Revealing AI usage and content to users
Project Overview
Music Transparency focuses on revealing how AI music is being implemented into users playlists and songs without users being aware, and aims to bring awareness to this by adding clarifying features to aid in the recognition of AI music and giving users the option to learn more about how its used, while giving users more control over AI's presence.
Ever listened to a song and really enjoyed it? You go to find the artist behind it and after several minutes of confused looking, you realize the artist generates their music through AI. Instant delete. So many people have become more and more annoyed at the inclusion of AI music in their recommended and playlists and frustrated at the lack of regulation of it. With this, I plan to tackle three issues:
The
Solution
To ensure AI usage is apparent to users, I focused on using YouTube Music as a platform to implement features to ensure users are able to recognize AI songs, learn more about them, and have more control in blocking AI music in their feed and recommended.
1
Users not wanting to see or hear AI music, having to take extra steps to ensure their music isn’t AI.
2
Ethical concerns with AI steer users away from platforms with no regulation.
3
Users being forced to see AI in their recommended when they don’t want to.
4
Copy writing issues and the use of Artists likeness without consent in songs.
Why is there More AI Music?
Ease of Use
Generating AI music is as simple as pressing a couple of buttons, such as choosing the genre, mood, pace, vibe, and energy, and users will be able to generate hundreds of songs to their liking with little to no effort.
It's Fun!
And with AI music being so easy to generate, many also just do it for the fun of it. With it being easy to pick up and use, there are plenty of AI generated music out there made for fun.
Defining the User
When beginning my research phase, pivoted focus to the target audience to be those of younger age since many young adults listen to music often throughout their day. Some other factors when establishing a target audience was:
Teen to young adult, about 35 years being the max age.
Highschool, college, or Grad students.
Listened to music constantly and/or uploaded their own music to platforms in the past year.
Current Competitors
After researching and understanding the issues with AI music as well as understanding the rise of it in the first place, I found two other music platforms that have attempted or implemented systems in place regarding AI music and it's usage alongside YouTube, the two being Deezer, and Spotify. With these platforms, three major questions needed to be answered:
1
Implementation
How did they tackle AI music transparency? Was it obvious what was AI and what wasn't?
2
Acts of Control
Do the users have control over what AI music is being shown to them? Can they limit the presence of AI music?
3
Ease
Are there ways users can easily access additional information regarding AI and it's usage and application on music platforms?
After establishing some of the main questions and goals needed, research on the different companies began, understanding users frustrations with them and what they do well when it comes to AI music regulation. Below is information of the different companies.
Deezer
Pain Points
No control
Can't turn off AI music from showing up in recommended.
No opting out
No way for musicians to stop their music from being used or combed.
No obvious section for users to learn more about the use and regulation of AI.
Thoughts
Deezer, implemented a tag on each song that portrays AI in it, which makes it easier for a user to either avoid or listen to it if the wish. The issue is that it doesn't do anything else besides alerting users that a song is AI, may be better to have it link to somewhere for users to learn more about it. Maybe understanding the terms and conditions and AI regulation policies.
Youtube Music
Pain Points
Weak enforcement
Though required, songs AI generated aren't being tagged.
No opting out
Isn't anyway for the user to opt-out of AI music in their feeds.
Impersonations
Deepfakes of musicians voices are prevalent and not addressed as so.
Thoughts
Youtube Music, they have systems put in place for AI music regulation, but it's rarely enforced, leading to many songs getting away with being AI without users noticing. There needs to be an obvious way to notify users if a song is AI. Also, musicians need a way to know if their likeness is being used without their permission as well.
Spotify
Pain Points
AI or not?
Hard for user to see if a song is actually AI or not.
Inconsistent
Individual tags aren't present for songs but are for playlists.
No opting out
No way to opt-out of AI music showing up in feeds.
Thoughts
Youtube Music, they have systems put in place for AI music regulation, but it's rarely enforced, leading to many songs getting away with being AI without users noticing. There needs to be an obvious way to notify users if a song is AI. Also, musicians need a way to know if their likeness is being used without their permission as well.
Key Insights
After understanding users perspectives and issues personally viewed in regards to AI transparency, I found three main patterns/pain points users were having:
1
No Blocking AI
No easy way to opt-out of AI music from showing on users feeds.
2
No AI Transparency
No easy way to see if a song is AI generated or not, making it tedious at times to find real songs.
3
Info Restrictions
Lack of easily accessible information regarding AI and how it's being used and regulated on the platform.
Tackling Interviews
After understanding some of the context of the rise of AI, it's lack of regulation, and how other music platforms were handling the issue, there was enough of an understanding to move on to interviewing the users. The main points I wanted answer from each user was:
1
Opinions
How do they feel about AI generated music?
2
Frustrations
Have they encountered any annoyances or issues with AI music?
3
Suggestions
How would they regulate or deal with AI music? How do they deal with it now?
4
Its Use
Do users believe AI music is doing more harm than good?
Interview Findings
From the interviews, the results showed that:
80%
Of interviewees view AI negatively, mentioning how its hurting more than its helping.
100%
Mentioned wanting to have options to decline their music being used in AI.
100%
Of interviewees wished to know how their music is being used and why.
100%
Of interviewees believe AI needs to be regulated, mentioning how many need copywriting.
Comments from users that stood out mentioned some key issues with AI:
"I feel like AI music has no soul whenever I hear it in a playlist… Its really dumb that there isn't an easy way to know if a songs AI before listening to it…"
"The fact that AI isn't regulated enough is… annoying cause they keep producing tens of hundreds of songs and because I clicked one by accident, I'm getting recommended more…"
"Wish AI music wasn't even on platforms, but if I had to regulate it I'd at least want a way
to turn it off…"
Key Insights
Interviewees general idea of AI music was widely negative, mentioning how it's hurting actual musicians and creating soulless music that wasn't needed. From the results and additional conversations with the interviewees, it became apparent that there was three main issues that users wanted addressed:
1
No AI in Recommended
A way to opt-out of AI generated contented on their feed.
2
Understanding AI Use
A better understanding of why AI is being used on the platform in the first place.
3
Transparency
A way to see what song is AI and
what isn't.
Personas
After getting feedback and questioning users of their takes and opinions with AI music, I moved to creating two personas, one for the group I interviewed that actively listens to music, and the other, the group that creates music.
Tobias Dones
Age: 22
Family: Single
Occupation: Designer
Gender: Male
Located: San Marcos
Education: Bachelors
Bio
Tobias is a Graphic Design student at a university. They enjoy playing guitar and posting their music online and attending music concerts on the weekends.
Goals
Obtain a bachelors
Improve music career
Work with different musicians
Frustrations
AI undermining music language and soul behind it
Reaching a wider audience is difficult
Unaware what music is and isn't AI generated
Personality
Chill
Introverted
Kind
Comical
Quote
"It's sad that people replaced something that comes from the soul…with AI. Their music is lifeless…"
In App Suggestions:
A way to turn off AI recommendations in their feed.
Bethany Jones
Age: 23
Family: Single
Occupation: Student Worker
Gender: Female
Located: San Marcos
Education: Bachelors
Bio
Bethany is currently a University student learning to become a Graphic Designer. She enjoys going out with her friends and being able to enjoy life at a leisure.
Goals
Graduating ang getting an apartment and job.
Learning from peers to better design skills.
Travel
Frustrations
Finds a nice song only for it to be AI then blocks them.
Annoyed with the AI in her recommended.
Personality
Extroverted
Caring
"Winging-it"
Comical
Quote
"Having AI in my music is so annoying, because its people that do nothing and just prompt to get a result…"
In App Suggestions:
Wants to know what songs are AI clearly in her feed.
Once a solid foundation of user pain points and issues from many users was established, then began the iteration stage of ideas and concepts, focusing on what problems to solve and how to tackle them.
Aid users in controlling AI content shown to them?
A toggle or button that users can use to switch between showing AI content or not in their feed or settings.
An option in settings that turns off and on when wanted.
In playlists, having a toggle that excludes AI music so you can download it without AI music.
Help users be aware of changes made to the app regarding AI?
Tags on songs saying "AI generated"
An icon that communicates its AI.
An overlay popup that mentions how some songs are AI generated. Could be within a playlist specifically.
Inform users of changes made to the app regarding AI?
A short tutorial or showcase of what's new when users log on for the first time after the update.
Highlighting the different changes briefly when they log in before fading away.
Exclamation marks on added features that go away after being clicked or when the user closes the app.
Inform users of AI usage on Youtube Music?
Having it prominently stated in settings.
User Flow
After establishing the features to put into Youtube Music, two different user flows were created, one for when the user wants to learn more about AI usage through the songs, and one for when the user wants to turn off AI recommended music.
Turning off AI Recommendations
Recognizing and Learning about AI
Designing an interface that encourages AI transparency meant abiding to YouTubes design guidelines, or at least doing the best that I can. Fortunately, someone designed a YouTube template shown below that could be worked on.
Though the template was useful, it was outdated, so the first step was to update the YouTube template and make it as identical to the current as possible, since YouTube doesn't have an open library of assets and elements to use.
Provided Template
Updated Ver.
To communicate to users that songs or playlists are AI generated, YouTube Music's specific AI icon was used to inform users. The icon is only slightly revamped to be slightly more noticeable and present in the designs.
YouTubes AI Icon
AI Indicator
With the icon, I went through and added the AI icon to all places where users would see and interact with songs, such as putting them on thumbnails songs, active playing songs, and descriptions of them as well.
Once the flows were established, I moved on to user testing among six individuals that listen to music constantly on YouTube as well as some that don't to get an idea if new users would be able to understand and navigate the interface.
"Navigating AI Details" Observations
75%
Of users initially scrolled before coming back up to see the AI icon.
50%
Of users tried tapping the “related” tab to see the AI details instead of the three dots.
50%
Of users tried tapping the “related” tab to see the AI details instead of the three dots.
"Turning AI Music Off" Observations
100%
Of users went to settings instead of channel to turn off AI Recommend.
50%
Of users scrolled by turn off button multiple times before seeing it.
From these findings of what worked and what didn't, I then tweaked and refined the flows and accounted for the multiple ways users tackle the flow at hand, results which you can see below in the final outcome.
Lets do a quick recap of how we got here:
01. The Goal
Improving how users deal with AI music on their music platforms.
02. Pain Points
Users hated lack of control with AI music. Not being able to opt-out of AI music or learn about how their platform uses AI.
03. Solution
Adding features to Youtube Music that enables users to recognize what songs are AI, learn more about the AI policy, and the option to opt-out of AI recommended music.
AI Info
This flow showcases how AI songs are identified as well as an option for users to learn more about Youtube Music's AI policy so they can be more informed.
AI Recommended
Showcases a feature allowing users to turn off AI recommended music from showing up in their feed or playlists in the future,
Reflections and Next Steps
This was one of the few case studies that was fun and chaotic to work on since I had a deadline of two weeks to complete the project. Learning about everyone's perspective and opinions on AI has been honestly hilarious seeing how everyone I've talked to hates it but eye opening seeing how others see it's use in helping beginner artists brainstorm ideas.
As fun and chaotic as this project was, there are still areas I know I can improve in and even add to this case study, such as:
1
Interviewing more musicians to get an even better understanding of artists views on AI.
2
Creating a sort of tutorial screen for users when the first open the app to show them what's new.
3
Providing more info on what specifically is AI in a song. The whole thing? One instrument? The voice itself?
4
Giving additional features to artists that upload songs to music platforms so that they have more say in AI usage.
