Here to stay: the dilemmas of AI in education
By Ernestina Alvarez Corona
ChatGPT has sparked widespread debate and interest since its debut in early 2023. This cutting-edge chat-based large language model (LLM) has sparked considerable interest as well as numerous debates about the future of work and education.
In a previous article, we addressed the rising anxieties that the OpenAI software triggered in the education sector regarding its implications on learning and plagiarism and whether ChatGPT should have a place in schools in the UK.
Now, we are beginning to see more institutional responses to the dilemma of AI and schools. Students enrolled in the International Baccalaureate programme may use content generated by ChatGPT in their essays as long as they credit the source and do not claim it as their own work. The usage of the AI chatbot in evaluations will not be prohibited by the IB.
According to Matt Glanville, the International Baccalaureate’s head of assessment principles and practise, artificial intelligence technologies should be viewed as a standard component of modern life, analogous to spell-checkers, translation tools, and calculators.
Glanville has reported that the use of AI language models can be seen as an opportunity to switch from essay-writing assessments to more analytical tasks. In his own words, “we need our pupils to master different skills, such as understanding if the essay is any good or if it has missed context, has used biased data or if it is lacking in creativity”.
At the end of the day, education aims to give pupils the analytical and moral skills to live righteously in contemporary communities. As AI is widely implemented in more fields every day, the younger generations should develop competencies that allow them to thrive in a society in which AI and other digital tools are embedded in its infrastructure.
A matter of style
Over the last few months, there has been a certain hype regarding the obsolescence of essay writing after the launch of ChatGPT. Although thought-provoking and stimulating for debate, the hype may not be entirely accurate. Admittedly, the traditional, heavily structured approach to essay assessments has been challenged by AI language models.
LLMs excel at creating well-written and structured texts. This is exactly why some teachers are integrating chatbots in their classrooms as a brainstorming or drafting tool. LLMs can provide a decent skeleton for formulating organized sequences of arguments. The reason behind this is that LLMs can provide a strong foundation for organizing a logical sequence of arguments in written work.
However, it is essential to acknowledge that there are limitations to LLMs when it comes to producing writing that is personal and reflective. While these tools can be useful for producing a framework or outline for written work, they may not necessarily capture the “essence” of the writer's experiences and perspectives. As highlighted by Darren Coxon in his latest newsletter, there is an element of the uncanny in the texts generated by ChatGPT. The writings produced by these technologies do not have the same feeling as those written by a human being. Instead, they are written in a detached and austere manner, influenced purely by logical sequencing and prediction rather than actual perception and understanding.
Given that the LLMs do not have sentience and lack the experience of the phenomenological world, it is not game over for educators and essay production. There is still a lot to teach in terms of writing and communication style at schools now that textual coherence and cohesion seem easily granted. Educators can play a vital role in helping students to cultivate their communication skills by encouraging them to engage in reflective and engaging writing that grapples with the complexities of the lived world.
Mind the hacks
I must admit that most of my social media feeds have been filled with AI content for quite some time. Prior to the launch of ChatGPT, my TikTok, Reddit, and Instagram homepages consisted mostly of news about OCR and image recognition and people experimenting with DALL-E or Midjourney. Now, it is not only about news but also about know-how.
With the boom of LLM tools, users are sharing different ways of using AI tools with the hope of exploring the limits of AI. From producing prompts that "induce sentience" to the joint use of multiple AI tools, young users are exploring and sharing tricks and hacks on how to write their assessments.
Recently, I saw a TikTok video of a girl who used ChatGPT for a university assignment submitted through the Turnitin platform and got caught. Evidently, now the assessment submission and plagiarism platform can detect GPT-generated content in documents.
Nevertheless, the mentioned TikTok was then stitched with another creator's video that explained how to "hack" Turnitin. To avoid getting caught cheating, social media users suggest using other AI-powered tools to alter the text and paraphrase it to evade ChatGPT detection in their submissions.
Another common practice among students is to pass ChatGPT content as their own by crafting more complex prompts. It is not uncommon to find students sharing how they induced "personality" or "sentience" by giving the LLM a series of detailed prompts that would emulate human expression.
Users who engage in more elaborate prompt writing, create a context for the AI to generate more detailed or personal outputs and share which writing strategies create the most appealing results. Some even combine this contextual approach with other AI tools, like PromptPerfect, to balance the subjective approach with the AI’s programmatic language to get the most out of the language model.
As AI users experiment with the capabilities of LLMs, educators should be aware of how pupils use these tools and encourage them to use these tools with a more analytical approach. With or without chatbots, understanding and developing different hues of verbal and written communication remains a fundamental part of a good education. This means teaching students how to differentiate a variety of written styles, critically evaluate the output of LLMs and understand how to use these tools in conjunction with their own ideas and insights.
