Since ChatGPT was launched in November 2022, a whole slew of generative AI tools have been developed. I took on the challenge of writing a post about this topic, realising very well that anything I write today can change completely by tomorrow. Between mergers and changes in the companies that offer these products, the discussion on how much of the source code should be open code, and the rapid change of the products themselves, the possibilities seem to change day by day.
At the same time, generative AI tools are out there. The reality is that our students are using these tools and that more and more professionals are using these tools as well. As such, I think it is our responsibility as academics and educators to understand the pitfalls and opportunities of these tools, reflect broadly on the ethical considerations, and prepare our students for a labour market that will rely increasingly on generative AI tools.
The first step for us as academics is to explore how we can use generative AI tools in our daily activities in a way that helps us understand the possibilities, see the limitations, and reflect on the ethical challenges. In today’s post, I will identify some applications that I have attempted myself, to inspire you to try out various applications and learn about the possibilities for your day-to-day activities.
Opportunities for teaching
In my teaching, I have used generative AI (and in particular ChatGPT) for the following tasks:
- Check my exams: As my students are allowed to use their computer during the exam, it is difficult to control if they will throw my exam questions into generative AI. I am one step ahead by putting my exam questions into ChatGPT. If ChatGPT struggles to come up with a sound answer, I know that I designed a good question. At the same time, I know that today ChatGPT may struggle with my answer, whereas tomorrow it may be better at solving it.
- Turn course text into multiple-choice questions: When I needed to make a lot of multiple-choice questionnaires, I asked ChatGPT at some point to help me generate ideas. The outcome was a bit of a mixed bag: some questions were good, most questions required tweaking and rewriting, some questions were absolute garbage. But, having something to start from helped me move forward when I ran out of steam.
- Develop a GPT for my class: I would like to have a custom-made GPT for my courses, that is well-trained on all my material, and that can really support my students’ learning. I have created such a GPT with all my material, but it requires much more training (and much more time and effort from my side) to reach the desired outcome. So far, I am not happy at all with my custom GPT.
Opportunities for research
Using generative AI for our research may sound like cheating. By all means, I am not suggesting that you use ChatGPT to write your papers (embarrassing examples are already out there), nor do your research work. However, there are some applications where generative AI can be of assistance:
- Debugging code: Generative AI is already so far that we have Devin the software engineer. While I have not outsourced complete programming of code to ChatGPT (I like to design my code structure myself, so that it makes sense to me), I have found it a very useful tool for debugging my code. In particular, as I recently did efforts to learn Python, I found that I could progress more quickly with my learning by asking ChatGPT why my code was giving me a certain error. Sometimes, it was something trivial that I simply did not see (such as mistyping my iterator i as 1), but in some instances the error really came from not knowing the inner workings of Python code yet.
- Understanding call for proposal requirements: Some calls for proposals are combined with many annexes of requirements and stipulations. I recently uploaded all the PDFs of one call into ChatGPT (after reading the documents myself, of course, but struggling with the legalese), and then asked it targeted questions regarding whether certain types of costs could be expenses on the project or not, with the argumentation, and where I could find that particular information in the document. I ended up confirming all these questions with an expert as well, and ChatGPT had given me proper guidance in the majority (but not all) of the cases. My advice here would be to proceed with caution, and always check with an actual human in the grant office.
Opportunities for administrative tasks
The idea is that generative AI is most suitable for relatively repetitive tasks. I am still exploring how I can leverage the tool in some of my administrative responsibilities. So far, I have only found one good application:
- Turn spreadsheets into a task list: As editor in chief of a journal, I keep a large spreadsheet with the status of each manuscript, and who is responsible for which activity each week. I have used ChatGPT to turn this spreadsheet into a list of responsibilities for each person, which makes it slightly easier for all involved: instead of having to search for their name in a long spreadsheet, they now get a list of actions. Still, I need to retrain ChatGPT every week for this task, as it seems to forget what to do, even when I upload the updated sheet into the same chat. Certainly, there is room for improvement in the tool there.
Staying on top of developments
The generative AI sphere develops super fast, so what can we do to stay informed? Realistically speaking, as academics, we do not have hours available each day for learning about all the ins and outs of the AI world. Here is what I try to do to stay informed:
- Talk to others: Perhaps the most important way to stay informed is by talking to colleagues about the topic: how are they using generative AI? How do they see their teaching and students influenced by it? How do we move forward in higher education with this tool around us?
- Take a prompt engineering course: I took the Coursera Prompt Engineering specialization to get a better understanding of this whole field, and many universities are offering short half-day introduction courses for their faculty development.
- Listen to podcasts: I am not able to stay up to date with all the news in the AI world, but every now and then I will listen to an AI-themed podcast. Some of my favourites include the Everyday AI podcast, How do you use ChatGPT, and the AI Chat.
- Follow experts on LinkedIn: In addition, I follow on LinkedIn thought leaders in my field and in higher education who are leveraging AI to learn from their expertise.
Conclusion
Generative AI is changing higher education, and it is our task as academics to understand its implications. The first step in this task is to explore the tool in our work so that we are aware of its use and the opportunities, as well as the tasks it cannot carry out properly (yet). We need such insights to update our teaching, and to be able to discuss the ethical implications.
I enjoyed reading your piece. As a retired academic, I found your analysis interesting. I’ve been using ChatGPT recently to develop the plot of a novel I’m working on set during Covid. I’ve found it useful for generating potential ideas and themes for chapter. Thanks again. (P.S. I wrote one of the other articles just published here-https://career-advice.jobs.ac.uk/career-development/working-overseas-the-pros-and-cons/)