Case Study: AI as a design/ research tool
Task
How can I use AI tools to help me in research and design within a problem area close to heart:
“Gender challenges within design”
The Problem
Having grown up around technology/ games and other spaces that have traditionally been more male dominated. I have always wondered why this is, and if my current career as a UX designer is influenced by these same causes. I have been curious to explore how this area has been not as accessible for women and if there are patterns in design that influence this. This question however is a very large one with potentially a lot of reasons. With AI becoming more and more helpful I wanted to see if it would be a good way for me to explore this.
Asking the AI a question
For this case study I decided to use two AI tools:
ChatGPT and DeepAI
The prompt I began with was:
“What are the current challenges in designing apps and software products for different genders”
AI Responses
Response Summary from ChatGPT
Response Summary from DeepAi
Critical Thinking and Fact checking
One of the things I have learned through exploring AI is that you should always fact check answers as well as look for your own research to back it up. In fact, ChatGPT even states not to trust it completely and it may create errors.
To compensate for this I used google to search for similar topics as well as use the sources linked within ChatGPT’s answer. (Unfortunately DeepAI does not reference where it gets its information from which is a bad point from me)
The answers turned out to be pretty relevant and accurate. A lot of it I had known from my own experience or research in the past but the articles linked were new to me and provided some interesting information. I believe the answers given to be pretty generic however.
One area that was particularly interesting to me was the lack of female non-binary and transgender specific research. It is clear that there needs to be more inclusivity targeting these groups as it makes up more than 50% of the population.
The AI did a really good job of showing how gender bias appears in tech whether it’s in the data we collect, how we think about inclusivity, or even the stereotypes that unitentially appear or are used in design. They dig into the big areas like how UX flows and also the smaller minute details including colour, AI voices and even the tone of text and speech. The AI also mentions areas like research gaps, privacy concerns and company culture.
However the AI response can sometimes simplify more important or complex issues. One area it lacks in is how all of the above interact with each other and the issues that arise from 2 or more of these things insersecting.
I also felt there was a lack of information on the progress that is being made.
Using AI to help me build a ‘hypothetical product’
From here I wanted to see if the AI could help me in creating a product. As I work with online meetings I decided to focus on this area.
From my own experience I know that meeting experiences should provide inclusivity in ways such as:
- Pronouns and custom names/ avatars
- Non discrimination of audio recognition from microphones
- Closed captioning for different languages
- Language change options
- Accessibility options
- Video recognition of all participants
Prompt response from ChatGPT
I wanted to see what the AI response would be and if it were similar to me so I prompted the AI:
“Id like to make sure online meeting experiences are gender neutral and provide the same experience for everyone. How would you recommend I do this?”
As can be seen from the answer to the right ChatGPT answered quite well. When I asked it to provide sources it replied with a list of very well credited websites and articles. It is quite impressive. Sources included places like:
- CareerFoundry, Interaction Design Foundation, Standard Beagle Studio
- Industry standards from Zoom, Prezi, ViewSonic
- Academic research from ACM Digital Library, DiGRAA, PMC (PubMed Central)
Using AI to Prototype
The next step for me was to see how AI could potentially help me in creating prototypes. I wanted to see if these tools would be able to take research to design in a way that transfers my learnings and helps in creating a beautiful product.
I decided to focus on one area which was to allow users to create and edit their pronouns, names and avatars within meetings.
I decided upon 3 prototyping tools:
Figma Make
UX Pilot
Uizard
The prompt that I gave the 3 tools was:
”Create a prototype flow for a video conferencing application used in large companies. This flow should focus on allowing users to change and edit their pronouns, names and avatar images. This should be inclusive and gender neutral and not show any bias.”
Figma Make creation.
Results
Figma Make
Please feel free to view the link above and full creation.
Figma make is perhaps my favourite design. The font, colours and ideas are all nice. I particularly like the preview area as it shows 3 different views of how you will appear to others to allow the user to tweak it to their liking.
However, this design to me is also a bit overwhelming. There is a lot happening on screen, a lot of text and some hierarchy concerns such as having the avatar section at the bottom. I do believe this would need testing and tweaking to make it usable.
UX Pilot creation
UX Pilot
This version is quite simple ‘does what it says on the tin’ It provides all basic functionality like allowing name change, avatar change and a simple preview.
However, it is very basic. It is also a little confusing as it allows you to pick avatars that ‘appear’ to be other humans, which to me is a very strange design decision and not one that would comply well with security.
To me this design is the worst out of the 3. It is incomplete and I do not believe I would recommend this tool. The designs are extremely basic and unfinished.
Conclusions
It is very clear to me that using AI for design and research can be extremely helpful.
From researching topics It has provided me with concise, accurate information with appropriate linked articles (ChatGTP) that make it easy to store and organise information on topics. It helps to back up prior knowledge but also link to things that you may not be aware of. However it does need to be fact checked. This is mentioned by ChatGPT and it is clear it can get confused if you over prompt it. I did not really like DeepAI as it did not have as many disclaimers as ChatGPT or provide sources. The way it organised answers was also not as neat in my opinion.
For Prototyping I believe Figma Make to be the best option. The AI is clever, it provides further steps, a list of the code but also the reasoning behind why it creates the prototype the way it does. This is very important in my opinion as you can see what it is ‘thinking’. I believe UX pilot is okay, it creates simple designs but does not think ‘outside the box’. I would not recommend Uizard. It provided incomplete designs with missing areas. It seems to be behind the other 2.
Final Thoughts
I do believe that I will be incorporating AI into my designs more often, especially for research reasons. Having the sources of information available to me quickly is extremely helpful in my busy work life. However I will be fact checking and tweaking any information gathered from it as it is not 100% accurate at times.
I am very excited to see how AI will continue to grow within this industry 🤖