Modern CS: The Age of Minors and Certificates

I started a draft of this essay last month, but forgot about it until now.

It seems that every other week or so, somebody from one of the frontier AI labs would drop an apocalyptic prediction on social media and interviews regarding how AI is (quickly becoming) capable of replacing human in different types of jobs. While these kinds of statements should be taken with a large spoon of salt, a study from March 2026 from Anthropic found that while there are jobs that are significantly more exposed to AI integration, there is no impact on unemployment rates for workers in the most exposed occupations, although “there’s tentative evidence that hiring into those professions has slowed slightly for workers aged 22-25”.

Rather than focusing exclusively on whether jobs are disappearing, in this essay, I want to echo the sentiment of a recently blog entry by Andrew Ng that it may be more productive to examine how the structure of computing work itself is changing. If more people outside traditional CS roles can now perform routine computational tasks, then the educational question becomes less about replacing software engineers and more about how universities should structure computing education for everyone else. How should traditional CS programs think about alternative pathways such as minors and certificates when computing is no longer confined to students who intend to become computer scientists?

The Overloaded CS Major

The traditional CS major was built around a certain image of the student and a certain image of the profession. The student would learn programming, then discrete mathematics, then data structures, algorithms, computer organization, operating systems, software engineering, theory, and a set of upper-level electives. This structure still makes sense for students who want to become computer scientists, software engineers, systems engineers, security specialists, or researchers.

However, the same curriculum has increasingly been asked to hold many different bodies. A student interested in computational biology may not need the same path as a student interested in compiler design. A business student who wants to build forecasting workflows may not need the same path as a student who wants to design distributed systems. Yet, institutionally, we often treat all of these demands as if they should flow through the same doorway. If someone wants serious computing knowledge, the default answer is often: take the CS major, or at least take the same early CS sequence that was designed for majors. This is not unreasonable, but it becomes strained when computing becomes a general capability that many professions now need.

The rise of AI has reduced the friction for many routine computing tasks such as writing small scripts, modifying existing code, and creating prototypes. These are not trivial tasks, but they are also not the same as maintaining a cloud platform or building a reliable operating system. In other words, AI has made it harder to demand that every useful computing task requires the full identity of a computer scientist. Some tasks still requires deep expertise, but there will be those that require practical computational fluency or just enough knowledge to use AI-generated output responsibly. While these tasks are related, they do not share the same educational goal.

The Rise of Domain-Computing Professionals

With the rise of data science and big data in the early 2010s, we have seen the arrival of the computational X movement. A computational X-ist is someone who is not quite a software engineer, but also not merely an end user. This person may be a biologist who can automate lab analysis or a financial analyst who can build Python workflows to interact with some remote AI backends. These people are expanding the surface area of computing as they understand a domain well enough to know what problems matter and computing well enough to improve how the work gets done. In many organizations, especially outside of large technology companies, this hybrid role may be more valuable than a narrowly trained programmer who does not understand the surrounding context.

This is especially true for regional companies, public institutions, nonprofits, and smaller organizations that need people who can move across boundaries. More specifically, they need someone who can talk to domain experts, understand messy workflows, evaluate technology, and build practical solutions that are good enough to improve daily operations. Increasingly, that someone may not be a traditional CS major but a domain professional with enough computing knowledge to know what is possible, what is risky, and what is worth doing.

For CS education, this creates both an opportunity and a warning. The opportunity is that computing can serve more students than ever before. The warning is that we should not force every one of those students into the same professional mold. Not everyone who needs computing needs to become a software engineer and not everyone who uses AI-assisted programming needs to become a computer scientist. But many more people will need enough computational understanding to avoid becoming passive consumers of tools they do not understand.

The Age of Minors and Certificates

This is why I think minors and certificates deserve more serious attention. Typically, CS minor includes the core sequence (programming, data structure, algorithm) and one or two electives. Given that data structure and algorithm are the gate-keeping courses of the CS major, this raises the question that what is the point of a CS minor? Without the advanced technical electives that truly define a computer scientist, struggling past the difficult lower-level courses does not make much sense.

A good computing minor or certificate should answer a different question from the CS major. If the CS major asks what a student needs in order to become a computing professional, then a minor or certificate can ask what a student in another discipline needs in order to apply computing responsibly and productively within that discipline? With these different questions, we should arrive at different curricular designs. For example, at fictitious university W, a CS minor requires 6 courses: CS1 (Python), CS2 (Java), CS3 (OOP), Discrete Math, Calculus I, and Data Structure and Algorithms. A more applicable alternative (built on existing courses) would be CS1 (Python), Calculus I (prereq), Linear Algebra for Applied Statistics, Introduction to Data Science, Introduction to Information Visualization, Introduction to Machine Learning. This alternative would be more attractive, and more importantly, accessible, to non-CS students.

Certificates are slightly harder to manage, as they are typically intended for major only. First, for CS major themslves, creating more certificates makes sense. Taking one step back, a certtficate indicates that a student has pursued a dedicated technical path. Isn’t that already the purpose of technical electives? In some sense, a certificate lets a student say that I am not simply taking random courses, I have a solid intention of doing A or B or C, hence I am getting this certificate. In that case, let’s make it easier for students by explicitly defining certificates. It will also help prospective employers and hiring managers who are not necessarily CS to better understand an applicant’s academic background. This set of courses is for Data Science Certificate. This set of courses will net you the Data Engineering Certificate. If you want to get Cloud Engineering Certificate, you must take A, B, C, and D. I am already eligible for graduation, but I can get a second certificate if I take one more extra class!

Second, and more importantly for this essay, certificates can be designed non-CS students through cross-departmental collaboration. For example, CS1 is often also classified as a science general education class for non-CS students. Business, Math, Biology, etc most likely will have introductory data analytic courses that have similar contents as a second-year Introduction to Data Science course. Two more courses, perhaps one taken as a free elective, and the Data Science Certificate is within reach. Similar discussion/negotiation can be done with Math and Engineering departments for AI Engineering or ML Engineering Certificate.

At regional public universities, this may be especially important. Many students are not trying to become frontier AI researchers or engineers at large technology companies. They are preparing to work in hospitals, schools, government offices, small businesses, banks, manufacturing firms, and regional organizations. For these students, a computing minor or certificate may be the difference between being someone who merely uses technology and someone who can shape how technology is used.

What About Rigor

The example in the section above will immediately raise the alarm bell for many CS faculty. Without data structure and algorithm, how can one understand proper techniques in data analysis or machine learning? This concern is quite valid, and it represents another hard choice that CS departments has to make: which subfields of the discipline can we tighten the traditional rigor and which subfields can we relax?

At institutions everywhere, it is almost certain that CS departments want to have a say in all things computing. The problem is that we cannot have our cake and eat it too. If CS wants control over all computing-related courses, then CS has to be able to support all students interested in computing-related courses. If we do this, then applying the same level of rigor for all is simply not sustainable. This does not mean we remove rigor. Instead, we need to define rigor differently for different educational goals. For a CS major, rigor may involve algorithmic analysis, systems design, formal reasoning, and deep implementation. For minor or domain-computing certificates, rigor may involve correct tool use, data interpretation, reproducibility, ethical judgment, workflow design, and the ability to recognize when a problem requires deeper technical expertise.

CS as the New Mathematics

The analogy I keep returning to is mathematics. Everyone needs some mathematics, but not everyone needs to become a mathematician. Basic arithmetic, algebra, statistics, and quantitative reasoning are part of general education because modern society requires them. At the same time, advanced mathematics remains deep, difficult, and specialized. The fact that everyone learns some math does not make mathematicians unnecessary.

Computing may be moving in a similar direction. Everyone may need some computing, but not everyone needs to become a computer scientist. A student should be able to understand data, automation, AI tools, basic programming logic, privacy, security, and computational limits without necessarily completing the full CS major. At the same time, we still need students who go much deeper into systems, algorithms, theory, security, machine learning, embedded computing, and software engineering. We need to reach the same level of maturity as mathematics and stop treating every computing pathway as either “real CS” or “not CS.”

AI makes this more urgent because it increases both access and risk. More people can now produce code, analyze data, automate workflows, and build prototypes, which is a good thing. However, more people can also produce broken code, misread data, automate bad assumptions, leak sensitive information, and trust systems they do not understand. Computing should be taught more broadly and more carefully instead of being locked inside the CS major. We can help design the broader computational education that the university now needs.

Conclusion

The current AI discussion often begins with the wrong, or rather narrow, question, which is about whether AI will replace software engineers, programmers, or other computing professionals. A better question is about how AI is changing the distribution of computing work across society. From that perspective, it is possible to see that AI is making routine computing work more accessible to people outside traditional computing roles. As a result, the difference between deep CS expertise and general computational fluency becomes more noticeable, and therefore, the learning paths for these two areas are more separable.

This is why minors and certificates should be taken seriously as they may become one of the main ways universities respond to a world where computing is needed everywhere, but not everyone needs the same computing education. If the rise of computational X in the last decade was about convincing everyone that computing matters, the next decade may be about building the right educational structures for that reality.