UTTELTThe Reality of Artificiality

The Reality of Artificiality A breakdown of AI in education from internal experts

Since the initial release of ChatGPT in November 2022, interest and news stories have been booming in the topic of artificial intelligence. From the various applications to the ethics of it: you have seen it all zoom by. However, the baseline is quite simple, with the release of ChatGPT AI has become widely accessible to the general public. The educational consultant from telt and the e-learning specialist from eemcs help you get an overview.

Education is already affected by this in these news stories and the release of ChatGPT has marked a sudden influx of further conversations about the usage of AI in education. The tone in these conversations is marked by variety, ranging from seeing opportunity in the development of such a tool, to a certain apprehension of the possible wrench this tool might throw in the cogs of our educational system. In this article we will outline the different currents of conversations and discuss what this might mean for our education.

State of the Union

But before starting this conversation, it is good to be aware of what is currently out there. While ChatGPT is currently the most prominent AI platform, it is by far not the only platform that is available to a wide audience. ChatGPT offers, as the name suggests, a chatbot-like experience. Yet there’s more AI-driven tools, such as DALL-E: the image generator by OpenAI, in which you are motivated to give highly specified descriptions of imagery to be generated. The banner of this article was also generated.

Furthermore, Replika has also been released. Replika is a chatbot that, instead of just generating text, offers companionship and friendship to people. Then there’s existing tools that (start to) incorporate AI in their offerings. Take for instance Photoshop with Neural Filters, allowing you to manipulate images through neural networking. With all these different developments you can only expect the boundaries of what is possible to be pushed further.

ChatGPT

Let’s focus on ChatGPT, with it currently being the most popular contender. You can ask it simple search-engine prompts, such as “Where is the University of Twente located?”. But beyond that you can also ask it to draft an essay for you on any topic of choice. Because of its chatbot nature, you can reply to it and ask it to adapt its output. You can even give it your already written text and ask it to ‘make it more attractive’ or give it a code you have written and ask it to clean it up. With the release of this platform, it’s not the question whether AI will become part of our education, but how. Let’s explore a couple of use-cases where both teacher and student can benefit from using ChatGPT in education as of now.

Generation

The specific feature of AI that has stirred so much discussion over the last few months, is the ability to generate texts. As part of experimenting with it, we tried to generate the script for a microlecture. This video is the result of it. Next to the script, the graphics used in the PowerPoint have also been generated by AI input, using DALL-E. The drafting of such texts can be applied for multiple purposes in education and (partially) automate time-intensive tasks, such as scripting, example texts and codes. As already demonstrated, AI can write a decent script for a microlecture and even the AI suggested the script could be finetuned by the end-user.

Besides instructional materials, you could also generate assessment with ChatGPT. For instance, specific questions for exams or more open-ended philosophical questions/assignments for an essay. Generally speaking, it’s best practice to specify your prompt with details you want included. But just like we’d like our students to remain critical in their thinking, we should do that of any output generated as well.

Manipulation

Besides generating content, AI can also manipulate or comment on existing content. This can be for both technical and non-technical applications. For instance, completing texts or even explaining programming code. As an example, we took the Shaping 2030 mission of the University of Twente and asked to change it into a fairytale.

Falsification

Finally, there’s the elephant in the room: plagiarism and using AI to write up assignments. The ability to compose a text from just a prompt can already be uncanny. All the previous examples can be used for fraudulent intent as well. From generating necessary programming code to a full essay of reflection a student must do. As an example, we’ve asked it to reflect on a group project. Some extra details we gave included an experienced difficulty, the project involving system design and reflection on what they would do differently.

Because of its AI nature, you don’t have a standard input-process-output model with the same output every time. That makes automatically detecting it a lot more difficult, and even though there is development on checking whether text was AI generated, this would just seem like a game of cat-and-mouse on the longer end with more different types of AI being developed.

Conclusion & evolution

All this being said, these are still early stages of what is yet to come in AI. Yet, these applications are already available for the public, which begs for acknowledgement by the public of its existence and usage. Or to put it simply: how will we in education facilitate this?  While options like detecting, restricting, and even banning might offer a short-term solution, this will not stop the development of this technology. It is better practice to welcome AI and see how we can use it.

Think of a past development we’ve embraced, like the calculator. In many ways, it made us ‘lazy’ yet for learning the basics of calculations you just need to do it by hand. Even beyond that, you can only use a calculator if you know what buttons to press and what they mean. It’s in this analogy the type of reflection and critical thinking that are required, even more so with the use of AI. This requires some rethinking of how we do our assessment: sure, you can let AI do work for you. But as Descartes once described it: cogito, ergo sum. That’s something an AI cannot do for you.

B.T.M. Ernst MSc (Babs)
e-Learning Specialist EEMCS
R.H.M. van Emmerloot BSc (Robin)
Educational Consultant @ TELT