With the goal of providing personalized support and empowering learners with autonomy in the learning process, Harvard University demonstrates its commitment to leveraging technology to enhance the learning experience.
Harvard University has announced its plans to incorporate an artificial intelligence (AI) instructor into its curriculum, allowing students to be taught by AI from Fall semester (September) this year.
Specifically, the popular intro-level coding course at Harvard, known as CS50, will be experimenting with a ChatGPT-powered teaching assistant. Professor David Malan, who oversees the course, explained that CS50 has a history of integrating new software into its syllabus, and the introduction of the ‘CS50 bot’ aligns with this approach.
@ ‘Our own hope is that, through AI, we can eventually approximate a 1:1 teacher:student ratio for every student in CS50… providing them with software-based tools that, 24/7, can support their learning at a pace and in a style that works best for them individually.’
Professor Malan mentioned in an interview with Harvard’s newspaper, the Crimson, that they are currently experimenting with both GPT-3.5 and GPT-4 models for the AI teaching assistant. However, developers and software engineers outside the Ivy League have encountered difficulties incorporating OpenAI’s ChatGPT-4 into their workflow.
This has raised concerns about the AI’s coding abilities.
@ ‘Is it just me or GPT-4’s quality has significantly deteriorated lately?’ asked one user on the Hacker News forum, associated with the Silicon Valley start-up incubator Y-combinator.
@ ‘It generates more buggy code,’ the user continued, ‘and overall it feels much worse than before.’
Critics within the software development community have expressed dissatisfaction with the AI’s performance, describing its software skills as ‘significantly worse’ than previous versions of ChatGPT. They have pointed out that it often provides superficial responses and its answers to coding prompts are sometimes inadequate.
Considering the high cost of a four-year degree from Harvard, which currently amounts to around $334,000, students are likely to expect the kinks in the CS50 bot to be resolved by the time the program begins in September.
CS50 is one of Harvard’s most popular offerings on the online learning platform edX. The platform was launched in collaboration with MIT in 2012, but both universities sold it to educational technology company 2U for $800 million in 2021. The agreement stipulated that edX should continue operating as a public benefit entity, providing courses that are also available for free auditing.
While Professor Malan acknowledged that early versions of AI programs like ChatGPT may occasionally underperform or make errors, he believes that the AI teaching assistant will reduce the time spent on assessing the design of students’ code. This, in turn, will enable teaching fellows to allocate more time to meaningful, interpersonal interactions with their students, resembling an apprenticeship model.
In his closing remarks regarding the new CS50 bot, Professor Malan emphasized that college education is not solely about teaching students what to think, but rather how to think critically. He encouraged students to apply critical thinking skills to information they receive, whether from humans or software.
@ ‘We’ll make clear to students that they should always think critically when taking in information as input,’ Malan said, ‘be it from humans or software.’ Higher education is geared towards teaching students how to think rather than what to think about.
Harvard’s introduction of AI instruction demonstrates the university’s commitment to leveraging technology to enhance the learning experience for students in the coding course. While challenges remain in implementing AI teaching assistants effectively, the ultimate goal is to provide personalized support and empower students in their educational journey.
(By Matthew Phelan and Bennett Leckrone)
Read more:
Top 20 Top Quality International Certificate Courses for Students – Data Science & Machine Learning
