TESOL-SPAIN, Sevilla 2026 - Ben Knight: Should we even try to assess our students’ writing skills in a world of Gen AI?
TESOL Spain Convention 2026, Sevilla
Should we even try to assess our students’ writing skills in a world of Gen AI?
Ben Knight
Head of Language Research and Pedagogy, at Oxford University Press
Ben started off his talk by presenting a couple of statistics:
- 95% of students report using AI tools in their writing
- 94% use AI for assessed assignments
He pointed out how writing has changed: as attention spans get shorter, so do text and the more likely they are to be accompanied by image and video.
There is nothing new in this: we have long been using machines and services to do tasks we don’t want to (wash up, cut grass) BUT why should students bother to learn how to do something if a machine can do it better?
One reason, Ben argues, is that writing is a gateway to success. To illustrate this, Ben showed a picture of All Souls College Oxford, apparently the most difficult college in the world to get into.
image: pexels_adam-lalam
If everyone is now using ChatGPT to write applications, the bar has been raised! If you have used ChatGPT, you can be sure thousands of others have too. But AI is generative and iterative, NOT creative.
Only a student who writes their own, original content will stand out. Writing allows us to express ourselves with human individuality.
Here are four key areas where students will have to demonstrate that individuality to succeed:
Writing involves skills that AI is not (yet) capable of: using higher order thinking skills and creative and original expression. Ben used the analogy of lifting weights. Think of writing as your mental gym. If you use a tool to help you, you will never develop your own ability. See the OUP Research Paper: The Human Connection: Motivation and Social Learning
What skills do students need to be good at for writing in the future?
Here, Ben highlighted four key skills:
- Draft: Generate, evaluate and organise thoughts
- Articulate: Use accurate and rich language
- Influence: Design for audience and impact
- Review: Assess, edit and improve
This is where we can put AI in the loop, to create ‘human-generated, AI-assisted’ texts
Upgrading
You can use AI to upgrade language, e.g. with a gap-fill activity: Ask AI to provide adjectives that mean ‘very bad’ and explain how they differ in meaning. Students have to choose the one that is most appropriate for their context. Only a person can justify why they have chosen that adjective.
Rewriting
Another example would be to rewrite texts with a specific audience and impact in mind, e.g. each learner (or group of learners) is given a ticket with a different audience (teacher / another student / the local mayor). The other students read the text and have to work out who the audience is.
Reviewing
Obviously, learners can use AI to correct mistakes. They then need to focus on the corrected version and noticing the difference (see below for an example)
The next step would be to take the corrected version and ask AI to improve it (see below) and ask learners to analyse the changes.
Evaluation of writing
The increased use of AI has increased the difficulties for teachers when evaluating student writing.
Teachers are increasingly turning to oral evaluation of students’ writing. If we know, and they know, that they haven’t written it themselves then one way of ensuring they learn something from the process is by asking them to summarise and explain what they’ve written.
This can be done one-to-one or in class in various formats, e.g.
- Q&A
- Pitch presentation
- Recorded presentation (video or message) to explain how & why it was done
- Group or pair presentations
Assessment of writing can be measured against the following criteria:
As a wrap up, Ben summarised the main points of his talk in the final slide (below).
Thanks to Ben for a clear and practical overview of some of the issues with learners' L2 writing in the age of AI.
Comments
Post a Comment