A new study from the University of East Anglia (UEA) has found that while AI-generated essays may appear polished and coherent, they fall short in one critical area: human engagement
The research, published in the journal Written Communication, compared essays written by university students with those generated by ChatGPT, the popular AI tool developed by OpenAI.
Despite the impressive fluency of the AI-written texts, researchers concluded that they lacked depth, interactivity, and personal connection in real student work.
Spotting the difference between human and AI-generated essays
The study involved 145 essays written by students and another 145 produced by ChatGPT. Researchers focused on how writers engaged with their readers, a key feature of compelling argumentative writing. This involved analysing rhetorical questions, personal commentary, and direct appeals to the reader.
Human-written essays stood out for their dynamic and persuasive style. They frequently used questions to provoke thought, offered personal reflections, and addressed the reader directly, techniques that contributed to a more engaging and persuasive narrative.
In contrast, while grammatically accurate and structurally sound, the AI-generated essays were more mechanical in tone. They followed academic writing norms but lacked a distinctive voice or clear stance on issues. The absence of these “engagement markers” made them feel impersonal and less compelling.
Understanding the limits of AI writing tools
The findings showed the limitations of current AI writing technologies. ChatGPT, for example, is trained a lot of text data and operates by predicting word sequences based on patterns. This allows it to generate coherent and fluent text, but it struggles to replicate how humans express opinions, interact with readers, and build arguments with personality and conviction.
The study also shows the importance of human thinking and creativity in academic writing. While AI can assist with drafting and grammar, it can’t be used for the critical thinking, originality, and personal insight educational institutions aim to cultivate.
Helping educators detect cheating
As AI tools become more widely available, concerns about their misuse in academic settings have grown. Teachers and lecturers worry that students might rely on AI to complete assignments, slowing the learning process. The study gives insights that could help educators identify machine-generated content by focusing on signs of authentic engagement and personal expression.
With no proper method yet available for detecting AI-generated text, the research is a good resource for teachers trying to maintain academic integrity. By becoming more aware of what makes human writing distinct, educators can support their students better by developing their voices and writing skills.
AI as a tool, not a replacement
Rather than dismissing AI-generated essays outright, the researchers suggest it can be a helpful educational tool when used appropriately. They advocate for its use in classrooms as a support for learning, not a shortcut. Tools like ChatGPT can assist with brainstorming and grammar but shouldn’t replace the cognitive skills students are meant to develop through writing.
![Figure 1: Sketch of the evolution of the Universe over the last 13.77 billion years. It started with the Big Bang, followed by an extremely short period of rapid exponential expansion. The furthest we can see is the cosmic microwave background, when radiation decoupled from matter, approximately 380,000 years after the Big Bang. This is followed by the ‘dark ages,’ during which this radiation redshifted from the visible regime into infrared and sub-mm wavelengths. The occurrence of the first stars, about 400 million years after the Big Bang, ended this phase, spearheading the formation of galaxies as we see them today. [Credit: NASA/WMAP Science Team, public domain]](https://www.openaccessgovernment.org/wp-content/uploads/2025/05/Fig-1_1200-100x70.jpg "How did the first stars form in space?")
