Correction: A previous version of this article misstated the name of ACM. It’s official name is the Association for Computing Machinery.
“In the past, some projects used to…take like about a month. Now they take…two to three days.”
This is how Dr. Vibhav Gogate, UTD computer science professor, described the effects artificial intelligence, or AI, coding tools have had on some of his classes.
The rise of artificial intelligence tools and the large corporations which create them over the past few years has been precipitous, and has affected a wide variety of very different parts of society. One of the sectors it has changed most dramatically, however, is coding and computer science. Even without considering that many of the largest economic players in that field now focus to a large degree exclusively on AI, the AI-based tools released for coding have had an astounding effect on the field.
The most obvious example of this is the rise of vibe coding, defined as generating code using AI rather than creating it by hand. The term was created by OpenAI co-founder Andrej Karpathy in February 2025, and already this practice has become very common among coders, with some sources reporting that as many as 75% of programmers regularly make use of these AI tools.
Such a rapidly developing sphere is, naturally, rife with competition. From industry giants like GhatGPT, Gemini and Copilot to more focused tools like Replit, Cursor and Claude Code, there are a large number of different companies creating tools for vibe coding.
These tools provide a lot of benefits to coders who utilize them, most notably reducing the amount of manual work necessary to complete any project.
Elijah Walker, president of the Association for Computing Machinery, or ACM, chapter at UTD, said “I remember a year ago…we would be on Stack Overflow for hours trying to find, like, some specific question that somebody had. But now…it takes two seconds…to use a specific prompt and get [the answer] back.”
For students, this can mean a significant change in how assignments are approached. According to Walker, AI coding tools are now capable of doing up to 80% of the work required for some classes.
“Most CS students, at least as far as I’ve seen, have started using AI tools to…complete their projects,” said Anirudh Vayalali, a director of ACM’s development arm.
Gogate was clear on one of the ways professors have responded to these changes. “Let’s say ten years back or five years back even, the projects used to be like about 50 to 60% of the grade, right? Now…they are only like 20% of the grade.”
While there are some classes where AI tools can handle most of the workload, it does not extend to every course. While AI tools may be very efficient when it comes to solving simple, structured problems, the more complex queries that come up in higher-level or research-based courses are entirely beyond these programs, said Gogate.
However, the usage of these AI tools is not purely beneficial. In a study conducted by Anthropic, the company behind Claude Code, one of the most used vibe coding tools, developers were tasked with completing a set of objectives using a new set of skills none of them were familiar with. One group was told to learn how to do these tasks on their own, and another group was permitted to use AI tools. When taking a short quiz on concepts they had just used, participants in the AI group had scored 17% lower than those who coded by hand, without AI assistance.
But this one outcome makes the results of the study look simpler than they really are. “Importantly, using AI assistance didn’t guarantee a lower score. How someone used AI influenced how much information they retained,” the study said.
“It’s the people that really press into the learning process that are going to get something more out of these, like, coding tools, right? Rather than kind of using it as a substitution and relying on…general intelligence to replace their own,” said Walker.
Gogate said something similar, and remarked on how students who were really serious would now have a much easier time learning as much as they wanted now that they have access to these AI tools.
Using these new tools unwisely could lead to seriously unfortunate learning outcomes. But for those who are committed to learning and getting better at coding and programming, these new tools can be extremely powerful at helping them not just get results, but improve their own skills as well.
A computer running Cursor, one of the leading “vibe coding” tools. Cursor generates code in practically any coding language based on prompts in simple English, as well as 11 other languages.
The Engineering and Computer Science South building on the UTD campus. Both students and professors in the Computer Science department are grappling with the rise of AI tools.
The Engineering and Computer Science South building on the UTD campus. Both students and professors in the Computer Science department are grappling with the rise of AI tools.
The Engineering and Computer Science West building on the UTD campus. The rise of AI tools for coding has had a dramatic effect on classes in computer science.
The Engineering and Computer Science West building on the UTD campus. The rise of AI tools for coding has had a dramatic effect on classes in Computer Science, as well as many other majors.
