Student-generated AI imagery: Enhancing geographic vocabulary acquisition through creative technology

VICTORIA MARSHALL, HARROW INTERNATIONAL SCHOOL HONG KONG, HONG KONG SAR, CHINA

Teachers face the persistent challenge of ensuring students master Tier 3 vocabulary while maintaining engagement in the subject and encouraging active learning to promote retention of understanding. The national curriculum (DfEDepartment for Education - a ministerial department responsible for children’s services and education in England, 2014) emphasises that teachers should develop vocabulary as an integral aspect of teaching every subject, with geography requiring particularly precise disciplinary terminology. Traditional vocabulary acquisition methods often fail to verify deep understanding or generate enthusiasm, while assessment frequently focuses on ‘higher-order’ geographical skills, assuming vocabulary mastery is already complete. I started the school year with these issues in mind and wondered if the influx of AI (artificial intelligence) education products could help to make vocabulary learning more creative.

Beck and McKeown (2007) point out that explanation of a word can result in static information but for students to truly engage with geography they must be able to understand and use specific vocabulary in order to become geographical thinkers (Draper and Bailey, 2022). Tier 3 vocabulary is that which is specific to a subject and less likely to be used in common conversation (Beck et al 2013). Physical geography topics have obvious Tier 3 vocabulary – the learning of a coastal process requires vocabulary such as attrition, slumping or concordant. However human geography is often taught with more generalised and Tier 1 vocabulary – city, housing, office blocks. This can mean that students don’t ‘speak like a geographer’ in the human topics and underestimate the importance of Tier 3 vocabulary in physical topics.

The research

At Harrow International School Hong Kong we focus on embedding seven essential skills through the curriculum regardless of the subject or topic being taught, based on research presented by the World Economic Forum’s future of jobs report (2023):

• collaborative problem solving
• creativity
• critical thinking
• cultural competency
• digital literacy
• effective communication
• leadership.

The metacognition of these skills is important and careful consideration of their integration into the curriculum is expected from all staff. I saw an opportunity to develop digital literacy while teaching a microunit on sustainable cities through the incorporation of Canva’s AI-powered image generator. I wanted students to use key vocabulary in a creative way without it taking up a huge amount of time and while keeping the focus on the vocabulary.

The AI activity was set in the third of three 55-minute lessons on sustainable cities for Year 9 (ages 13-14). The key outcomes were to introduce the Sustainable Development Goals (SDGs) and sustainable cities within the context of ‘thinking like a geographer’ and reminding students of the key social, economic and environmental themes. In the earlier lessons, students had completed a guided reading to build up ideas and vocabulary which would help them in the activity.

Students were given 20 minutes during the lesson. They were asked to write prompts to generate a series of images using AI depicting how they imagined various government buildings, residential buildings and office buildings might look in an eco-city. They were then instructed to: ‘Paste the images into the PowerPoint below. Include the prompt and a couple of sentences explaining what makes the buildings in your images sustainable. You will be marked on the geographical vocabulary you have used and your geographical understanding.’ Students were shown prompt examples and how to use Canva to generate images to scaffold the activity before they made a start. I ensured that I circulated at all times to observe and help with the process. I was also able to see their work developing in real time as it was written up within a PowerPoint in a Teams assignment.

There were four criteria when considering bringing AI into the classroom which guided the selection of Canva as the image generator, addressed below.

1. What are the pedagogical benefits?

The integration of AI image generation in this activity aligns with several important pedagogical principles that enhance vocabulary acquisition. First, it addresses the cognitive principle that students learn vocabulary best through active engagement with terms rather than passive reception (Nagy and Townsend, 2012). By requiring students to craft prompts with geographic terminology, the activity transforms vocabulary from abstract concepts into visual representations, creating stronger neural connections for retention.

From a constructivist perspective, this method enables students to build their own understanding of geographical concepts by experimenting with terminology in a low-stakes environment. Rather than simply memorising definitions, students must apply terms with sufficient precision to generate meaningful visual outputs. This creates what Ritchhart et al. (2011) call ‘visible thinking’, where students’ conceptual understanding becomes externalised through the AI-generated images and their subsequent analysis.

The activity also addresses the challenge identified by Brooks (2016) that geographic vocabulary instruction often lacks contextualisation. By enabling students to see how specialised terminology manifests in visual representations of sustainable urban environments, abstract concepts become concrete and relevant. This approach transforms what Graves (2006) calls ‘definitional knowledge’ (knowing what a word means) into ‘contextual knowledge’ (understanding how to use the word appropriately), which is essential for subject mastery in geography.

2. What practical and technical steps are required, e.g. log in access/cost?

Our school has a 1:1 laptop expectation. I trialled the use of AI with a Year 9 class who have been using laptops for their entire secondary school experience.

The majority of students were already familiar with using Canva, which reduced cognitive load, and already had access to all of the features via the Canva Educator scheme reducing login issues in the lesson. Furthermore, Canva had a limit on the AI scope. There are sufficient credits for students to play around with options and four images are generated each time to give the students a chance to critically analyse the images produced.

3. What potential challenges might there be?

There was a worry that the image generator would be too good and thus would produce results regardless of the quality of the terminology the student used. To overcome this the activity was marked using a rubric which focussed on two criteria: ‘use of geographical vocabulary’ and ‘explanation of sustainable design’. Students were required to share the prompt that they used alongside the image they selected from the four options Canva gives. They then needed to explain the sustainable elements within the image further bridging the focus of the assessment between definitional and contextual knowledge (Graves, 2006).

4. What ethical and safeguarding implications are there?

Canva’s AI image generator is not linked to any community creation meaning students are not exposed to the work of strangers thus ensuring safeguarding. Canva has put a lot of work into the safeguarding of their AI use since they have an education focus and ensure that their safeguards undertake the following: review the input, review the output and reduce bias (Canva, 2024).

Findings

The integration of AI dramatically streamlined the creative component of this task, allowing students to focus more deeply on applying geographical vocabulary rather than struggling with visual creation ­– all students had a relevant visual by the end of the lesson. By emphasising prompt construction over image production, the activity effectively centralised the importance of subject-specific terminology. While lacking a control group comparison, the assessment data still provided valuable insights into students’ ability to deploy key geographical terminology in meaningful contexts. The rubric assessment revealed that when students needed to craft effective AI prompts, they engaged more deliberately with Tier 3 vocabulary than in traditional assignments ­– the students achieved an average grade of 5.1/6. This is substantially higher than a usual piece of work for that group where the average sits at 4.2/6.

Real-time observations of engagement, time management and verbal feedback during completion were all positive with all students completing the task and peer discussion remaining focussed on the images.

One limitation was that this activity opened up the use of AI in the classroom. The students assumed that since I had introduced AI to them it was acceptable to use any AI and when they wanted. This required a reset and expectations were reestablished at a later point.

Overall it is clear that with AI integrated software there is potential to increase the creativity of students while maintaining a focus on vocabulary acquisition. Critical to this is building in effective assessment criteria which specify that students must include their prompts and the quality of these is part of the assessment. It is hard to predict the impact AI will have to the classroom, or indeed, the workplace these students will enter meaning it is important to give time to instruction of effective prompt writing which in turn gives more opportunity for subject-specific terminology to be harnessed. AI does not have to mark the end of academic rigour but teachers will need to adapt to the ways we can now set and assess work. Furthermore, AI does not need to be considered as a creativity swap but as an additional tool for teaching abstract concepts that benefit from visual representation.