Artificial intelligence (AI) is ubiquitous. Its productivity benefits make it an invaluable tool for many institutions across the world, including governments, companies and universities.
At GA, though, the use of AI for assignments and assessments is discouraged, given its potential to undermine academic integrity. However, research suggests that the effective use of AI can visibly improve student academic performance.
To maximize the benefits of AI while mitigating its risks, GA should formally integrate AI into its curriculum, ideally through adapting an existing AI model to reflect GA’s standards of honesty. By doing so, GA students can develop the academic skills necessary to become more successful and capable.
Reassuringly, adapting AI for educational settings is not unique. Numerous educational institutions and online platforms have employed modified AI models to supplement online instruction.
“[We] tried to implement our own version of ChatGPT…rather than just take away this very nascent technology, which is undoubtedly going to be useful,” computer scientist and Harvard professor David Malan said. “The goal then, through AI, has been to provide an approximation of a teacher [always] being available to students,”
For CS50, Malan’s renowned online computer science class, his team imposed some pedagogical guardrails on ChatGPT to ensure that the technology remains educationally productive for students.
“We instruct the [AI] to follow certain guidelines.” CS50 Senior Preceptor Rongxin Liu says. “Importantly, the [AI] should not be giving any solution under any circumstance and should not be giving too helpful code to the student.”
A survey conducted by CS50’s team found that roughly 73% of users found the tool “helpful” or “very helpful”, suggesting that AI is more than just a professional tool; it enhances student learning.
A meta-analysis by Dong et al. researched student academic achievement across studies investigating “AI-based interventions” in a classroom setting, identifying an effect size of 0.924.
Quantitative and qualitative evidence alike affirms AI’s substantial positive effect on student academic performance. From a purely logical standpoint, an AI-driven approach to learning should be implemented immediately. However, to ensure continuity, I would emphasize a gradual, incremental implementation.
First, AI’s benefits and risks (in education) should be publicly addressed in classes. More often than not, AI is unfairly disparaged, but understanding its benefits are equally, if not more, important for appropriate use of this technology. In many GA classes, this process is already happening.
Second, an existing AI platform should be suitably refashioned to seamlessly integrate into GA’s course curricula. This process would resemble the steps taken by Malan et al. to create their CS50 AI assistant. Specifically, AI could be instructed to only answer questions relevant to GA’s classes, while also guiding students to the answer, instead of immediately providing it.
A case study by Rashid et al. identifies numerous industries that are currently being transformed by AI, including agriculture, education, healthcare, manufacturing, software, among others. AI is not limited to a few, highly-specialized fields. The scope and scale of its growing presence is vast, and many GA students will inevitably be affected by AI in their careers of interest.
Ultimately, through a thoughtful integration of AI into the classroom, rather than a complete restriction, GA can preserve its standards of academic integrity while officially providing students with the skills to learn, adapt, and succeed in the world.
Sources:
Embracing Artificial Intelligence in the Classroom | Harvard Graduate School of Education