A student, a screen, and a future suddenly in question.
My son is concerned about his future.
Several years ago, when I was teaching in the Technology Management program at BCIT, I was introduced to a YouTube video called Humans Need Not Apply.
Its subtitle did most of the work:
What happened to horses is happening to us.
And I recommended he watch it.
When he was in Grade 8, generative AI became commonplace. He took one of his school essays, fed the topic into ChatGPT and said, “write an essay on this like a Grade 8 student.” He compared the result to his essay and immediately was concerned. Having watched Humans Need Not Apply and then seeing what ChatGPT could do, the conclusion came quickly: he was the horse, and there would be no job.
His reaction both surprised and alarmed me.
The video was posted in 2014. It argued that automation was not only coming for repetitive physical labour. It was coming for drivers, clerks, analysts, professionals, and eventually creative workers too. The horse became the metaphor. Once machines replaced the economic role of horses, the horse population collapsed. The video’s conclusion was blunt: we needed to start planning for a future where vast numbers of people would not have jobs because human labour would no longer be required.
A future where Humans Need Not Apply.
Eleven years later, the prediction looks both wrong and more relevant than ever. The timeline was too simple. Mass unemployment did not arrive on schedule.
But the mechanism the video suggests is useful.
In my work with networking technologies, I have watched this happen more than once. Software-defined networking did not remove the need to understand networks. If anything, that understanding is needed more than ever. But the new technology changed where the knowledge had to live. Knowing commands was no longer enough. Students also needed to understand abstraction, automation, APIs, templates, and systems thinking.
The additional skills are not beyond the students I teach. But they do need to be taught, named, and practiced. Adaptation is not magic. It is curriculum, time, support, and a reason to believe the effort is worth it.
The same pattern appears elsewhere: dial systems reduced the need for telephone operators; ATMs changed the work of bank tellers; software changed the work of network administrators. The job does not always disappear. But the valuable part of the work moves.
Movement is not the same as disappearance. That is where the video moves too quickly. The horse analogy is powerful, but it is also misleading.
Horses were not workers in the human sense. They did not apply for jobs, negotiate wages, develop careers, retrain, retire, or ask what their work meant. They were domesticated living technology. People bred them, trained them, bought them, housed them, and used them for transport, hauling, agriculture, and war.
Then another technology arrived.
The car did not defeat the horse in an argument. It did not become more deserving. It simply became more useful for the systems people were building.
That is the part of the analogy worth keeping.
Replacement does not require moral superiority. It only requires functional adequacy.
That is where AI feels different. The speed is faster, the target is broader, and the institutions adopting it are not always clear about what they still expect people to know.
A program does not need to be brilliant. A report does not need to be insightful. Each only needs to be acceptable to the institution using them.
This is why the AI question is uncomfortable. AI does not need to be human to replace some human work. It does not need to understand in the way we understand. It does not need to care. It only needs to produce something an institution is willing to accept.
But this is also where the analogy breaks.
People are not horses.
A society that treats people as replaceable labour technology has already made the central error. Not a technical error. A moral one. AI did not create that error. It reveals it.
If a person’s value is defined only by output, then AI becomes a direct competitor.
If value includes judgment, responsibility, care, accountability, relationship, institutional memory, and moral agency, then the comparison changes.
Not because AI is harmless.
Because the question becomes institutional:
What are we willing to replace, and what are we willing to protect?
Top comments (0)