Researchers from Dongyang Mirae University in South Korea have released a new paper on arXiv that looks at ChatGPT and its educational impact in the field of computer science. The findings support the growing idea that generative AI (gen AI) can be used to enhance learning for students, despite worries by some that the technology is abused by cheaters.
In this study, researchers Sunhee Hwang, Yudoo Kim, and Heejin Lee took 36 computer science students and began a two-round software development competition that ran for four weeks with students utilizing ChatGPT.
During the first round, all 36 students were evaluated based on performance metrics such as code quality, innovation, and adherence to project requirements. The top 15 who got the best scores were sent to the second round where they were given final rankings and a winner selected.
The researchers said they found that those students who used ChatGPT a lot through the different stages of the project including ideation, documentation, software development, and quality assurance, had higher completion rates and better scores. A comparative analysis between the first and second-round students uncovered a significant difference in their experience with gen AI for software development.
In a post-study survey, the researchers found that 78% of participants agreed AI had helped them develop a deeper understanding of complex topics. Three-quarters of the students said that ChatGPT significantly improved their practical skills and the same number said it had positively contributed to their career development.
Generative AI has had a mixed response from people since its launch. Some have seen clear benefits and others are wary and warn that it's not reliable, uses a lot of energy and water, and as mentioned, can be used to cheat on homework. No matter what your opinion, the fact is that it is here and it is disrupting.
One sector that has been disrupted is educators teaching computer science. Back in May, Neowin reported that many professors are now updating their courses to move the focus away from the syntax of programming languages and to more higher-level skills such as testing and debugging code so that AI-generated code can be validated.
At its current level, gen AI plays a more synergistic role with developers, helping them get over issues they're having with their code and helping them write code faster. The study undertaken in Korea also shows the role it can have in education and it clearly looks as though those students who use it see better outcomes.
Source: arXiv
3 Comments - Add comment