By Josh Siatkowski | Staff Writer
In computer science, universities are falling behind the rapidly evolving AI ecosystem — even professors admit it.
“I think computer science is still a very promising major,” said Dr. Chen Zhao, assistant professor in the computer science department. “But the only thing I worry about is [that] the things students learn in school are outdated.”
When the release of ChatGPT brought large language models to the forefront in 2022, it also sparked a flurry of private-sector breakthroughs. Bigger and better models and the introduction of AI agents have created a challenging disparity between newly relevant skills and what’s taught in the classroom.
“I think this is hard because if you want to fill in the gap, you need to ask the universities to have a more advanced class,” Zhao said. “For example, we do not have a class that’s only focused on large language models.”
Zhao, whose work focuses on fairness in machine learning and AI, is at the forefront of this revolution. He said that breakthrough papers — increasingly published by private tech companies and not universities — are coming out so fast that even researchers are struggling to keep up.
“I tell my own PhD students, ‘If one project cannot be accepted as a paper in six months, then you do not have a chance to resubmit,’” he said.
For students who want a career at the center of technological innovation, the gap between education and industry innovation has made the most competitive jobs even tougher to secure. Upland, Calif., senior Omar Darwish, an incoming data engineer for IBM, had to do most of his AI learning outside the classroom. He knocked on the door of every computer science professor looking for research opportunities as a sophomore, landed multiple internships — including a machine-learning role at financial firm Hilltop Holdings — and even started his own company to understand the state of the industry today.
“It definitely was a very, very large amount of commitment and work that started very early,” Darwish said.
Darwish agreed with Zhao that Baylor’s curriculum is outdated. But that’s not unique to Baylor, nor is it necessarily a problem. The challenge, he said, is the lack of a “build culture,” in which students are encouraged and expected to start projects outside the classroom.
“If you’re at a school like the University of Waterloo, or a school like Georgia Tech, … it really doesn’t matter if the curriculum is outdated or not applicable because there’s a build culture there,” Darwish said. “There’s a culture where every student that you talk to at the University of Waterloo is a founder … In that culture, it doesn’t matter if the curriculum is outdated because you are a part of an atmosphere that encourages learning more, that encourages learning what isn’t necessarily being taught in class.”
But even as new research puts fundamental computer science skills further back in the annals of technology, it doesn’t mean those classroom lessons are irrelevant — even if students don’t end up using them on the job.
“I can think of so many things in my classes that I’m just never going to use,” Lubbock junior and computer science major Carter Lewis said. “I think it’s to teach you how to learn new technologies and how to learn new skills, so that whenever we enter the workforce, we’re able to keep up with new technologies. And so because of that, I think I will have the ability to keep learning and to stay on pace.”
Lewis is currently a software engineering intern at Texas Farm Bureau Insurance, and this summer he’ll join the IT department of defense company Textron. For him and the many others who opt for computer science roles outside the realm of Big Tech, Baylor’s curriculum and some personability were all it took to land a solid job. And while AI will definitely impact both software development and IT roles, Lewis doesn’t see it as something that can replace the numerous human-led tasks required to get a project from beginning to end.
“I think a couple of years ago, I really started to research [AI] and look into it,” he said. “I was a little worried. Once you start using AI to complete whole projects, you realize that AI is just not very good at completing entire projects.”
For Zhao and Darwish, both of whom believe in AI’s capabilities, this is also true. Software jobs might look different in the future, but they won’t go away — not for the ones who adopt the revolution, at least.
“I’ve never been particularly afraid because I know that there’s a level of intellectual proficiency and creativity that’s required to be a software engineer, that’s required to be a data scientist, that’s required to be a data engineer,” Darwish said. “You have to be able to solve problems, complex problems. That’s something that I don’t think artificial intelligence is at the point of being able to do right now. But even when it does get there, I think there’s always going to need to be an engineer there to interface with it, and also, part of being an engineer is being somebody who’s personable, being able to talk to people.”
And it won’t just be soft skills that matter as AI adoption increases, Zhao said. As industry-specific LLM development (as opposed to typical universal models) grows more common, so will jobs in that area.
“A lot of companies actually, they are recruiting [software development engineers] that can develop those kinds of software for them, for a specific business, for specific domains,” Zhao said. “That’s a long-term job. … There are a lot of downstream tasks.”
But at the same time, the job market for tech could still compress to a “very, very competitive” level, Zhao said. The key to standing out, Darwish said, is not only to use AI for technical projects but to refine nontechnical ones to become irreplaceable. That idea is why he signed on for the client-facing, consulting-based position at IBM instead of a strictly code-based job at Cisco.
“Cisco is very technical — headphones on, hood on, slinging-code-all-day type of role,” Darwish said. “Those are the ones that are threatened … Those are the roles where you can be replaced. But if you’re an engineer that can speak the business and also solve problems on the back, that’s not replaceable by artificial intelligence.”
