“I do not understand your concern,” Horse 1 says, presumably via telepathy or whichever language horses speak. “You should be happy about cars.”
“I don’t know,” Horse 2 says. “If humans have cars, what would they need us for?”
“Quit horsing around,” Horse 1 replies. “Every other human innovation has only made our jobs easier and better. Guns made sure that humans could fight others at a distance without risking us getting caught in close combat. Telegraphs reduced the need for us to run back and forth across the country. The automobile just means even fewer redundant trips for us.”
“I suppose you’re right,” Horse 2 concedes, unaware that the automobile would, indeed, eventually eliminate the economic necessity of animal-based transport in the developed world.
The rapid obsolescence of the domesticated horse is often used as an analogy for artificial intelligence and its relationship to broader human society. Of course, it is flawed (humans are, of course, not horses). However, it demonstrates a fundamental principle that I believe is pertinent to understanding within the current discourse around AI: there is a limit to how much technology can truly benefit mankind.
AI is often compared to other significant technological inventions of recent history, whether it’s the mechanization of manual labor or the spread of the internet. These inventions have, ultimately, made our lives easier, and we have eventually adapted to them. But if we continue to make labor easier for humans to perform, will we, too, experience our “horses-and-automobiles” event? Will AI make humans obsolete?
My motivation for writing this article primarily comes from the advancement of the agent architecture in many cutting-edge AI models. The conventional understanding of AI among the general public still largely remains limited to generating essays and creating uncanny-valley images. However, AI agents are capable of much more than that. Agents use large-language models (LLMs) like ChatGPT or Anthropic’s Claude to autonomously perform tasks with minimal to no human intervention.
Imagine having an AI model grab data from a spreadsheet on your computer’s file system, then summarize all the new messages in your work group chat and finally create a presentation from the materials gathered. With the current state of AI technology, you could do all of that on your phone while on the train. At this point, almost anything that is performed on a computer by a human can be automated away.
There are very clear economic ramifications for these AI advancements. Entry-level jobs within the U.S have fallen by 35% recently, largely due to AI. It makes sense from a hiring standpoint—why pay a full salary to an employee for the lower-level grunt work if AI could do it for much cheaper? At large tech companies like Meta, this is already a reality. Thousands of people have lost their jobs as AI has taken over more junior-level work, such as generating code. Recent reports from Anthropic’s own labor market research show that AI implementation within the corporate world may still be in its infancy.
At the same time, a total AI takeover of the labor market could be grossly exaggerating reality. Some of these companies may be engaging in “AI-washing.” In essence, artificial intelligence may be granting large companies a “get out of jail free card.” Widespread automation will take considerable amounts of time and money, but in the meantime, it is a pretty convenient scapegoat for broader company issues such as increased costs due to AI deployment or financial concerns that may have nothing to do with AI.
As with everything, there is considerable nuance within this subject. Perhaps it is futile to even discuss this topic as an editorial given the impossibility of predicting the future, especially when both overenthusiastic tech bros and anti-AI Luddites present valid arguments. But as the Class of 2026 approaches their high school graduation, it remains beyond frustrating to me that the current prospects for the future are so volatile.
The conventional advice for children is to stay in school, get an education and work in a high-paying white-collar job. I vividly remember playing around with block coding in elementary school during the “Learn to Code” movement of the 2010s. There have also been various government initiatives to get children interested in learning about STEM. Much of our generation has essentially been socialized into pursuing technical work which requires a deep education. Ironically, at this point in time, the work that we always thought would be safe—the work that requires formal education and pays top-dollar—seems the most unstable.
There are so many plausible paths for AI to progress at this point that it seems practically impossible to know whether a career or educational decision will be beneficial or not. While cushy corporate jobs have long been the goal for high-achieving students, the most stable options at this point tend to be in the trades, including construction and plumbing. But telling kids to just “go into the trades” is about as fruitless as telling them to just “pursue STEM.” As with any field, the trades will only appeal to specific groups of people.
As a student who is going to college to study computer science, and as someone who has genuinely enjoyed programming for years, I am often faced with these difficult questions. Should I jump ship and pursue something that I will inevitably enjoy far less? Should I continue with my current path and essentially gamble, hoping that the work I pursue will evolve with AI rather than be replaced by it? Should I even base my decisions on a technology which may not pan out the way I expect it to?
While I aim to be an optimistic person, it is quite difficult to see a universally favorable outcome here. Should the tech bros be correct in their claims of continued exponential growth in AI technology, technology could pose an imminent threat to thousands of jobs. Alternatively, if such jobs end up adapting to AI, the Jevons paradox may strike again. The paradox states that, as technology makes the use of a resource efficient, overall consumption of that resource will grow, rather than fall. This has historically been seen with coal; as steam engines continued to increase in efficiency, the demand for coal increased given how much cheaper it became to travel longer distances with less fuel. In the context of the labor market, this means that particularly AI-exposed fields such as radiology and programming could potentially see increased demand as AI makes them more efficient.
Essentially, doing work more efficiently just means that companies can get more work done, and so they will assign more work. This scenario would essentially force AI to become a ubiquitous part of many workflows in jobs to enable rapid output. However, the constant delegation of work to AI agents, as well as tightening deadlines with companies expecting more output per employee, may just create more workload for employees in the future.
So, the best outcome would seem to be for this AI thing to just die out, right? That may not be true either, particularly given just how influential AI is on the U.S. economy. The technology is driving so much investment that its collapse could be catastrophic for the economy.
It is an extremely uncertain time for those entering employment currently and within the next couple of years. Beyond AI, colleges continue to explode in price while wages have continued to stagnate and it remains extremely difficult to buy a house. It is a perfect storm that I anticipate will impact young and upcoming professionals gravely, no matter what path AI progresses in.
I did not write this article intending to fear-monger or encourage individuals to doomsday-prep (as mentioned, that would be impossible to do given the unpredictability of the current economic climate). Think of this as a venting session. The individual is powerless against a system; there is likely not much that we ourselves can do. However, as the Class of 2026 approaches graduation and, by extension, voting age, I implore any and all readers to stay aware of the current situation and evaluate how the political trailblazers of tomorrow plan to address it. It is undeniable that good leadership through the fog of this greatly uncertain time is a non-negotiable. Unlike horses, we have the power of democracy on our side.
