By Casey Kennington
Now that some time has passed, we want to revisit the topic of how AI is affecting Computer Science (CS) and see if we have changed the view we had in July 2025. AI models have become even more capable than before and upgrades to old models have resulted in major advances. In short, everything still seems to point to AI being the future of software development. Therefore, if AI is writing code, what happens to computer science (CS) as a field of study or as a degree program for hopeful computer programmers? A recent article in TechCrunch, for example, looks into a resulting noticeable decline of CS enrollment at universities. An interesting note:
…parents who once pushed kids toward CS are now reflexively steering them toward other majors that seem more resistant to AI automation…
More AI, Less CS?
Indeed, there seems to be a widespread belief that most software will ultimately be written by AI. However, as a faculty member of a CS department at a large state university who teaches CS, data science and AI courses, and as someone who researches how to improve language models, I don’t see the causal connection between “AI will write more code” and “I should convince my smart child to go into a field other than CS.”
First of all, the amount of code that has been written and is currently being written is not zero-sum. That’s what has made software so powerful: there’s always something new that can be created. For example, at a fairly recent point in history, there was no such thing as word processors like Microsoft Word or Google Docs which are so commonly used now. Someone had to write, and someone needs to maintain the software for those programs. Still more recently, there wasn’t an Internet or smart phones; each in turn brought a new need for software from websites to mobile apps. AI is likewise a new platform for code, but also a new way of writing code. AI doesn’t mean less CS, it means more code written by everyone, from physicians who are making an app to help with their patients to an entrepreneur who is making a prototype that will turn into a product that changes the world. Importantly, many of those products are being made with code that needs to be meticulously tested and maintained by trained software engineers.
What about the CS Job Market?
If AI means more CS, then where are the jobs for CS grads? We have seen some pretty noticable layoffs in recent years. While blamed on AI, most (if not all) were really due to normal economic fluctuations, and not all (or even the majority) of those laid off software developers. Recent data from the U.S. Bureau of Labor Statistics revised the total jobs added in 2025 to be only 181,000, overall, not just CS. Moreover, AI hasn’t really made employees that much more productive. Despite massive adoption in AI tools at companies across the U.S., some adoption has stalled. Other evidence points to the layoffs being a necessary correction for companies that wanted to hire software developers, and instead of hiring properly educated developers, they hired anyone who spent a weekend or two learning to code on their own and marketed themselves as an experienced software engineer. The scrutiny of education and experience is higher now—a good thing for CS graduates who are better prepared for such roles.
As soon as the term “vibe coding” (i.e., just using AI models to generate runnable code using prompting–no coding experience required) entered the vernacular, it was quickly followed by “vibe coding paralysis” which is less about the AI tools themselves, but how humans harness them; not always productively in the long run. In a similar vein, one particular software engineer (the developer behind OpenAuth) posted a “scathing review” of AI in software development, stating:
…the bottleneck facing companies isn’t coding productivity — it’s a lack of good ideas, unmotivated employees, corporate bureaucracy, and “the dozen other realities of shipping something real.
In other words, vibe coding has its limitations, again highlighting the need for properly trained software engineers (i.e., CS grads). In fact, CS grads are projected to earn more in 2026 than ever before, likely due to the need for better scrutiny over the code that AI tools produce.
Changes in AI Perceptions Point to More CS
IBM is tripling their entry-level hires because AI alone isn’t a long-term strategy for software development. Yes, CS graduates should know how to use AI tools, but they also need to know the core CS principles that are taught in every accredited CS program. A recent high school graduate told Business Insider why he will study CS, despite hearing people say that CS will be replaced by AI. The interviewer’s summary is powerful:
Simply learning how to code no longer guarantees a stable or well-defined career path, but there are still many opportunities by staying within what I would call the eye of the typhoon and studying these machines, even if they are increasingly unpredictable.
He [the student] told me the biggest risk right now isn’t unemployment — it’s becoming irrelevant as tech accelerates.
In other words, according to this peer of current high school, university hopefuls, if you want to stay relevant in technology and not get behind, CS is still a good option. It may be the best option. A Google exec, who develops with AI, recently said that “this is the best era to be a [software] developer” because AI tools will help enable CS graduates to do more and focus on interesting problems.
Finally, despite the drop in CS job openings compared to huge increases of only a few years ago, a CS degree is still in higher demand than many other degrees. A recent survey showed that of the 10 most in-demand bachelor’s degrees, CS isn’t at 10 or even 9, it’s number 3.
In short: no, we haven’t really changed our minds about AI and CS. We agree that AI is in many ways the culture of CS, but that doesn’t mean we will need fewer CS graduates.
Boise State CS and AI Science: Preparing Graduates for the AI Future
CS programs are seeing more AI integrated into their curriculum to give CS graduates a solid understanding of what AI can do to increase productivity, but also AI’s limitations. We as a department have looked at what can be done in our own CS curriculum at Boise State to ensure that students know about tools that will make them better software developers and we have already made changes, some big and some small. Idaho has even introduced legislation to put some resources behind teachers and students learning about AI, an echo of what was done before when people started using personal computers and the Internet, showing more widespread recognition that learning how to use AI is a skill we cannot ignore.
As important as it is to learn how to use AI tools, I would argue that it is just as important for students to have the option to learn how to build AI tools. Since 2020, CS majors at Boise State University have been able to specialize in machine learning—the skills needed to build AI models. As the AI world has advanced, so have our degree offerings. Speaking of not becoming irrelevant as technology advances, we now offer a Bachelors of Science in AI Science aligned with industry needs, where students go beyond anything that has been offered to date on how to understand data, understand how machines learn, and to build the AI tools of the future.
You may believe (and you may very well be right) that the current AI market is a bubble that is about to burst, like the dotcom bubble of 25 years ago. However, popping the dotcom bubble did not result in less need for websites—the need dramatically increased. So will it be with AI, and our new degree program in AI Science is designed to help our students–your child–not just be ready for that future, but be someone who shapes that future.