ChatGPT and the Classroom Shift: How AI is Reshaping Student Essays

The academic world has been buzzing since the arrival of ChatGPT in late 2022. This revolutionary tool, capable of generating coherent, human-like text, has sparked conversations, debates, and a healthy dose of concern regarding its implications in education. On one side, educators are exploring the potential of AI as a learning ally; on the other, there’s growing anxiety about the moral landscape of academic integrity.

But what does the data say about this transformation? A recent study sought to unpack the effects of this technology on how students write essays in a Massive Open Online Course (MOOC) focusing on AI ethics. This research is particularly timely as we navigate the uncharted waters of AI in education.

Unpacking the Research: What Was Analyzed?

The study, authored by Leo Leppänen and colleagues, took a close look at over 56,000 student essays submitted before and after ChatGPT made its grand entrance. Conducted in the context of a free online course, this research aimed to answer three crucial questions about student submissions:

1. How has the style and length of student writing changed?
2. What shifts have occurred in vocabulary unrelated to essay topics?
3. Are there any noticeable changes in the essay topics themselves?

Given the extensive dataset and established timeline, this study provides a fascinating glimpse into the evolving landscape of academic writing in the AI age.

The Language Evolution: Style and Length

One of the most striking findings from the research is that, on average, student essays have grown both longer and less complex since the introduction of ChatGPT. Before its release, essays averaged about 150 words; a year later, they ballooned to over 230 words. That’s an increase of nearly 50%! And while sentence counts also rose significantly, the length of individual sentences saw only a tiny bump, indicating a shift towards longer essays rather than deeper sentence construction.

You might wonder what this means in practical terms: students are writing more, but potentially with less variety. Their essays have also become easier to read, a detail evidenced by the increase in the Flesch Reading Ease score, which measures how comprehensible a piece of writing is. Higher scores are associated with texts that are more understandable for most readers.

So, while students are producing more text, they’re also opting for simpler, perhaps less original language. This trend raises questions about students’ writing skills and whether the ease of generating longer answers through AI might lead to a dilution of their vocabularies and critical thinking.

Vocabulary: A Shift in the Language of AI

The research also reveals significant changes in vocabulary. Several key terms related to AI and its ethics saw a boost in usage post-ChatGPT launch. For instance, words like “Generative AI” skyrocketed by over 53 times, while terms like “hallucination” (in AI terms, not the dreamy state!) also surged.

Interestingly, more traditional terms like “important” saw a notable dip, suggesting that the incoming wave of AI-related discussions has altered how students express their thoughts. In a nutshell, students are now peppering their essays with trendy AI jargon, highlighting perhaps how they’ve adopted language patterns influenced by AI tools. This shift might mean that students are mirroring AI output, adopting its quirky word choices as part of their own vocabulary.

Did Topics Change?

You might imagine that with all these changes in style and vocabulary, the thematic landscape of essays might also shift. Surprisingly, the study found no significant changes in the overarching topics students tackled in their essays. This suggests that while students are writing longer and perhaps a little differently, the core subjects they’re discussing haven’t transformed dramatically.

So why is that significant? Well, it could indicate that while students are changing their writing styles and using different language tools, the fundamental questions surrounding AI ethics remain constant. They might be experimenting with new expressions but still circling back to the same themes and ethical dilemmas inherent in AI discussions.

Real-World Implications

What does this all mean for the future of education? As we see students increasingly relying on AI tools, we need to ask the tough questions about academic integrity and learning efficacy. If students use ChatGPT to flesh out their essays without engaging in the underlying material, are they truly learning? Or are they just churning out longer essays?

Universities and educators will need to develop countermeasures to ensure that academic integrity is maintained. This could include strategies like increased writing-focused assignments or more in-depth assessments where critical thinking and personal engagement come into play. Perhaps we should encourage students to use ChatGPT as a writing aid, rather than a crutch that undermines their abilities.

Key Takeaways

• Essay Length and Complexity: Since the launch of ChatGPT, student essays have become significantly longer but less complex, raising questions about the depth of their writing.

• Vocabulary Changes: The vocabulary of students has shifted to include more AI-related terminology, suggesting an adaptation to modern discussions around technology and its implications.

• Stable Topics: Despite changes in style and lexicon, the central themes of student essays, focused on AI ethics, have remained consistent, indicating a recognition of fundamental issues despite the changing method of expression.

• Implications for Education: The reliance on AI tools in education necessitates a reevaluation of how we assess and promote learning and academic integrity.

As we move forward, educators can encourage students to think critically and engage deeply with their subjects while also adapting to the new realities of AI technologies. The rise of tools like ChatGPT is not just a challenge; it’s an opportunity to rethink how we approach learning in a tech-centric world.

If you are looking to improve your prompting skills and haven’t already, check out our free Advanced Prompt Engineering course.

This blog post is based on the research article “How Large Language Models Are Changing MOOC Essay Answers: A Comparison of Pre- and Post-LLM Responses” by Authors: Leo Leppänen, Lili Aunimo, Arto Hellas, Jukka K. Nurminen, Linda Mannila. You can find the original article here.