The education world was stunned last year with the advent of ChatGPT, a Large Language Model that is engineered to generate text and hold conversations. Seemingly overnight, students were able to utilize LLMs to draft essays out of short prompts and complete homework in a fraction of the time. After the dust settled, many teachers responded by restricting its use. Others opted to change their curriculum to ward off generated writing. A year later, many teachers are still leery about ChatGPT. However, a few promising use cases are appearing.
At my school, the entire English department followed many others and moved back to in-class written essays with paper outlines in 2023. The combination of a powerful new technology along with a hyped media frenzy caused many teachers to take a step back and take stock. Given the possibilities of ChatGPT, new methods of teaching and learning were needed.
During a recent meeting, a collection of teachers and administrators from my school had an opportunity to voice their concerns with CEOs of AI startups. Humanities educators were specifically wary. Writing is a core skill in identity development, they explained, and it lays the groundwork for how we develop meaning. The writer, Flannery O’Connor, once noted, “I write because I don’t know what I think until I read what I say.” If LLMs write for students, how could they ever develop an authentic voice?
In response, the CEOs had little in the way of comforting news. Machine Learning, the technology behind AI, is advancing at an incredibly rapid rate. While early models had problems with “hallucinations”, or false information that the model asserted to be true, recent advancements have nearly eliminated them. Algorithms that promised the ability to sniff out AI-generated text have also fallen quite short. Many of them have far too many false positives to be trusted in matters of educational honesty.
My takeaway from this meeting was that the short-term future of education is complicated. That said, some new perspectives on what it means to learn and show mastery could be a way of bridging this divide. One reality of this new landscape is that the process will be far more important than the end product. The days of turning in a research paper or project sight unseen on the last day of the term are numbered. In its place, educators will need to place the focus on revision, the small, incremental steps that students take along the way to an eventual goal.
LLMs like ChatGPT are uniquely poised to be helpful in this regard. Students could bring an outline of a research paper to the chatbot and ask it to find the best order of information. It could also scan an essay searching for redundant points or help to evaluate sentence variety. After an essay is written, the student could use the LLM to develop a reverse outline or a summary of their argument.
Another new reality is that LLMs are incredibly powerful, and this power needs to be utilized in service of the learner’s cognitive thinking skills – not to replace them. Again, If used correctly, AI can be helpful for developing a robust understanding. In a Computer Science class, students can direct questions to a chatbot asking about how a piece of sample code works. They can ask questions over and over drilling down to a deeper level of understanding. Then, once they understand the code, they can then write their own program and elicit feedback.
While this process is tedious and more difficult than having ChatGPT write the code for the student, it provides the opportunity to learn a topic more thoroughly than reading a textbook and drafting a code example. The process of problem-solving with computational thinking is the real work of programming, and if students can bolster those skills with LLMs they will have a broader skill set to work from.
These realities have certainly been disruptive for educators, although the outlook for learners is rather positive. Never has there been a more powerful tool for increasing understanding at their fingertips. In the coming years, educators will be instrumental in crafting the dynamics between the student and ChatGPT. The trick will be in making sure it serves authentic and critical thought.
learn more
ChatGPT Has Entered the Classroom
https://www.nature.com/articles/d41586-023-03507-3
Using ChatGPT to Support Teachers
https://blogs.worldbank.org/education/how-use-chatgpt-support-teachers-good-bad-and-ugly
ChatzGPT as a Learning Tool
https://www.apa.org/monitor/2023/06/chatgpt-learning-tool
How ChatGPT Can Improve, Not Replace, Your Writing
https://www.wired.com/story/chatgpt-writing-tips/
ChatGPT and the Future of Writing
https://aicontentfy.com/en/blog/chatgpt-and-future-of-writing-1
Why ChatGPT Won’t Replace Human Creativity
https://www.linkedin.com/pulse/five-real-facts-chat-gpt-why-wont-replace-human-creativity-rasal/
The Dark Side of ChatGPT
https://poolmarketing.medium.com/the-dark-side-of-chatgpt-has-real-world-consequences-90bff03a00bf