User research findings
Hannah Wallwork
February 2025
Young people & Generative AI
About this research
How might we support young people to understand, examine and positively influence how they use and engage with Generative AI, becoming thoughtful makers and critical users of this technology?
We are working with 3 youth sector organisations, �Beats Bus, Politics Project & Warrington Youth Zone, �over a 4 month period to produce a proof-of-concept to answer:
Methodology
About this project
About this stage of research
We showed 13 young people aged between 15 - 17 years old 2 early concepts in individual interviews.
If you would like to:
Please email me hannah@neontribe.co.uk for permission
Concept 1
Mystery story
This was an interactive choice based mystery story where young people discover a rumour being spread about 2 teachers having an affair.
They need to make decisions about what to do next, whether that’s deciding whether to believe the rumour or asking for proof. There are a number of other decisions they make that affect the storyline.
At the end of the story they learn that a deep fake AI image was created by another student as revenge. They are signposted to BBC bitesize.
Think AI is an image and text generator that encourages young people to think about the potential issues behind the image/text they have generated.
We tested one version created by our designer testing the core concept. We also had a technical prototype actually connected an AI system, however without the design considerations of the first prototype.
Concept 2
Think AI
Finally we ended by asking young people’s opinions on what direction they felt we should take with the branding.
Final activity
Branding
Key findings
Think AI was the winning idea
8/13 Young people preferred Think AI because they found it:
Those who preferred the mystery story found it:
“Game is one and done can't see myself using it again"
“Would say to the teacher: Can I just read a book, rather than clicking through a whole thing"
"More impactful, more relatable”
“I prefer the think one. It's more personalised to you.”
Prefer [think ai] because the other is a planned story whereas this one YP get more input into it themselves
Young people are knowledgeable on many, but not all, of the risks.
Whilst there was some new information, many were aware of many of the risks mentioned on ThinkAI. Some also felt there wasn't enough educational information within or at the end of the mystery story about deep fakes
This research suggests young people need to be introduced to emerging or overlooked risks that are not widely discussed among young people, or to be introduced to a new perspective on an issue they are already familiar with.
It appears bias in datasets, environmental issues and tools that can be used to identify deep fakes may be areas that are less known by young people.
How might we…
make sure young people are learning something new about generative AI?
Sees the environmental risks that they hadn't thought about but thinks is very important.
"i think i've heard those before"
Quotes & observations
“I hadn't considered the 1st one and haven't heard it discussed as much”
Being able to relate to the risks of AI is important to young people.
Many preferred the relatability of the mystery story in comparison to the ThinkAI answers. Many suggested showing how it could directly affect young people, for instance one of their friends.
Research suggests that this may help them better understand and relate to the real-world consequences of generative AI.
How might we…
make sure that young people are able to relate the risks of generative AI?
"Soo relatable"
"Something similar happened at my high school"
Quotes & observations
“But maybe the plot should be only about students to make students think - i don't want to be in that situation”
Says Think AI should make people interpret with real life example
It may not be clear how to facilitate this in a group setting.
Whilst both tools are designed to be used in group settings with a facilitator, we tested them in individual calls and therefore weren't able to test how it might be run as a group. This is something we will need to explore further in our next set of research.
We had questions about how this may work in a group setting, with some feeling the facilitation side may not be clear.
In addition, young people didn't spend time reflecting on the potential issues on the 'BUT' page, instead clicking straight through to the answers. We could be missing an opportunity to help group facilitators engage young people in critical discussion
How might we…
make it clear for facilitators to know how to use this solution in a group setting?
Quotes & observations
How might we…
encourage a group of young people to spend more time thinking critically about the issues, before finding out the answer?
suggests making Think AI more "group friendly"
Says mystery story involves "more conscious thought" as need to convince people why you'd choose your answer
It needs to be clear if there are next steps are for young people.
Many young people either didn't notice or weren't interested in the BBC bitesize article at the end of mystery story.
After the ThinkAI risks & issues page, young people have a choice to generate something new. However, it wasn’t always clear what to do next, or what to take away from the experience. Some also felt like it ended on a negative note by focusing on the issues.
Considering this is designed to be used in a group setting we may want to consider how those next steps could involve a facilitator and what we want or they may want young people to take away from the experience.
How might we…
help young people & their facilitators know what to what to do next, whether that’s applying the knowledge, reflecting, or continuing the learning?
Quotes & observations
“No, I don't use BBC Bitesize and read long pages about AI, as I am already quite aware about it”
Says it might turn people away from AI so they might not want to use it again
“Reading text is not interesting, I'd just skim and click again”
There is a balance between creating an element of surprise versus being transparent about the purpose
The purpose of both ThinkAI and the mystery story as being an educational tool around AI was not obvious at the beginning, but instead revealed itself part way through.
Whilst this 'element of surprise' was sometimes well received, there were criticisms that the purpose should be clearer so that:
How might we…
help young people & their facilitators understand the purpose & benefits of the tool, whilst not giving away every detail?
Quotes & observations
"The story was more interesting to me, as it was hidden"
Liked Think AI more even though it was more “deceiving”
We need to think carefully about how we name any solutions and the language we use around it.
Using words that people associated with communication like 'whisper' 'secrets' & 'rumours' led young people to think they might be speaking to a service. 'ThinkAI' made young people assume the AI would do the thinking or analysing for them, which confused them.
Gov.uk have guidance on how to name services, including using verbs rather than nouns.
How might we…
ensure the naming of solutions is clear and doesn't mislead expectations?
Quotes & observations
All the words you can pick out say it's going to be a chat with a teacher, or with a friend, or someone from CAMHS
what is it for: "something like chatgpt for generating stuff for you" because name is
"think-ai"
Young people are quick to point out when a solution, or part of a solution, feels childish.
Many young people felt the mystery story was aimed at people younger than them. They pointed out the concept of an interactive story; the storyline of teachers having an affair and type of information provided was simple and perhaps juvenile at points.
On the flip side, other young people felt that it might not be aimed at the youngest children because they saw some of the language as more complicated.
Some felt Think AI was aimed at younger school children because of the graphic elements, like the star and the eyes.
How might we…
ensure any solutions feel age appropriate for 15-17 year olds?
Quotes & observations
"I think I take back what I said about college students, they might find it a bit juvenile"
"it looks like its' for younger people" "googly eyes" "it's got that diamond"
There are individual differences in how young people like to consume information.
Visuals are important to young people. They help them engage with and process information, however they can be confusing is they don't appear relevant or have accessibility issues.
Whilst some like to read detail, others don't and may completely skip over text, for example the speech bubbles on the Think AI tool. In addition some found the language confusing on the mystery story.
We need to think about how we can cater to individual differences, prioritising accessibility and those that struggle with reading, whilst also engaging with young people who have an appetitive for more detailed information.
How might we…
balance individual differences in how young people prefer to process information?
Quotes & observations
“I'd made it more fun, reading text is not interesting”
“”people cant be arsed to read"
Now that the young people have chosen Think AI as the tool to take forwards…
We have questions to consider with design & development:
Social cohesion is an important part of this project. How might we ensure the solution fosters social cohesion in 15-17 year olds? NCS has a specific definition of social cohesion:
User need statements �& user stories
What are user needs & user stories?
User stories are short, simple descriptions of a feature or requirement from the user's perspective.
User need statements articulate what the user needs rather than what the product should do. They are concise, human-centered, and framed to avoid prescribing solutions.
Both tools help:
User needs & stories
As a 15-17 year old, who is already knowledgeable about some of the risks of generative AI, I need to be introduced to emerging or overlooked risks that are not widely discussed among young people, so that I can deepen my understanding around generative AI. |
As a 15-17 year old, I need to learn about AI risks in a way that highlights their impact on people like me or those close to me, so that I can better understand and relate to their real-world consequences. |
As a 15-17 year old, after finishing an educational experience, I need clear guidance on what to do next, whether that’s applying my knowledge, reflecting, or continuing my learning, so that I can make the most of what I’ve learned. |
As a 15-17 year old, I need information to be communicated clearly and concisely in a way that is appropriate for my age, so that I can understand and engage with it. |
As a 15-17 year old, I need guidance & examples when crafting image or text prompts, so that I can feel confident exploring generative AI. |
As an adult facilitator, I need guidance on how to use an educational AI tool effectively, so that I can confidently facilitate engaging group activities with 15-17 years |
As a young person, I need a clear understanding of the purpose and benefits of an educational tool, so that I can decide if it's relevant and valuable for me to spend time on. |