Recently I attended the International Association of Teachers as a Foreign Language, IATEFL conference. I intentionally selected generative artificial intelligence, GenAI, sessions to attend over four-day event. I watched and participated in fourteen of these presentations and workshops. After the conference, I reviewed my notes and identified commonalities offered in these sessions. I found some of these predictable while others were more obvious.
Ban AI?
All of the speakers agree that this is impossible. Learners and teachers have access to hundreds of GenAI tools and are using them to various degrees.
Students and GenAI usage
In societies that have the wealth, infrastructure and means students are using GenAI, however not all students are using it. Clear as a muddy puddle, right? Within these societies, at the higher tiers of education, more than half of students are actively using GenAI to assist their pathways through study guides, study buddies, practices assessments and even to generate assessment submissions.
Societies without the means have a much lower usage of GenAI.
Deepening the digital divide
Throughout the short history of digital education technologies, unequal access to digital tools, networks and digital literacy training have all resulted in a digital divide. GenAI tools are deepening this state. Access or lack of access to the infrastructure required by GenAI technologies, as mentioned in the previous paragraph, contribute to widening the digital divide. Another reason for this, among many, is the English first nature of GenAI tools. Students with native English language skills, including those with accents, hold advantages when engaging with GenAI tools.
Hallucinations
If I only had a pound for every time I heard the term hallucination at this conference, I could have afforded a deluxe tour of Edinburgh Castle. Hallucinations are inaccurate responses through GenAI’s fabrications or over confidence when filling in the LLM’s knowledge gaps. Some of the IATEFL speakers reported that this occurred because the GenAI tool was trying to “please” the prompter with a response. Odd as it sounds, they were serious about this statement. GenAI tools sometimes make up answers. Sometimes hallucinations occur because it does not have the requisite information to form an accurate response, but the LLM has a confidence to believe that it is correct. It is starting to sound like the GenAI is a little too much like us!
Biased Responses
Since most of the training of GenAI large language models, to this point, is based on Internet data, there is a strong bias for English and western culture, race, gender, political, socioeconomic, geographical, religious, well you get the idea. An example of this is GenAI generated images. Typically, these portray male, Caucasians if specific modifying prompts for characteristics are not provided.
Teacher materials development
Teachers are now streamlining the development of original learning materials with GenAI tools. This is also resulting in more and more enhanced learning opportunities as GenAI tools function as a codeveloper in this process. GenAI tools can take on the mundane tasks: creating quiz question distractors, generating the first draft of a listening script, creating visuals, seeking out sources of materials, or assembling rubrics for activities are potential examples of these tasks. The debate on the quantity and quality of these learning objects is passionate and ongoing, so I will leave this to better minds than my own.
Final Thoughts
In this first part post of common issues related to GenAI and language educators I briefly raise awareness of potential challenges for our sector and the adoption of GenAI. I am relaying these commonalities to you to promote thought and discussion. You may take more time to research and engage in discussion about GenAI and your teaching to become an ethical GenAI practitioner. In June, part two will complete the common issues of GenAI sessions at IATEFL.
Relevant Link
IATEFL, https://www.iatefl.org/about
John Allan @mrpottz
Hi—I'm John Allan. I am an educator who works in the technology enhanced language learning field. I create online learning opportunities on various projects. I have ESL and EFL teaching & training experience in Canada, the United States and the Middle East. I hold an MSC in Computer Assisted Language learning, a M.Ed. in Distance Education, TESL B. Ed., a B.Ed. (OCT), and a variety of TESL relevant certifications from TESL Canada, TESL Ontario and the Ontario Ministry of Education. For more articles, learning objects, projects and blog links see https://www.linkedin.com/in/johnharoldallan