The Mechanical Elephant in the Room
By Chris Oestereich

Artificial intelligence has made quite a racket of late. Much of the attention has been focused on platforms that convert simple requests into written works and others that can turn short descriptions into vivid images. There is plenty of excitement and apprehension in the air over this seeming quantum leap in AI’s capabilities.

When I started testing these platforms, my initial reaction was disbelief. At the time, I didn’t know how they worked, so when I would type a request and hit “send” and the systems began to respond in a matter of seconds, it felt like magic.

It was a trick.

Writing is, at least for me, a challenging pursuit. It takes time, focus, and effort, which often goes for naught. It starts with an idea and a draft, but revision is where it comes together. Adding and subtracting. Cutting and pasting. Setting it down and coming back to it. Considering feedback and deciding what to do with it. All this is done in service of improving the whole, to deliver a message that lands properly and has a meaningful impact.

AI doesn’t do any of this because what it does isn’t writing. It can’t write because it doesn’t think. It can’t think because that’s not what it’s designed to do. It just receives instructions and generates outputs. These systems hoover up an unfathomable amount of data and then use massive processing power to approximate, triangulate, and regurgitate what they’ve ingested based on complicated algorithms. I like to think of them as elaborate digital blenders making information smoothies.

Some of the recent advances in AI have been unsettling. They suggest a path forward in which an increasing number of workers might have their jobs on the chopping block in an era of dwindling government support. Given these circumstances, humanists need to think about what may come, prepare to react, and advocate for beneficial change in things such as legislation that could help avoid undesired outcomes.

It’s not hard to imagine how recent developments in AI might lead to job loss. Why would a devout, profit-maximizing capitalist pay an artist when an intern can approximate work from grandmasters in the time it takes to pour a cup of coffee? The text-to-image functionality of the online graphic design platform Canva produced the associated graphic in about thirty seconds (along with three others) when I asked it for “a Van Gogh-style painting of dogs playing poker on a train.”

Forced Obsolescence
Jason Colavito, an author who writes at the nexus of science, pseudoscience, and religion, recently shared a story about a company he regularly writes for. They wanted to use OpenAI’s ChatGPT platform to create a draft document and then have Colavito edit the text for a fraction of his regular fee. Sci-fi author Ben Jeapes relayed a similar story as he surveyed our circumstances, writing, “It makes no difference that the product is flat, insincere, dry and devoid of any human feeling. It’s also cheap.” Firms may like the idea of using writers to inject life into ChatGPT’s outputs, but what happens to the writers?

And while driverless cars continue to make headlines for the wrong reasons, driverless trucks seem increasingly likely. There, the aim is to displace labor via a “transfer hub” model in which human drivers would deliver shipments from their origin to a hub located near an interstate. AI would then handle the long-haul portion of the trip (the part where drivers usually earn higher wages). At the far end, another human driver would handle the tricky bits. Back in 2021, the U.S. Bureau of Labor Statistics estimated that there were 2.1 million truck drivers in the United States and that the industry would add another 90,000 jobs over the next decade. Meanwhile, the people working to bring driverless trucks to market estimate that the transfer hub model will reduce the human-driven portion of long-haul deliveries by 90 percent.

Is this model starting to sound familiar?

Self-checkout machines are another job killer. As of 2018, cashiers were the third largest group of U.S. workers with 3.3 million people employed. As this work disappears, the effects are widespread. The obvious impact is in the form of lost jobs, most of which are held by women. But as an industry insider noted, the competitive nature of the retail grocery industry ensures that mass adoption of self-checkout machines would put “pressure on the whole industry to lower labor expenses, regardless of whether holdouts adopt the technology.”

We live within an economic system in which jobs are a necessity for most. For those without jobs or wealth, life is hard. Lives spiral downward as resources dwindle. Pandemonium ensues at the individual level.

It’s worth considering how these systems might impact our communities. What happens if hundreds of thousands of jobs, if not millions, disappear? All this is happening amid porous safety nets, while investment in higher education (was it meant to be an investment?) is increasingly questionable. How can we expect people to survive in “everyone for themselves” circumstances if there’s not enough work for everyone?

Economist Peter Morici views these matters very differently. Instead of viewing tech as a zero-sum game, he believes “it creates new higher-paying jobs to replace those it destroys.” While he propped up that argument with anecdotes, he did point to places with a higher ratio of robots deployed per worker than the United States—Germany, Japan, and South Korea—which also have a significantly lower portion of workers making less than two-thirds the national median wage. This might be good news if these four countries were identical. That said, the far more individualistic nature of the United States, and its frighteningly short-term orientation, suggest that these nations are not carbon copies. And while he says we should expect tech-related cost savings to initially be captured by “those who own the machines,” Morici wants us to believe that the rest of us will eventually benefit thanks to trickle-down theory. No further questions.

When asked about AI’s potential to displace workers, Ashraf Elnagar, professor of artificial intelligence at the University of Sharjah, claimed, “There’s a misconception that AI will replace humans, but AI can be looked at as an empowering tool for those people who are using it.” This is a parlor trick that’s popular in the AI community. They tell you the systems can’t do everything a person can do, so there’s nothing to worry about. AI will just assist us by handling mundane tasks while we focus on the important stuff. I’d love to ask those folks how they think AI was empowering Jason Colavito. I’d also ask them what exactly the driverless trucks were freeing up truckers to do with 90 percent of their time and what it would do to the wages for the remaining 10 percent of the work, with far more than 10 percent of the truckers scrambling for work.

In writing about tech’s advances, Cory Doctorow harkened back to the Luddites. You may know them as being anti-technology, but Doctorow would have you know that you’ve been misinformed. The historical Luddites were nineteenth-century artisans whose craft gave them a decent income. Their beef was not with technology’s existence but rather the division (or lack thereof) of its spoils. As inequality has been spiraling in recent years, the Luddites might give us a hint of where we’re headed. Doctorow’s thoughts on where this might lead are sobering. “There is no empirical right answer about who should benefit from automation, only social contestation, which includes all the things that desperate people whose access to food, shelter, and comfort are threatened might do, such as smashing looms and torching factories.”

Much like framing Luddites as anti-tech, there are ongoing efforts to frame AI in beneficial ways. When AI systems such as ChatGPT produce “convincing but completely fictitious answers,” its supporters call it a “hallucination.” This is an attempt to humanize these systems. Don’t fall for it.

The Oughtness of Things
The potential impacts of AI are so vast that Stanford University is hosting an ongoing study that aims to anticipate AI’s impacts on our lives that’s scheduled to go on for at least 100 years. The effort produced a report in 2021 that listed the four most pressing dangers related to AI: techno-solutionism, dangers of adopting a statistical perspective on justice, disinformation and threat to democracy, and discrimination and risk in the medical setting. While all those are obviously critical, it’s worth noting that the potential for disruption to jobs, sectors, or the economy didn’t make the cut. It’s also worth noting, as Elizabeth Weil did in an interview with linguist Emily M. Bender, that a handful of firms “employ or finance the work of a huge chunk of the academics who understand how to make LLMs” (large language models). Make of that what you will.

For Silicon Valley, it has long seemed that the unofficial motto is: Always, “Can we?” Never, “Should we?” Who needs things such as ethics and morals when you’re reinventing the world? But when a move-fast-and-break-things approach is coupled with a nihilistic orientation, the “things” that get broken are often people. The ongoing acceleration of technical “progress” creates the potential for unmitigated disaster. To paraphrase my former professor Timothy Weiskel’s familiar refrain, “We need to consider the oughtness of things.” That doesn’t seem to happen much in Silicon Valley.

It seems unlikely that technologists will shift their priorities in the near term, but their choices will continue to deeply affect us in an increasingly divided world. Sam Altman, OpenAI’s CEO, is reportedly hiring an “army of developers” to train his AI systems to replace entry-level coders. Consider what that might do to an industry that has been a bright spot for wages in recent years. Altman recently acknowledged that the work he is leading has the potential to “break capitalism.” He believes that doing so would necessitate changes to governance and company structure, but he isn’t working on those things.

Now, I’m not here to save capitalism, but if it’s going in the woodchipper, I’d like to find ways to work together to avoid collateral damage in the form of destroyed lives.

We need to do something about this. We are meant to do something about this.

Humanists are natural bridge builders. It’s our nature. Professor Michael Kelly notes this has long been humanism’s role in Europe:

Humanism has been a concomitant of almost all universalising projects in the last five hundred years of European history. In the present century, it has particularly served as a support for projects of political unity, brought to prominence as a general ideological bond at times when coalitions of interests have wished to sink their differences in a common political objective.

Humanists, our mission, should we choose to accept it, is to tirelessly work to bring people together. This is who we must be. Our circumstances necessitate action. Lacking (or refusing) the sort of tribal affiliation conferred by religion, we’re the ones who can build bonds and combat those who work to sever them.

Before you commit to this project, I’ll ask that you consider the possibility that my fears about our circumstances may be misplaced. If that’s the case, we might commit ourselves to needlessly building better communities. Perish the thought.

Chris Oestereich is a lecturer at Thammasat University's School of Global Studies and a Global Ambassador of the RSA. He is also the publisher of the Wicked Problems Collaborative and a co-founder of the Circular Design Lab. In all things, he is an advocate for sustainability and equality. Chris writes about the need for a paradigm shift towards interdependence in human relations and the environment, and believes that such a shift is necessary for a better future. Although he is pessimistic about humanity's current state, he remains optimistic about the future we can create. In the words of Marshall McLuhan, "There are no passengers on spaceship earth. We are all crew." Chris adds, "There's turbulence ahead. Grab a helmet."