The Artificial Gets Real

The overlap of cliches, innovation, and irony

Ronald Reagan once said that a recession is when your neighbor loses his job, and a depression is when you lose yours. While I'm not depressed by my now-former employer's "restructuring," it's not particularly thrilling, either. The upside is that this movie has played before and the protagonist comes out of it okay. Some of the scenes within the film are unpleasant, but meaningful change is impossible without discomfort.

It's hard to be 100% certain as to why my position was impacted, but there is a strong suspicion that AI is involved, which leads to the heart of this article. While it has been comical to watch the Google people wriggle beneath the weight of their own stupidity in training an AI image generator to think white people never existed, the reality behind this tool is a bit more sobering.

People with four-year degrees are twice as likely to be impacted by AI compared to individuals with a high school education. This is going to produce some ripples that tend to be overlooked. Roughly 20% of the workforce has what are termed "high exposure jobs;" in other words, these positions are relatively well-paying, and IT is among the hardest-hit industries.

The snarky admonition of "learn to code" is not quite as cute as some thought, especially when the non-coders at whom it was directed now find themselves in a better economic position. So what happens when the affluent and their disposable income are suddenly out of circulation?

How they spend and invest will be directly affected. The investment aspect may stop altogether in order to funnel funds toward the basics, such as housing, food, and energy. The loss of income will have an economic ripple, impacting the businesses that rely on those dollars and cascading to govts that rely on the tax revenue from productive activity. This may not matter to a federal govt with a printing press, but states and municipalities do not have that luxury. They typically have statutory requirements to balance their budgets. When projected revenue streams unexpectedly dry up, that's a problem.

There is an element of irony in all this. A significant part of the company’s future growth was predicated on harnessing the value of AI for operational efficiency, improved data management, and other things that this technology does well. I know that because I wrote a fair amount of marketing collateral, blog posts, and other material extolling the virtues of this innovation, particularly how it would blend with humans in a synergistic case of man meets machine. Perhaps I paid less attention than I should have to the value of ChatGTP and other authoring tools that spit back documents of varying lengths with a simple prompt. The potential benefits make it easy to overlook the red flags.  

Further digging revealed other curious things, namely AI's insatiable need for energy. The investment community's darling of the moment, NVIDIA, is projected to ship out about 1.5 million server units annually for the next few years. If those machines run at full capacity, they would suck down more power than some countries use. That's not an exaggeration. One projection says that by 2026, electricity consumption in data centers will equal that of Japan. And it doesn't stop there.

In the Portland suburb of The Dalles, Google operates three massive data centers that collectively use more than one-fourth of the city's water supply. That hardly sounds sustainable, particularly if the community grows. Along with the effect on people, there is the corresponding impact on fish and wildlife that depend on the same water sources that people use, not to mention plants and trees. Google has put money into upgrading the municipal system, but neither the company nor anyone has yet figured out how to create more water.   

Beyond the immediate concerns, this massive drain on resources is entirely at odds with what we're told about the need to be more mindful of the earth, greener, and move from fossil fuels to alternative energy sources. If the current grid struggles to keep up with those horribles known as coal, oil, and gas, how will it manage in this fantasized version of the future? Where is the potable water going to come from? Perhaps most startling about the Google case is how the city fought to keep that information secret, claiming that the data center's thirst amounted to a trade secret. It took a lawsuit to learn how much of a drain, literally, technology can be.

Please don't mistake any of this as the musings of a Luddite. Like every other digital immigrant of my generation, I have adapted to the new tools introduced over the past few decades by necessity. There was no debate about figuring them out; it was just something we did. These tools are marvelous in managing massive amounts of data, maintaining records, and providing new communication channels. But like anything else, innovation comes with tradeoffs, including human costs and environmental concerns. While we are hectored about controlling our individual footprints, more resource-intensive facilities come online.

The study from Pew Research that was cited at the beginning deliberately emphasized "exposure" to AI instead of predicting how many positions will be lost. The term is relevant because the study also accounts for the importance of job functions not associated with AI. While work classified as high skill and requiring an advanced degree will be at the highest risk, the odds of displacement drop the more the human touch is needed. None of which did me and some others any good, but it opened the door for a look behind the AI curtain to some things that are seldom discussed.