Don’t believe everything the doomsayers say

Stephen Addison

AFTER DECADES OF growth in both jobs and student numbers in Computer Science programs, the emergence of generative A.I. into the mainstream is coinciding with apparent stalls in admissions into college Computer Science programs and stagnation in the growth of Computer Science jobs—particularly at the entry level. As a result, pessimists are now out in force with tales of doom and gloom. For example, on June 21 The Atlantic published an article by Rose Horowitch entitled The Computer-Science Bubble is Bursting. “Artificial Intelligence is ideally suited to replacing the very type of person who built it,” wrote Horowitch, and then she went on to note that the number of Computer Science majors is falling at Duke, Princeton, and Stanford, among other places. Similar stories, including a narrative about a softening in the Computer Science job market, have appeared in the New York Times and other media outlets.

While enrollments and graduation numbers are down in many places, and the job market is indeed softer than it has been, there isn’t a clear causal link between these numbers and the growing use of generative A.I.—and it’s too soon to declare that there’s a trend. Other causes are affecting both the job market and college enrollments, making the contribution of A.I. to this situation unclear.

First consider the job market: The U.S. software industry is increasingly mature, and software tools are widely available and stable and have been adopted across most businesses. Entry-level coding and programming roles are now frequently outsourced globally, and the widespread adoption of remote working technologies has reduced the need for fulltime, in-house employees. In general, companies are not looking for or building new systems—the focus is on making efficient use of existing systems and software.

This trend is also reducing the number of entry-level coding and development positions. In addition, we have experienced high inflation, increasing interest rates, and global instability. High inflation and higher interest rates lead to cost-cutting and decreasing investment, which translates to layoffs, hiring freezes, and less hiring; that, in turn, increases competition for the jobs that are available. This all leads to uncertainty, and this uncertainty is increased by disruptions caused by US-China trade tensions, including tension related to Taiwan, as well as disruptions caused by the widespread involvement of many countries in the war in Ukraine. We should also remember that cyclical hiring and layoff periods have long been common in the computer industry, even in times of prosperity.

***

NOW LET’S CONSIDER students enrolling and graduating. The Computer Research Association tracks both enrollments and graduations in Computer Science; if we examine their data from past events, we can see that Computer Science enrollments tend to lag behind trends observed in the job market. After the bursting of the dot-com bubble in 2001, for example, the number of Computer Science majors declined, with the largest effects being seen from 2005 to 2007. In contrast, enrollments grew starting with the recession of 2008, as IT companies weren’t struggling when many other industries were.

Today, new graduates often find it more difficult to find positions because of the availability of experienced candidates who have been laid off with many prospective employers requiring experience that many new graduates do not have. Today’s new graduates are also competing with potential employees who don’t have degrees but who have developed the needed skills through bootcamps or by earning certifications. These candidates are qualified for many entry-level jobs. But the number of degrees awarded in Computer Science doubled between 2013 and 2023, and those who specialized in in-demand areas like cybersecurity, cloud computing, software engineering, and A.I. more easily secured their first positions in industry.

Enrollment and graduation numbers are still suffering from the aftereffects of the pandemic. Today, and for some years to come, the deficits in mathematics preparation in schools over the pandemic years will affect readiness for college-level mathematics courses across the STEM disciplines. This means the effects of the pandemic are now appearing directly in graduation numbers, and indirectly through inadequate preparation in mathematics, which is reducing both initial enrollments and persistence in Computer Science and other STEM disciplines. We should also remember that Computer Science has long been a popular choice of major for international students, and international enrollments in U.S. universities are declining, which will result in a lower number of graduates in the future.

So, while there are downturns in both the number of graduates and the number of entry-level jobs, there are many causes for that—and while A.I. is one of them, it is far from being the only one.

***

RETURNING TO THE question in the title of this piece, in terms of employment opportunities and enrollments in Computer Science degree programs, is the sky falling? My answer is a resounding no! While there’s a current downturn due to factors outlined above (as well as skill mismatches between some recent graduates and a growing regulatory culture around data privacy, cybersecurity, and A.I.), it’s likely that we’re seeing a temporary pause rather than a long-term decline.

Like the computer itself, Generative A.I. is an example of a general-purpose technology. Beginning with the industrial revolution and continuing through today’s information revolutions, general-purpose technologies have been drivers of economic growth. Along with this growth, these technologies have also transformed the workplace—old jobs disappear, and new jobs are created in greater numbers. Work activities will change as we adopt Generative A.I. in the workplace. There will be little space for web designers or low-level coders, but the demand for people who can perform code-reviews and deploy A.I. tools will increase.

The changes will be far-reaching and will extend beyond IT occupations. Generative A.I. has the potential to replace translators, sale representatives, accountants, editors, financial advisors, and a whole host of other professions. The adoption of new technologies tends to eliminate more physically oriented jobs and replace them with knowledge-based jobs, but while we have seen great advances in tools like ChatGPT over the last three years, changes in the job market may take longer than we expect.

Let’s look, for example, at the tasks of an editor. Many parts of an editor’s job have already been replaced—grammar checkers and spell check have largely eliminated copy editors, and LLMs are reducing the need for many editing functions. But have books and articles improved? It is common to find modern books filled with repetitions, anachronisms, the use of incorrect words, and continuity errors—all of which would have been prevented by a good human editor. So A.I. isn’t the answer to all our problems, and not all uses of A.I. lead to improvement. Books edited by a professional editor offer readers a far more rewarding experience than books edited by A.I.

We should also consider the cost of A.I. Goldman Sachs estimates that investments in A.I. will be close to $200 billion in the current year, but such investments aren’t generating profits, and results are few and far between. On August 18, Sheryl Estrada, writing in Fortune, reported that 95 percent of Generative A.I. pilot programs are failing. Her report is based on an MIT study that found only 5 percent of A.I. pilot programs seeing rapid income acceleration—most such pilot programs produce no measurable impact on profits. In light of such data, we’re seeing more and more reports of companies abandoning plans to reduce their workforces and replace them with A.I.

There’s no doubt that technology eliminates jobs—but there’s also no evidence that the deployment of advanced technologies has caused widespread, long-term unemployment. Technology typically creates increased employment by creating new tasks and occupations. New technologies also produce new products, and the demand for those products leads to expansion, thus leading to higher employment. Cost reduction through technology also leads to increased profits that can increase wages and salaries and reward investors. Throughout history, technology has created new jobs and expanded the workforce, replacing physically demanding jobs with jobs that are less physically demanding.

Experience to date suggests that the number of IT jobs will continue to increase, but that students should focus on computational thinking and upgrade their mathematical skills to prepare themselves for the new careers that A.I. will create as we develop better methods of deploying it. I have no doubt that the success rate of business A.I. deployments will increase, but those companies will need more and more employees with the appropriate skills.

I expect that after the current interruption, Computer Science departments will continue to expand, and I plan to continue my efforts to broaden their scope. At the University of Central Arkansas, we’re aware of emerging trends and are focused on producing graduates who are experts in computational thinking. This fall, three new tenure-track faculty members joined our Department of Computer Science and Engineering. All three made extensive use of A.I. in their doctoral programs, and I expect that they will play a major role in producing the Computer Science talent needed to continue the growth of IT industries in Arkansas.
___________________________________________
Dr. Stephen Addison is Professor of Physics and Dean, College of Science and Engineering, University of Central Arkansas. He is also a Senior Member of IEEE, a member of the Arkansas Academy of Computing, and Secretary of the Arkansas Academy of Science.