AI Taking Jobs

As I write this, I will be entering my final year of studies to finalize a degree in Computer Science. I wanted to write down my thoughts on Artificial Intelligence, as someone who has used it and been working in an industry disrupted by it.

Firstly, here's a list various questions and comments people have told me related to the rise of Artificial Intelligence:

• "Are you not worried that AI will take over the world?"
• "Why are you doing Computer Science? Do you not have concerns that AI will take over your job?"
• "Why would a company hire a programmer when they can have AI do it"
• "I bet the job market sucks because of AI right now"

I'll try to tackle these in a way that is comprehensive and appropriate.

"Are you not worried that AI will take over the world?"

No, not in the slightest. The fact of the matter is that the traditional AI that people associate with taking over the world are generally programs that can think for themselves and have some sort of conscious. The programs that are being developed now have no ability to think on their own (more on that later). They simply regurgitate data based on their trained dataset. There is something to say that maybe LLMs will take over how people do their work and change that, but enslaving humanity? That is not likely given what is out there right now.

"Why are you doing Computer Science? Do you not have concerns that AI will take over your job?"

I do CS because I enjoy solving problems on computers. An engineer does engineering because they enjoy solving the problems of whatever their discipline is. Part of this breakdown is that people view a CS degree as something different from an engineering degree. The reality is that modern Computer Scientists have two choices that they can go into. Either they are solving problems (typically by writing code), or they are exploring theory. In some sense, the theory is still a type of problem solving (for instance, is there a way to efficiently find the most effective route in the traveling salesman problem, without estimating).

I am not worried that AI will take over my job, because as a student, I have learned enough skills to keep myself valuable. AI will, and has already begun, to replace CS students who do not have a respect for the craft. It's not gatekeeping when I say that future programmers have a need and obligation to understand the frameworks and languages they are working in. An AI writing code does not consider if a react component will remount an infinite number of times. To the AI, it just generates text to meet your prompt. There is no logic in that decision. Perhaps the code you get works because many intelligent developers prior have published enough resources that got trained into this AI. In essence, I think that AI will be used to replace underperforming programmers, but to those of us who know how to use it, we can use it effectively.

AI in my Workflow

When I develop personal projects, I often use Visual Studio Code to write it. Pretty often I have Copilot sitting up next to me. Some people might ask why, but the reality is that Copilot functions much like my own Google. As I write a program, I may want to know more about the specifics of something I am working with. In today's modern search engine, Google will already use Gemini to give me the answer I am looking for. Why waste time switching windows to go there, when I have it embedded into my IDE? People might also argue that Copilot (or any other AI) is prone to giving me wrong answers. This is true, but I will discover that as I go. In a traditional sense, I might Google my question and end up on Stack Overflow. Stack Overflow is never 100% accurate either. Frankly, there are times when documentation for a tool is unclear or nonexistent as well. It's nice to have a tool to bounce ideas off of and at least cut down on the amount of time I might spend searching the internet for a solution, and get back to coding.

"Why would a company hire a programmer when they can have AI do it"

The hottest new word today is "vibe coding". Basically, anyone could give an AI a prompt and it will code something up for them. AI companies aim to make programming accessible to everyone through their products. In short, a company that opts to have AI do all its programming will have issues. The latest news story today relates to an app called Tea. This app had a pretty significant data breach that exposed selfies and licenses of many women using the site. It is unclear if the data that was taken was accessed as a result of portions of the app being vibe-coded. In any case, it is clear that the maker of the app relied on AI too heavily, and did not prioritize security and a proper "human in the loop" approach. There are countless other stories out there of developers who utilize AI to code their applications, and experience breaches as a result of leaked API keys or other problems.

Beyond this, AI struggles to architect systems. Perhaps the AI can create your dream app, but adding new features? That might require a significant rewrite and code restructuring to even be possible. This goes back to my original point, for a programmer to continue to be valuable, they need to double down on these skills. A programmer who can write code is replacable, but one who can architect systems? They are more valuable and are harder to replace.

"I bet the job market sucks because of AI right now"

The job market does suck, great observation! However, I do not think AI is to blame entirely. Certainly companies are scaling their workforces back since they can make their stronger members more productive. However, I think that the tech field is approaching another bubble with AI. Part of the bad situation for CS jobs right now is a direct result of the pandemic. Many companies saw an increased demand for digital services, and scaled accordingly. When the pandemic went away and life went back to a somewhat normal sense, companies saw that demand drop. That overhiring has led to a correction in the market, and now many programmers are losing their jobs or struggling to get a job in the field.

Going to school however, I do know countless students who have graduated and cannot find a job. In part, these students have an unrealistic expectation for their first job. They believe that they can land a $100k+ job that lets the work remotely and has many benefits. Reality sucks, and the fact is that many of these students have too high of an expectation for their first job. The other issue is that many of these students lack the drive to have their own personal projects or skill growth outside of the classroom. Their skills end there and they believe a company will teach them the rest. In my experience, a company expects you to come in with more than what a classroom teaches. The company might have patience to teach you company specific platforms and workflows, but the ability to learn that quickly comes from your own sheer will and determination.

In conclusion, there are still plenty of merits to being a Computer Science major. What needs to happen is that universities need to strategically adapt to the changing market to teach skills that matter to students. My wisdom to anyone wishing to get a degree in CS is to actually want to do it. CS should not be your fall back option, because you will struggle to find a job and continue to be outskilled.