Often when we think of AI, images of futuristic robots or voice-activated devices spring to mind, capable of executing complex tasks or answering queries with the speed of thought. However, AI’s impact is a little less science-fiction and a little more just making the tasks we do already, a little bit easier.
An area which AI is set to have a large impact on, with the advent of systems like ChatGPT, is English Language Training (ELT).
The global approach to learning languages is changing, and rapidly so, thanks to the use of AI systems in teaching, studying and assessing. No longer confined to traditional classrooms or bound by the limitations of conventional teaching. English learners across the globe can now access personalised and interactive resources through AI-driven language models. These programs not only offer immediate, personalised feedback but also foster an engaging learning environment that adapts to the individual’s pace and proficiency level.
But with AI providing a vehicle for change in the way we teach, learn and assess English, what impacts will this have? Will it make things easier for students or could it actually stand in their way?
The Rise Of AI In English Language Learning
When AI language models like Google’s Bard or OpenAI’s ChatGPT were released, they automatically presented opportunities in making ELT much more streamlined and accessible. They have an unrivalled ability not before seen to process and analyse large data sets at record speed, which means they can create tailored learning for students and assess them in real-time faster than any English teacher ever could.
Platforms powered by AI, which use incredibly sophisticated algorithms to assess a learner’s current proficiency level, learning style, and pace are then able to instantly use this data to great effect in tailoring learning content and then giving students instructions or tasks that are just right for their skill level and learning goals.
Another key feature of AI in ELT is the provision of immediate feedback. Unlike traditional learning environments where feedback can be delayed, AI systems offer real-time responses to learners’ inputs. It could be a case of providing immediate grammatical corrections or assessing the use of spoken vocabulary – having an immediate feedback loop between system and learner rapidly accelerates the learning process and helps stop common mistakes at source, before they become ingrained in a person’s use of English.
The final major change that is being noticed with AI’s involvement in ELT is how much more interactive it can make the learning process. AI apps and platforms can provide people with many different learning experiences, from interactive lessons and conversational practice to comprehensive grammar and vocabulary exercises, making language learning accessible and engaging to a global audience.
AI In The IELTS Exam
The International English Language Testing System, known more commonly as just the IELTS exam, is recognised as the global benchmark when it comes to assessing the English language proficiency of non-native speakers.
But how is this shift toward AI driven ELT affecting one of the world’s most well-known English tests?
Well, first of all, the learning methods and resources we spoke about above are the first things that learners studying for the IELTS exam are noticing. The AI Language models can offer these personalised learning experiences, instant feedback and interactive learning.
They can also go a little further if given the right prompts and guidance; for example, AI applications can mimic the format and difficulty of IELTS tasks, providing learners with an immersive experience that closely resembles the real exam.
However, there are some concerns on how widely AI should be used in the IELTS exam. As research for this article, we spoke with Andrew Turner from English With An Expert, who is an IELTS Tutor with decades of experience in teaching English to non-native speakers.
In our discussion, Andy spoke on both the opportunities and challenges presented by AI in language assessment. He highlighted research from MIT that raises concerns about inherent biases in AI systems. These biases stem from the training data used, which might not fully represent the diversity of English language speakers worldwide. Instances where native speakers, and even English teachers, have failed AI-marked tests due to accent recognition issues underline the debate on the fairness of AI in language assessments.
As well as this, Andy also stressed to us that “understanding the nuances of the English Language – such as slang, jokes and emotional cues is something that AI is not adept at enough for it to be considered as an acceptable way of grading candidates in the IELTS exam. Human examiners are adept at interpreting these subtleties, an ability AI has yet to master. Without human judgement, a candidate’s response in a particular area of the exam, which is perfectly acceptable and correct when viewed in the right context, could easily be misinterpreted by an AI system that does not fully understand said context. In an exam so costly and so crucial to many students around the world, replacing the human assessors with AI language models for the IELTS Speaking and Writing components is something that should be considered very, very carefully before any decisions are made. For the same reasons, test candidates should also beware of using AI without any human input when preparing for the test.”
It’s clear from Andy’s insight and the research done by MIT that despite the value AI platforms are adding to ELT globally, there are significant drawbacks and limitations here too! While AI can offer unparalleled opportunities for practice and preparation, the final judgement on a candidate’s language proficiency, for now, remains better served by human expertise.
What Are The Future Possibilities For AI In English Learning?
There are the obvious success stories here, as well as where AI systems have some shortcomings in assessing English. But what does the future hold? AI is evolving more and more every day, and whilst the internet took a long time to evolve into its current form, many people are aware that AI, with its ability to learn on its own with often minimal human interaction, will grow much more rapidly.
The next generation of AI in language learning is expected to feature even more sophisticated algorithms capable of understanding and interacting in natural language with unprecedented accuracy. Future AI systems may overcome current limitations in recognising nuances, slang, emotional cues, and cultural contexts, making AI-driven language learning tools more versatile and effective, which could make AI-assessed testing more of a possibility.
