Reinfo

The “Reinfo” app is an artificial intelligence (AI) tool that will scrape data from across the web to help people discern which information is factual and what is not. Use “Reinfo” to validate claims made on platforms such as news articles and Instagram posts.

Industry

Social Media, AI

Role

UX Research
UI Design

Themes

Combating
Misinformation

Tools

Figma

Duration

Aug-Nov 2023

Overview

It is difficult to validate much of the information one finds on social media and online news outlets today. Over the years, I have become more and more concerned about the frequency of what appears to be misinformation on social media. Therefore, I decided to conduct some research on this problem and imagine a potential solution.

Research

In order to explore the topic of misinformation on the web further, I read through various academic papers that have analyzed news as it appears on social media and other online outlets.

Some highlights:

In a study by Walker and Matsa, 48% of U.S. adults use social media as their primary news source

In a study by Vosoughi et al, 70% of Twitter users reported not being able to distinguish real from unverified news

In a study by the FTC, 770M were reported lost by 95K people due to social media fraud

Problem Statement

Internet users often fall victim to misinformation because a significant portion of the information found online is sensationalized to grab attention, not entirely factual, and/or difficult to validate.

How might we

provide internet users with the necessary

context needed to discern which information is trustworthy and which is not?

User Persona: Susceptible Sarah

I supplemented my research by interviewing 10 people to learn more about their experience with filtering through misinformation to find trusted sources of news and current events online.

Susceptible Sarah is an adult in her late 20's to early 30's, leading a career that she is engaged and satisfied with. She leads a fairly balanced life, mixing recreation with work, and wouldn't consider herself an expert in popular social topics of the day. She surfs the internet semi-regularly, but feels she increasingly becomes more skeptical of the information she reads online.

Demography

Age: 30

Education: Bachelor's Degree
Occupation: Regular office job

Frustrations

Cognizant of the existence of biased/untrustworthy information online
While concerned about misinformation online, not willing to invest time to fact-check

Motivations

Pass time with quick entertainment
Stay up to date with current events
Quickly research topics of interest

Behaviours

In spare time, browses social media outlets at an average of ~2 hours daily
Loosely follows the popular social and economic topics of the day, eg. gender equality, global conflicts

"Social media algorithms prefer images and content that are extreme or catchy, so naturally you start to get misinformation."

User Journey

Susceptible Sarah's user journey starts with her browsing through social media, finding some piece of information that seems reasonable, acting on that information, then realizing it was not correct. Please note that the "secret weight loss method" mentioned in the user journey is for example purposes, and other cases of acting on misinformation were mentioned during interviews.

Open social media, Youtube, other information outlets on the web

Consume news, friends’ updates, other information online

Find intriguing information on a topic of interest eg. “a fast, secret weight loss method”

Try the “secret method” without verifying its validity
See no results

Discover additional resources, sometimes behind paywall
Converse with friends

Friends believe the “secret method” is false
Become skeptical of sensationalized information

Give Sarah the tools to quickly fact-check intriguing information at the time of discovery.

Value Driver:

Design Solution

User Problems

It is difficult for a non expert to tell the difference between trust-worthy and false information online

Users do not want to spend time validating the accuracy of every questionable content found online

Design Solutions

Collect as much relevant context on key claims made in online written and filmed content as possible

Bring the relevant context to the user at the moments when they mentally process the information they find online

Task Flow

The solution I chose will allow users to read through online articles or social media content, and overlay the original content with relevant context that allows them to fact check the information they consume when they need to. On the backend, an AI system will scrape the web to look for trustworthy context to validate the information users consume.

Low Fidelity Prototype

In my low-fidelity prototype, users are first led to a homepage that leads users to a combination of news outlets, social media platforms, and articles on topics of common interest. Users would then leverage features that pull in relevant contextual information for the news and social media content they view through the prototype.

1. Simple description of the app to users

4. Add many popular topic in homepage, so users can easily find the sources when they visit the app

2. Information related to their search was automatically generated by search engines

5. After search the information provide users with verified articles

3. API social media platform so users can question any sources

6. Users had access to the complete source of validated data

User Testing

After refining the low-fidelity prototype further to add content copy and functionality, I enlisted 5 users from the initial user interviews to test the prototype. There were 2 key findings that I iterated my design on.

Before

Iteration Process

1. Users were initially confused about the intent of the button they would press to bring in AI-validated fact-checking information.

I made the analysis icon more prominent and moved it to a more noticeable location at the top of the content page.

2. Users said it was difficult to differentiate between the original content and fact-checking information. Given the intent of my design is to show users the ability to differentiate between misinformation and factual information, I color-coded the text to show misinformation in red and factual information in green.

After

Final Prototype Reinfo

You can explore the app here!

My final prototype, Reinfo, is a mobile app that allows users to browse through news and social media content from one convenient platform, and validate which content they see is factual or not with the click of a button.

Color Palette

With Reinfo's purpose being to help internet users differentiate misinformation from trustworthy information, I went with a color palette that leverages green and red, two colors that are commonly associated with states of being "correct" and "incorrect".

The distinct hues were used for each color to promote accessibility across a wide variety of users.

Challenges I Faced...

With Reinfo being an AI- powered app, I wanted to make sure the fact-checking experience was not unrealistic relative to the capabilities of today's AI systems. My task flow, as a result, deviates from the standard archetype to include a high level backend logic that Reinfo's AI would follow. For those from a technical background, please feel free to reach out with ways to refine my fact-checking logic.

Additionally, even with an AI system searching the web for fact-checking context, it should be noted that in many cases, information cannot be categorized as "correct" and "incorrect" in a black and white fashion. I have added various examples of this into the final prototype's UX to depict this limitation.

Key Learnings & Next Step

I found that, it designing for an app that more heavily leverages AI, I made a larger effort to design with technical feasibility concerns in mind. Reinfo also incorporates many capabilities, which has allowed me to develop a wider array of design skills.

I would like to add validation of content and news updated in real-time, as users often make mental decisions on the validity of online content at first exposure. Another capability that would be helpful

to add is to allow users to contribute relevant context directly as well, similar to Twitter's Community Notes feature. A key concern with this type of feature would be that additional validation will be needed to filter out unhelpful submissions.