How GenAI Became a Disruptive Force in CS Higher Education

For years, multiple signals have pointed to the need for change in computer science education: industry leaders have called for broader skill sets, tools have emerged to evaluate learning beyond syntax and correctness, and companies have increasingly hired developers without formal computer science degrees. Despite these signs, sustained demand for graduates and the comfort of tradition ensured that warnings were largely ignored, postponing meaningful reform. Now, GenAI has arrived, and the long-resisted change has become unavoidable.

This post outlines a timeline of these missed signals leading up to the arrival of GenAI, traces key phases in the evolution of computer science education, and offers recommendations for the future of higher education in the field.

A Timeline of Missed Signals

The following timeline of missed signals until the arrival of inevitable change presents key phases in the evolution of computer science education. While the dates are approximate and are meant to provide a general sense of progression, they highlight important shifts and trends over time, showing how the writing on the wall became clearer and clearer throughout a gradual process, until GenAI left no doubt about what was written.

1970s–2000s: A Stable Formula, While the Writing on the Wall was Faint

During this period, introductory computer science courses followed a consistent pattern, focusing primarily on coding, programming style, and the correctness of small programs. Changes were mostly limited to updates in programming languages and development environments. Assessments centered on code correctness, typically through handwritten exams and homework assignments involving small programming tasks.

While this formula seemed stable, the writing was already faintly visible on the wall, especially toward the turn of the millennium: the industry was gradually beginning to call for a broader skill set beyond mere code and program correctness, although computer science education had yet to respond.

2000–2010: Increasing Enrollments and Automation

This decade witnessed massive growth in computer science enrollments, both on campus and on online platforms like MOOCs. With increasing student numbers, the need to automate assessment processes became clear. Automated grading tools emerged that could check code correctness, style, and efficiency. While these tools sped up grading, they did not fundamentally change what was being assessed. Moreover, these tools were used primarily for low-stakes assessments rather than final exams, which continued to be handwritten.

The writing on the wall was growing clearer: automation was reshaping education, but deeper changes were still pending.

2010–2020: Industry Asks for More

During this decade, calls for competency-based computer science education grew louder. Employers increasingly emphasized the need for skills beyond coding, including agile development methodologies (e.g., teamwork, automated unit testing, and reflection), as well as human skills (such as collaboration, time management, and communication). To demonstrate these evolving requirements, the industry began hiring talented developers without traditional computer science degrees, focusing instead on skills, portfolios, and real-world problem-solving abilities.

Furthermore, computational thinking^9,10 gained recognition as an essential 21^st century skill, and discussions about it, as well as efforts to integrate it into computer science education, became a central theme in conferences and professional literature. Despite the growing industry demand for broader skills and the specific emphasis on computational thinking, pedagogy remained largely unchanged and students continued to be assessed primarily on individual coding tasks.

The writing on the wall was becoming impossible to ignore: computational thinking was emerging as a key 21^stcentury skill, and pioneering bypass efforts, such as Guzdial and Ericson’s (2016) approach⁶ that emphasizes creativity and open-ended programming for media creation like visual or audio collages, showed that rethinking pedagogy could both strengthen learning and broaden participation among underrepresented groups. These innovations pointed toward a more inclusive, forward-looking vision for computer science education. Still, the traditional model held firm, missing the opportunity to adapt before the emergence of an unprecedented technology less than a decade later.

Mid-2010s–2020: Degrees Start Losing Their Monopoly

During this period, major employers and companies across the tech sector began hiring developers without computer science degrees, placing greater emphasis on skill portfolios, real-world projects, and problem-solving abilities. The list of organizations adopting such practices quickly became too long to enumerate, making it clear that coding skills alone were no longer sufficient to secure employment.

The writing on the wall had grown bolder and unmistakable: traditional degrees were no longer the sole gateway to a career in computer science or in the broader tech sector.

2020–2022: Executable Exams, but No Transformation

The COVID-19 pandemic accelerated some changes in computer science education. The rapid shift to online learning made it clear that change was necessary, though the pandemic itself did not dictate the direction of that change. During this period, some institutions experimented with executable exams^5,2 within IDEs to better simulate real coding environments. However, these local innovations did not compel broader curriculum redesigns and failed to incorporate human skills into the assessment process.

The writing on the wall grew more urgent, yet systemic transformation remained elusive.

2022–Present: The Arrival of GenAI

GenAI has emerged as an inevitability technology¹—simple, versatile, autonomous, and impossible to ignore. Its proficiency in programming strikes at the heart of introductory computer science courses sooner than it does in most other disciplines. This rapid shift in capability has led to a contraction of entry-level coding roles, directly impacting higher education: students are questioning the value of a computer science degree, enrollment growth has stalled, and some institutions have even reported 20-25% declines in enrollment and course sign-ups.

The writing was on the wall, but GenAI’s arrival has made it impossible to ignore. Its impact on computer science is the textbook definition of disruptive technology^3,4: a technology that shifts value away from traditional offerings before institutions fully recognize the need to adapt, even if they do notice the preliminary writing on the wall.

Why is the Case of GenAI Different?

Past technologies gave educators the option to adjust; GenAI gives no such choice. Its strengths—code generation and problem solving—directly overlap with what computer science education has historically treated as the core outcomes of introductory courses. Furthermore, GenAI’s excellence in project prototyping, and specifically vibe coding, has overshadowed the appeal of traditional introductory computer science courses. As a result, the basic values of a computer science degree are under pressure.

Some computer science departments continue to claim they do not teach coding but focus on scientific thinking; yet, the content of most introductory courses and their assessment methods tell a different story. In most cases, the unique voice of these institutions goes unheard amid this turbulent period.

What Must the Computer Science Higher Education Community Do?

The computer science higher education community understands the challenge it faces and recognizes what must be done: embrace GenAI and reinvent teaching methods and learning materials, or risk becoming obsolete.⁷ Possible recommendations include:

• Integrate GenAI as a core tool, treating it as an essential tool of learning and teaching.
• Shift focus from syntax to skills and abilities, emphasizing problem solving, ethics, collaboration, and critical thinking.
• Explicitly value human skills by designing curricula that teach and evaluate teamwork, communication, and adaptability.
• Prepare for diverse career paths and reconsider admission requirements,⁸ recognizing that many students will work in roles that combine computer science, AI, and other disciplines, and will likely change careers multiple times throughout their working life.

Conclusion

The writing has been on the wall for years: industry demands, alternative hiring, new tools.

The community of computer science education ignored it while enrollments remained strong. GenAI has made the change inevitable, hitting and impacting computer science education earlier and more deeply than it has most fields, due to its excellent coding capabilities.

Now, as is the case with disruptive technologies, the question is not whether to change, but how quickly the introductory courses and, eventually, the entire computer science curriculum can be redesigned. Otherwise, computer science higher education risks becoming obsolete.⁷ One noticeable path is to prepare graduates for a future in which AI is their collaborator, not their competitor.

Disclaimer: This post was written with the support of ChatGPT, which helped organize ideas, refine language and terminology, and structure the narrative. Final content and viewpoints remain solely those of the author.

References

1. Armony, Y. and Hazzan, O. (2024). Inevitability of AI in Education: Futurism Perspectives for Education for the Next Two Decades, Springer.

2. Bourke, C., Erez, Y. and Hazzan, O. (2023). Executable Exams: Taxonomy, Implementation and Prospects, SIGCSE Technical Symposium 2023, Toronto, Canada, pp. 381–387, https://doi.org/10.1145/3545945.3569724.

3. Bower, J.L., and Christensen, C.M. (1995). Disruptive technologies: Catching the wave. Harvard Business Review, January-February issue, pp. 43-53.

4. Christensen, C.M. (1997). The innovator’s dilemma: When new technologies cause great firms to fail. Boston, Massachusetts, USA: Harvard Business School Press. ISBN 978-0-87584-585-2.

5. Erez, Y., and Hazzan, O. (July 7, 2021). 10 Tips for Implementing Executable Exams, BLOG@CACM.

6. Guzdial, M.J. and Ericson, B. Introduction to Computing and Programming in Python, 4th edition, Pearson (January 9, 2015). © 2016

7. Hazzan, O. and Erez, Y. (September 18, 2023). ChatGPT (and Other Generative AI Applications) as a Disruptive Technology for Computer Science Education: Obsolescence or Reinvention, BLOG@CACM.

8. Hazzan, O. and Salmon, A. (June 10, 2025). Should Universities Raise or Lower Admission Requirements for CS Programs in the Age of GenAI?, BLOG@CACM.

9. Wing, J.M. (2006). Computational Thinking. Communications of the ACM 49 (3), 33–35. https://doi.org/10.1145/1118178.1118215

10. Wing, J.M. (2010). Computational Thinking: What and Why? Unpublished Manuscript, Computer Science Department, Carnegie Mellon University. http://www.cs.cmu.edu/~CompThink/resources/TheLinkWing.pdf

Orit Hazzan is a professor at the Technion’s Department of Education in Science and Technology. Her research focuses on computer science, software engineering, and data science education. For additional details, see https://orithazzan.net.technion.ac.il/.