This article gives an overview of how a Chat bot works and how to write effective prompts to improve the likelihood of a useful output.
Copilot Chat and other similar chat based generative AI tools have been sometimes described as ‘an over enthusiastic intern with unlimited energy’. You might identify with this description and have countless experiences of opening a Copilot chat, starting a conversation and then finding yourself pretty quickly with a lot of text which is either too generic, useless or even plainly incorrect.
But is it just the bot’s fault or do I have a part in this? Well, a bit of both. The chat-based interaction plays on a model that is familiar to us when interacting through text with other humans, like texting on a mobile phone. Therefore, it is natural for us to assume that an interaction with an AI chat bot should function in the same way. But in reality, this is not how a large language model (LLM) works. At its core, a LLM is an incredibly advanced “next-word predictor.” It works by reading billions of pages of text to learn patterns, then uses math to guess the most logical next word in a sentence, repeating this process over and over. It does not understand what we mean, so we need to be very specific with our prompts.
Anatomy of a good prompt
There are various frameworks that you can use to write a good prompt. Since at UAL we are using Copilot Chat, we will look at Microsoft framework which has four parts:
- Goal: What do you want from Copilot?
- Context: Why do you need it and who is involved?
- Expectations: How should Copilot respond to best fulfil your request?
- Source: What information samples do you want Copilot to use?
The information and details you provide using these four parts will help Copilot to create a response that is likely to be more useful and accurate because you are defining the context and the limits of its response.
Context / Teaching Scenario
You might want to use Copilot Chat as a form of inspiration to create seminars, lectures or individual activities. This is an example of prompt that uses the Microsoft framework:
Goal: Create a 30-minute seminar activity.
Source: Use the attached Scheme of Work and Sawyer’s literature review on art and design pedagogy.
Expectations: Give me a plan with timings, 3 discussion questions, and 1 follow-up task. Keep it student-centred, practical, and lightly scaffolded.
Context: I teach BA Fashion Photography students on LCF’s Capturing Fashion unit. This is a level 4 unit (Year 1). Students are developing early ideas linked to inclusivity, representation, the gaze, and decolonising fashion photography. The activity should support research journal work and concept development.
This is the output of the prompt:
30-Minute Seminar Activity
Capturing Fashion: Whose Gaze? Whose Story?
Learning Aim
Students will interrogate early project ideas through the lenses of inclusivity, representation, and the gaze, using peer dialogue to strengthen concept development and research journal work.
0–5 mins | Individual Reflection: “Locate Your Idea”
Ask students to open their research journals and write brief responses to:
- What or who am I interested in photographing?
- Whose perspective is currently shaping this idea?
- What assumptions might I be making about my subject?
Output: Three bullet-point notes in their journal.
Tutor role: Minimal input. Encourage students to work with ideas they are already developing.
5–15 mins | Small-Group Idea Exchange
Students work in groups of three.
Each student has 2 minutes to explain their emerging concept while peers listen and make notes.
After each presentation, peers offer:
- One aspect that feels inclusive or representative.
- One question about power, authorship, or gaze.
- One artist, image, community, text, or reference worth researching further.
Output: Students record peer feedback directly in their journals.
Tutor role: Circulate and prompt deeper thinking rather than provide answers.
15–25 mins | Critical Discussion
Bring the class together for a facilitated discussion using the following questions:
Discussion Question 1
Who is being centred in your proposed images, and who might still be absent or overlooked?
Discussion Question 2
How could your photographic approach challenge rather than reproduce dominant ideas about fashion, beauty, gender, race, class, or identity?
Discussion Question 3
What practical decisions (casting, styling, collaboration, location, image selection, captioning) might help decolonise or diversify your project?
Encourage students to connect comments back to their own developing concepts.
25–30 mins | Journal Action Plan
Students individually write:
- One insight gained from discussion.
- One research gap they need to address.
- One concrete action they will take before the next session.
Examples:
- Research a photographer whose practice challenges conventional fashion imagery.
- Reconsider casting choices.
- Explore collaborative image-making methods.
- Investigate representation of a specific community.
If the original prompt had been something vague like:
“Create a seminar activity for fashion photography students.”
Then Copilot would have much less context about level, curriculum, themes, duration, pedagogical approach, assessment context, or desired outputs.
A typical response might look like this:
Fashion Photography Seminar Activity
Objective
Help students think about fashion photography concepts and share ideas with peers.
Activity (30 minutes)
Introduction (5 minutes)
Ask students to think about what fashion photography means to them.
Group Discussion (10 minutes)
In small groups, students discuss:
- What makes a successful fashion photograph?
- Who are your favourite fashion photographers?
- What themes interest you in fashion photography?
Idea Development (10 minutes)
Students sketch or write down an idea for a fashion photography project.
Share Back (5 minutes)
Students briefly present their ideas to the class.
Why this response is weaker
The activity isn’t necessarily wrong, but the prompt provides insufficient information, so the output lacks:
- Level 4 appropriateness.
- Connection to inclusivity, representation, gaze, or decolonisation.
- Alignment with the Capturing Fashion unit.
- Support for research journal development.
- Any reference to art and design pedagogy.
- Specific learning outcomes.
- Meaningful scaffolding.
- Criticality and disciplinary language.
The difference is useful to show because it demonstrates a key principle of AI prompting: the quality of the output is often proportional to the quality of the context provided.
The original prompt supplied:
- Programme (BA Fashion Photography).
- Unit (Capturing Fashion).
- Study level (Level 4 / Year 1).
- Topic focus.
- Pedagogical purpose.
- Activity duration.
- Desired format.
- Source materials.
Those details enabled a much more targeted and educationally robust response than a generic request could produce.
Author details:
Name: Giacomo Occhipinti
Contact: g.occhipinti@fashion.arts.ac.uk
Leave a Reply