Pages

Wednesday, January 14, 2026

Chapter 4 - Guidance on Use of Multimodal Generative AI Tools to Enhance T&L.pdf

 https://go.gov.sg/ai-chapter4

GENERATIVE AI GUIDE CHAPTER 4

MULTIMODAL GENERATIVE AI TOOLS


#GEN-AI #IMAGETOOLS


In Secondary Science, students can

generate images of models that represent

and explain phenomena (e.g., cell, atom).

They can then evaluate these models for

accuracy.


In History, students can generate an artefact to

demonstrate their historical understanding of an event.

They can reference historical evidence and/or other

concepts (e.g., causation, significance), to present how

the output reflects their interpretation.

to enhance teaching & learning


Guide students in writing prompts, e.g., with prompt templates as an initial scaffold.

Encourage students to use precise subject-specific vocabulary in their prompts, e.g. art terminology, and

evaluate the output to improve the accuracy of terms used.

Have students document each iteration of their product and considerations for each refinement.

Tips for Teachers

Students can use GenAI tools to generate new ideas and translate them into visual products, e.g.

design sketches and 3D models. This helps students experiment early with alternative or

unconventional ideas, before further refining their solutions (Computational Thinking^).

In Art, students can apply design strategies like SCAMPER or shape borrowing techniques in

their prompts to help them refine their product.


Generative AI (GenAI) tools can generate images, audio and videos based on a user’s prompt by mimicking

the style and characteristics of image sets they were trained on. These GenAI tools can be used to develop

critical competencies (e.g., adaptive thinking) in students or for teachers to create customised resources.


Official (Closed) \ Non-Sensitive


Foster Adaptive Thinking

Illustrate abstract concepts using GenAI tools to help students make connections and draw new

insights from different contexts.

In Language lessons, teachers can guide students in comparing images created with and without descriptive

language prompts and discuss their impact on vivid imagery. Students can demonstrate adaptive thinking by

adjusting their writing according to the purpose, audience, context and culture of each image.

Develop Information Skills

Images, audio and videos convey information. By comparing AI-generated outputs with other

available sources, students can exercise discernment by evaluating the accuracy, credibility and

relevance of the information conveyed (Digital Information Management^).


Tips for Teachers


Promote Inventive Thinking

How can I use multimodal AI tools to develop Emerging

21st Century Competencies and Digital Competencies?


Read more about the Emerging 21CC here (via OPAL login).

^These refer to Digital Competencies. Read more here (via MOE Intranet).


Have students evaluate and discuss the output using relevant criteria, e.g. historical accuracy, relevance to topic,

time period or approaches like cross-referring to other sources.

Have students consider why the output generated differ to deepen their understanding of how accounts are

constructed, and how data in GenAI tools affect the generated image.


1


Correct as at 26 Nov 2025


How can I use GenAI tools to create multimodal resources?


Oral Picture Stimulus Composition Stimulus


Comic strips to make content more

accessible and engaging.


Images and videos to support lesson

storyline and gamification, e.g. story


background and avatars.

for resource creation


Student is an [Asian boy in school uniform

with white shirt and black shorts]. Blind man

is [in his 60s wearing a white shirt and jeans,

with sunglasses and holding a walking cane].

Create a [4 panel black and white comic

strip in cartoon style].

Panel 1 is [student walking home from

school].

Panel 2 is [student next to blind man at

traffic light].

Panel 3 is [student holding blind man's hand

to cross the road].

Panel 4 is [blind man saying thank you to

student].

*Image generated with Microsoft Copilot was

further edited for consistency using AI tools

(erase extra canes) and non-AI methods (copy

and paste, edit text).


Teachers can use GenAI tools to quickly create lesson resources including

customised diagrams, illustrations, podcasts, videos and other stimuli.


MULTIMODAL GENERATIVE AI TOOLS


Official (Closed) \ Non-Sensitive


Be clear and concise in your prompts to prevent

addition of unintended details in the image

generated.


Add descriptors, e.g., style or mood of artwork,

traits like physical appearance and emotional

expression to help the GenAI tool more

accurately generate your intended image.

Tips for Teachers


Use prompt libraries like Lexica, which

contains images with attached prompts.


Use a text-based generative AI tool to

craft prompts for images or videos and

finetune before using.


Creating stimuli to promote thinking and discussion


Due to the probabilistic nature of GenAI tools, use of the same prompts may not yield the exact

same images, audio or videos.


Increase student engagement through multimodal outputs


Teachers can further explore generative AI tools to generate videos or music.

For more examples of such AI tools, click here.


Example prompt template for images:

Create a [image type – photo, illustration, drawing] of a [person/object] [action] [place/background].


Create a [photo] of a [man] [sorting out rubbish

into three recycling bins] [in Singapore].

*Image generated by Canva


Podcasts to help students review

their lecture content and reinforce


learning.


2


Correct as at 26 Nov 2025


Possible example of citation:

OpenAI. (2023). DALL-E (Version 2) [Large

language model]. https://labs.openai.com

Prompt: XXX


Infringement of Copyright

AI image and video generators are

trained on available content on the

web, which may include original

works of artists without their

permission. It is important to respect

the creative rights of artists when

using and sharing the generated

images (e.g. uploading to SLS).


Content Inappropriateness

Tool may potentially generate image,

audio- or video-based outputs that

are inappropriate for students.


Abide by the age advisories of GenAI

tools if the tool is to be used by students.

Teachers can also try out the tool to see

if inappropriate outputs could be

generated, and switch to other tools if

needed.


Accuracy of content

Fact-check, source-check, proof-read, verify

the output (e.g., check the number of

limbs, if the context makes sense)


safeguards


MULTIMODAL GENERATIVE AI TOOLS

potential concerns


&


Did you know?

You can use Google tools like reverse

image search and about this image

to check if the generated image is

similar to an original work? If so, you

can change the prompt to regenerate

a different image.


Respect creative expression and

academic integrity by citing images,

audio and/or videos generated. While

this may not absolve one of copyright

infringement, it is a good practice.


Bias in Generated Outputs

Due to overrepresentation of particular

groups in training data, there is a risk

that AI-generated images may perpetuate

biases or stereotypes.


Incorporate modifiers in prompts, e.g.

specifying ethnicity, age and contextual

scenarios to counter potential biases

inherent in the training data.

Teachers can encourage students to discuss

how biases in AI-generated outputs may

mirror the stereotypes found in society and

the media.

Avoid using prompts which carry a high risk

of infringing intellectual property rights

(e.g. produce an image similar to [Artist X]'s

works, or incorporate [Brand X]’s logo)


Official (Closed) \ Non-Sensitive


For more information, please refer to the MMDI circular on the Rules for the use of Generative


AI Applications by Public Officers and IPOS Guidance on AI related Resources

No comments:

Post a Comment