I've been thinking a lot about the future of Computer Science a lot lately. Specifically the question on my mind: Is it ethical to encourage students to consider Computer Science for their future studies and careers?
As a Secondary school teacher, I have the privilege of helping guide students through the transformative teenage years of values-formation, self-discovery and ambition setting. I enjoy witnessing students discover Computer Science as an exciting field filled with creative potential to innovate and solve problems for people in the real world. I love the shrieks of euphoria from students when, after wrestling with a problem or bug "for hours", it finally "works"! I get joy seeing students get excited about their projects, and dream about how they could use Computer Science to contribute back to society.
As their teacher, I want my students to live happy, fruitful lives according to their values. The last thing I wish is for any former student to find themselves stuck on a path of perpetual unemployment through redundant expertise.
For a long time I've argued that LLMs are just a fancy auto-complete on steroids. That argument really doesn't face scrutiny any more. Agentic programming is a very powerful tool and an essential skill for anyone wishing to maintain relevance in today's job market.
Honestly, it is rare for me to open VS Code for serious programming work these days. While I am looking forward to the coming Advent of Code in a few weeks as an excuse to put the LLM away and experience the joy of problem solving based programming again, the truth is that for productive projects it is agentic programming tools that I am turning to 99% of the time to get the job done.
Software development itself is certainly not about to go away. If anything it is the opposite. The Jevons paradox is relevant here. It states that "technological improvements that increase the efficiency of a resource's use lead to a rise, rather than a fall, in total consumption of that resource". Rather than companies producing the same output with 1/10th the people, I think it is far more likely we will see 10x the software created instead. As the barrier to development lowers, we are going to see more rather than less produced.
This is certainly true for my own personal side projects. I have produced dozens of new tools or added features to existing apps that I never would have had the time to do if I had to have done it all manually.
What does this mean for students entering the field? Are their skills useless in a world where there is a reduced barrier to entry? Let me pose the question this way: Would you trust an app with your personal information if it was vibe-coded by an amateur who just watched a couple of youtube videos? Therein lies the answer. This is the reason why there is a critical place for formal Computer Science skills.
There is another interesting question linked to this: How much do we, as society, want to subcontract the task of software engineering to the 6 or 7 mega corporations who currently occupy the frontier model space? We have numerous examples from very recent memory of massive supply chain disruptions from pandemics, tariffs and trade wars, to actual shooting wars. Do we want our ability to write and maintain software to have fragile dependencies that are at the whim of the commercial interests of a multi national corporation and the geopolitical environment in which they operate?
What happens when a prompt is denied due to "company XYZ already creates a tool with this functionality, I am prohibited from replicating its functionality"? "your country is denied access to our model"? or even just your monthly plan suddenly increasing by a factor of 10x once all the competition has been eliminated?
If the counter argument is to say "oh, I could always do it myself, I just use the AI to speed run it", then you have just made the case for human based Computer Science expertise. Is it wise for our experienced developers to let their skills atrophy, or for there to be an incoming generation of junior developers who have never known an alternative? Locally installable LLMs do exist, yes, but these will always significantly lag behind the capabilities of the frontier models. Author Frank Herbert's quote in Dune seems apt: "Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them." Once again, it seems the best insurance is to invest in strong personal expertise.
The other reality is that anyone who thinks avoiding Computer Science as a field as being the best way to avoid LLMs wrecking havoc on their profession, has a rude awakening coming. Software development is only the canary in the coal mine. All knowledge based professions have a LLM-shaped tsunami coming their way in the not-too-distant future. The entry level positions may be the first to face this (paralegals, accounts receivable/payable, graphic design, writers of all descriptions, translation etc) but it is coming for them all whether they see it or not. Jobs requiring a physical component have an additional window of time before the revolution comes their way but AI-powered-robotics is on the way.
Perhaps, then, it is the Computer Scientist with their detailed understanding of the workings and capabilities of the LLM who are best placed to capitalise on opportunities this transformation presents? And, so, this article has completed a 360 loop. We are back where we started, and I'm no closer to answering my question. I do worry for my students and the uncertain job market they will face, but I remain hopeful in their potential to master their futures rather than being a slave to it.
Ultimately while Computer Science may be the subject I teach, I hope my students walk away with far more than that. If facts and knowledge are going to be the realm of the AI, then what is the value the human brings? Curiosity. Empathy. Communication. The human touch. There is a quote attributed to Albert Einstein that I have used for years in the opening lesson of my courses: "I am neither especially clever nor especially gifted. I am only very, very curious." Perhaps he knew what he was talking about after all.
Paul Baumgarten October 2026