Don't you think it's absurd that most university IT researchers are focusing on AI and machine learning?

Nearly all breakthroughs in AI are not coming from university research. Regardless, it is not right for so many to jump on AI/ML research.

This article was selected by Medium curators as “Boosted” on 2025—02-03.
Medium Members: read this article on Medium.

Today, I had a conversation with my classmate, who is now a Computer Science Professor at the university (China) I attended. He said that nearly all research topics there, including him and others, are focused on AI and machine learning. His son is currently pursuing a PhD at another university, whose research topic is also Machine Learning. I think suddenly such an intense focus on AI research at universities is absurd.

Credit: the classic “Silicon Valley” TV Show, season 2 episode 3

I was not surprised to hear that though. When Courtney, my daughter, chose her thesis project for her honours degree (in Australia) ~2 years ago, over 80% of the options (I saw the list) offered by lecturers were focused on AI or machine learning. When studying the AI & machine learning courses, Courtney didn’t find the content interesting or practically useful (due to the knowledge of lecturers. I could tell from the assignments as I taught at Uni before), so she decided to propose her own topic (Mobile Test Automation) and find a supervisor willing to supervise her.

A couple of interesting things happened. Two senior lecturers from a top university in NSW showed interest in Courtney’s proposal and arranged an online interview (this was still during the COVID period). Courtney presented a demo and explained her approach. One associate professor later asked, "You mentioned Continuous Delivery—was that a term you invented?"

Throughout Courtney's studies, I advised her to often use Selenium automation in her presentations, which typically left a strong impression on the audience, including the lecturer. On one occasion, a lecturer asked Courtney, “Was that (referring to the automation execution in the browser, powered by Selenium) AI?” It was clear this lecturer didn't understand either AI or automation.

(By the way, these two Australian universities mentioned above are both the world’s top 50).

Software advancements are usually not driven by university research

I worked for 3.5 years at a university research centre, so I'm quite familiar with university research. In my opinion, much of the computer science research at the university level isn't practically useful.

The real software advancements have come from practical application and innovation, such as:

• JUnit Testing by Kent Beck & Erich Gamma

• Test-Driven Development (TDD) by Kent Beck

• End-to-End Test Automation by Watir

• Refactoring (by Martin Fowler, and materialized by JetBrains)

• Continuous Integration (by Martin Fowler) and the implementation of CruiseControl at ThoughtWorks

• Selenium by Jason Huggins at ThoughtWorks

• Selenium WebDriver by Simon Stewart (ThoughtWorks, Google, and Facebook)

• Ruby language by Matz

• Continuous Delivery (by Jez Humble and David Farley, ThoughtWorks)

• Convention Over Configuration by DHH (at 37signals)

One exception is Design Patterns by Erich Gamma, which originated from a Ph.D. paper (note, by a student, not a lecturer). A close one is Minix, which was created by Professor Andrew S. Tanenbaum for teaching OS, rather than strictly as a research project (for publishing papers).

Major AI breakthroughs almost exclusively come from well-funded companies