A great uprising, called the Butlerian Jihad, decided the risk of artificial intelligence was too great. And so, the entire Dune universe decided to ban computers, due to being an existential threat to all humanity.

Dune is a prophetic book in many ways, and the Butlerian Jihad is descriptive of our current time, when academics, technocrats and presidents worry about whether the new generative AI could spell the end of humanity. But Dune, like the pundits and leaders, misunderstand AI, and technological innovation in general.

More Automation Equals Less Human Labor?

As Dune shows, concerns about technology spelling the end of humanity as we know it are not new. A couple of centuries ago, Marx was worried the Industrial Revolution would unravel the fabric of society and send all but the richest down a freefall to wretchedness. In Marx’s Fragment on Machines, he predicts that capitalism inevitably will abstract away more and more human labor until it is entirely turned into a giant machine. More than a faint foreshadowing of modern concerns about AI!

Enter the book Ghost Work, by Mary L. Gray and Siddharth Suri. The two authors conducted many interviews with the world’s crowdsource workforce. These workers are responsible for powering the AI of the biggest tech companies, including Google, Microsoft, and Amazon. While to us these companies’ services appear to work autonomously to address our every need, behind the scenes is a vast number of humans who are manipulating every aspect of the AI algorithms, from labeling content, providing training data, and even secretly stepping in when the algorithms cannot answer the users. The authors dub this unseen workforce powering the modern Internet with the term “Ghost Work”.

This is a fascinating insight, which addresses the threat of human extinction from automation. Do we need to start a Butlerian Jihad and ban AI (and computers) for the safety of all humanity? Was Marx right? Will AI eliminate a massive number of jobs, and put most humans out on the street? Can the technotopias blame the rise of AI for their runaway homelessness problems? Ghost Work shows the answer to all these questions is a resounding No.

To understand why, Gray and Suri introduce the concept of “piecework”. Piecework originated during the industrial revolution to address the gap between what was promised, and what was delivered. The industrial revolution was founded on exactly the same premise as AI, that automation would eliminate the need for human labor. However, what the titans of the revolution discovered was that automation was always an 80% solution. The marvelous industrial factories, despite the ingenious mechanisms they housed, were always unable to completely deliver a product. Human labor was always necessary to put the finishing touches on what the factories produced and make the products fit for human consumption. This labor was known as “piecework.” It was often filled by women and children, who were not covered by the same labor laws as factory workers and were consequently easier to exploit by the factory owners.

From piecework, the authors derive the “paradox of the last mile.” Starting from the Industrial Revolution and progressing all the way to the modern AI revolution the authors see this consistent 80/20 tradeoff at work with all forms of automation, regardless of era. Whenever the technocrats promise a new way to automate human labor, what instead happens is the automation opens new venues of labor that only humans can fulfill. These venues are the “last mile,” and the last mile is never automated away.

Human Labor is Still Necessary

We can think about this phenomenon by looking at a bush and a tree. Imagine that the leaves on each represent human labor. With the bush, a very large percentage is leaves. Consequently, most of the bush is represented by human labor, and we can think of this as an area of industry before automation. On the other hand, with a tree a much smaller percentage is leaves. This is an industry that has been automated. Yet, despite the fact the leaves form a smaller percentage of the tree, if we were to count all the leaves on a tree compared to a bush, the tree would have many more leaves. Thus, even though automation reduces the percentage of human labor necessary, it paradoxically increases the amount of human labor required.

This is why, contrary to the Butlerian jihadists’ expectations, the paradox of the last mile means that as automation increases, the need for human labor increases, and as a corollary, so does the need for humans in general. In our current culture, this is an extremely counterintuitive finding. Our culture has gone so far as to create an entire religion, transhumanism, around the assumption that technology will automate all our needs, concerns, and humanity. This is despite the obvious fact the opposite is occurring. Contrary to the status quo, AI, and automation in general, is not a threat to human existence, but rather a benefit that will generate the need for greater human involvement and creativity.  The few contrarians who understand what is really happening stand to benefit greatly.

Eric Holloway

Senior Fellow, Walter Bradley Center for Natural & Artificial Intelligence
Eric Holloway is a Senior Fellow with the Walter Bradley Center for Natural & Artificial Intelligence, and holds a PhD in Electrical & Computer Engineering from Baylor University. A Captain in the United States Air Force, he served in the US and Afghanistan. He is the co-editor of Naturalism and Its Alternatives in Scientific Methodologies.

“Ghost Work” and the Enduring Necessity of Human Labor