Rethinking data science education in the age of genAI
how to we know what students know when chatGPT can do the tasks we have traditionally relied on to assess learning?
I spent a lovely few days in Melbourne last week visiting Monash and attending WOMBAT (Workshop Organised by Monash Business Analytics Team).
Di Cook and her collegaues in Department of Econometrics and Business Statistics at Monash have a tradition of naming software packages, events, and even their research group after Australian animals and the WOMBAT event is an annual excuse for R people from academia and industry to come together and share new packages, practices, and problems.
When Cynthia Huang (who was organising the program) invited me to speak about the impact of generative AI on data science education, I was excited to dive into how large language models (LLMs) are changing the way educators are thinking about what students need to learn and how they need to learn it.
If the #positconf2025 program is anything to go by, it seems that data science practitioners are “all in” on LLMs. The new Positron IDE has an AI assistant and a DataBot agent embedded and Posit are developing packages like ellmer and chatlas that make using LLMs within your data science workflow easy. The landscape is changing rapidly though, and Posit also launched a fortnightly AI newsletter dedicated to AI updates.
There is no doubt that using LLMs once you already know your way around R can be useful, but what about learning? I wanted to dig into what students are doing with LLMs, what universities are doing to counter misuse, and most importantly, what we as a community of data science educators are doing (and can do) to support students in learning to navigate the new learning environment they find themselves in.