Media Literacy Tools for the AI Age — Research collaboration with TrueMedia.org
TL;DR
Role
Sole Designer (Graduate Capstone)
Team
Solo (post-February 2024) · Research collaboration with TrueMedia.org
Timeline
7 months (Jan–July 2024)
Impact
~1,800 testers at Misinfo Day | 2,000-respondent survey | Concept validation
Key Skills
User research · Concept design · Gamification · Platform strategy · Academic rigor
01 — Snapshot
Platforms try to control misinformation, but users don't trust them. This capstone asked a different question: what if we equipped people to identify it themselves?
Agency over authority. People resist being told what's true, but respond better when given tools to decide for themselves. Not another fact-check label — tools that help people pause, verify, and learn, folded into daily habits.
"She took us along on her journey of learning from users as she went."
— capstone reviewer
My role. Sole designer on a UW graduate capstone (Jan–July 2024), started as a team of four. After February, the product, design, and direction were entirely mine — research, archetypes, all four features, and testing with ~1,800 people at Misinfo Day.
From March to May, I collaborated with TrueMedia.org, a deepfake-detection nonprofit, on research and market assessment. They analyzed 60,000+ pieces of media and shut down in January 2025 — not from bad design, but because nobody profits from deepfake detection. That's this case study's throughline: design alone can't make a public good sustainable (section 06).
Search & Image Search
Verify links, headlines, and images with layered credibility ratings
Scan
Camera-based verification for printed content — flyers, newspapers, ads
Literacy Quiz
A level-based game teaching users to spot manipulated content
Ask Us
Human-backed fact-checking for gray-area content algorithms miss
2,000
survey respondents
28
interview participants, two phases
14
moderated usability participants
~1,800
tested live at Misinfo Day
02 — The Problem
After AI, the same problem arrived at a different scale
Misinformation exploits human bias: believe what confirms your views, trust what feels familiar, share before verifying. After AI, the same problem arrived at a different scale — more believable, more volume, same fragile ecosystem.
Survey Results (n=2,000)
74% encountered misinformation weekly. 62% of parents felt overwhelmed. <20% trusted existing fact-checkers. Users wanted peace of mind, not just truth.
The Generational Sandwich
Mothers filtering misinformation for kids while protecting elderly parents from scams. Managing media literacy for two generations at once shaped the archetype strategy.
The Transparency Demand
Users wanted to see the incentive structure behind content, not just fact-check labels. Understanding who benefits mattered as much as knowing what's true.
The Platform Trust Problem
Testing rejected a Facebook-integrated solution immediately. Users wouldn't trust the platform spreading misinformation to also solve it. The tool had to be independent.
The platform that spread the problem was structurally incapable of being trusted as its solution.
03 — Features
The quiz builds the skill, Search and Scan provide the tool at the moment of need, Ask Us is the human fallback at the tool's limit
Each feature came from a specific gap the research exposed — not a feature list.
The pivot: from Facebook to standalone
The first iteration was a Facebook Misinformation Center — fact-check labels, community reporting, a hub in the feed. Functional, logical, and dead on arrival: users wouldn't trust the vector of the problem as its solution.
Feature 01 — Literacy Quiz
Tutorials and explainers weren't changing behavior. Testing on early "how to spot misinformation" modules showed users disengaging as if watching homework.
The first quiz format — read a claim, type a response — lost users by question four.
The constraint became the feature:
Under 40 seconds per session
One hand, on a train
Real social media content (not sanitized examples)
Swipe-based True/False interaction
Try it yourself ↓
Question 1 of 5
This photo shows recent flooding in Seattle caused by climate change.
The quiz was designed in the visual register of the platforms where misinformation actually spreads, not the register of an educational tool.
After Misinfo Day, completion states were redesigned around skill progression rather than score — "You're getting better at spotting this" outperformed a percentage.
Users wanted to know why an item was false, which conflicted with the 40-second constraint. The resolution: speed for the question, depth for the reveal, sources one tap away.
Feature 02 — Search & Image Search
Verification had to come to users, not the reverse. A unified entry point handled both links and images so users didn't have to choose a tool at the moment of uncertainty.
Four states, not two
True, Misleading, False, Satire. The Misleading state was critical: most misinformation isn't false, it's selectively true and context-stripped.
Badge + evidence
Fast badge upfront, one tap to reveal sources, fact-checkers, and reasoning. Users wanted the why, not just the verdict.
Multiple sources
Reuters, AP, a fact-check org side by side — institutional backing users could verify themselves.
Feature 03 — Scan
Scan came from Misinfo Day: what happens when the misinformation is a printed flyer, a newspaper, a poster?
Older users hesitated and erred on forms, but scanned instantly and confidently — a behavior already learned from restaurant QR codes. OCR converted headlines to searchable text invisibly.
Feature 04 — Ask Us
Some misinformation exists in no database — private WhatsApp forwards, local rumors, freshly manipulated images.
Automated systems can't catch what they've never seen. Users who distrusted platform verdicts wouldn't trust an AI verdict on emotionally charged content either.
Human judgment backed by journalistic expertise was the only answer.
Clear response-time expectations
Pending / Under Review / Responded status system so users knew what to expect.
Three-step simplification
After Misinfo Day, the multi-field form was reduced to three steps because older participants were struggling with entry.
The design system holding it together
Four features, one visual language — component states, spacing, and color built to scale across the app rather than to each screen individually.
Design system — scroll to see more →
The quiz builds the skill, Search and Scan provide the tool at the moment of need, Ask Us is the human fallback at the tool's limit. None of them tell users what to think.
"The prototype is very well crafted… the experience looks easy and also fun."
— capstone reviewer
Let's get into the details now.
04 — Market Research
Verification had to live where misinformation spreads, but platform ownership destroyed trust
Existing tools fell into two categories, neither worked:
Monetized Platforms — Credible but Compromised
Snopes had trust and rigor, but ran ads. Facebook labels and Twitter Community Notes had scale but were seen as biased. The entity profiting from misinformation's spread can't be trusted to flag it.
Nonprofit Resources — Trustworthy but Invisible
Literacy initiatives had integrity but no engagement — text-heavy resources in a short-form-video era, browser extensions desktop-bound and friction-heavy.
The gap: users who trusted Reuters and the Guardian, wanted multiple sources, and read news on phones inside apps they already trusted. Every existing tool asked them to leave and do extra work — friction that was fatal.
Verification had to live where misinformation spreads, but platform ownership destroyed trust.
The integration insight. I brought this paradox to my professors, who thought I was digging too deep for an academic project. The answer: a trusted utility embedded in trusted surfaces — Safari share sheet, Apple News, native camera — backed by institutions users can verify, owned by none of the platforms it lives in. Not another app; infrastructure that makes verification a natural step.
"Great idea to bring in UI scenarios outside of the app itself, like the lock screen widget."
— capstone reviewer
05 — User Research
Misinformation moves through trust networks — families, WhatsApp groups, neighborhood pages — in a specific generational pattern
2,000 survey respondents · 28 paid interviews · 14 usability sessions · ~1,800 tested live at Misinfo Day, from high schoolers to the creator of the SIFT methodology.
Early task-based testing — verify a headline in under 30 seconds — surfaced ambiguous labels and slow search response times, which shaped a later shift to color-coded ratings and a single, unified search/image-search entry point.
Misinformation moves through trust networks — families, WhatsApp groups, neighborhood pages — in a specific generational pattern.
The finding that changed the problem: Adults 55–80 receive misinformation from people they trust deeply. Teens 13–18 encounter it through peer dynamics where questioning carries social cost. Adults 25–40 sit between both — the most exhausted, managing it for two generations at once.
The archetypes emerged from that chain, not a persona template:
Truth Seekers (18–30) — already motivated, cross-referencing sources, wanting to be community validators, not just skeptics.
Overwhelmed Guardians (25–40) — desperately want verification tools, but any added friction fails with this group regardless of quality.
Vulnerable Believers (55–80) — trust what they receive because it comes from people they know. Design for them means meeting existing behavior, not demanding new habits.
The strategy: design primarily for the first two, since they become the human layer protecting the third — a mother who learns to verify a link becomes the person her parent calls before forwarding it.
"I loved that you used the quotes from interviews. It gave it a lot of meaning."
— capstone reviewer
What Misinfo Day revealed that recruited testing couldn't. Small-scale testing shows what people say; live testing with unrecruited crowds shows what they do. Younger participants abandoned any flow that felt like reading within seconds. Older participants trusted Ask Us over automated ratings — a person's judgment over an algorithm's output. The sharpest insight: engagement spiked when detection was framed as a skill, not a correction — people wanted to feel smart for getting it right, not told they were wrong.
This was also the moment Scan got decided. Attendees kept asking whether it only worked on stuff already on their phone, or could check something in front of them right now — a flyer, a printed article. That gap became Scan: camera-based verification for printed and in-person content a purely digital tool would have missed.
Misinfo Day — live testing with ~1,800 participants · scroll to see more →
06 — TrueMedia
Detection is the expensive end of the chain — authenticating content at creation is fundamentally more efficient, but nobody profits from implementing it
TrueMedia.org was a nonprofit building deepfake detection for the 2024 election. Through UW's partnership, I contributed research synthesis on what to prioritize. They shipped a dark, utilitarian tool for journalists; mine served a teenager asking "why should I care?" — two solutions to adjacent problems. They analyzed 60,000+ pieces of media, launched in September 2024, and shut down in January 2025, open-sourcing the technology rather than chasing funding. Founder Oren Etzioni: "We are not prepared for a large-scale, generative AI attack. It hasn't come yet. That doesn't mean it won't."
Two lessons reframed the market for me: platforms are ambivalent about detection because it drives engagement, and the real unaddressed threat — personalized scams — never reaches a community to verify it. Detection is also the expensive end of the chain; authenticating content at creation is more efficient, but C2PA proposed exactly that and stalled on incentives. A third came from testing: participants rejected a Facebook-embedded concept as too platform-tied and easy to dismiss as biased, settling the case for the integration vision over a standalone app.
07 — Reflection
A designer who thinks only about what users see is a UI designer. A designer who thinks about everything required to make that experience real and sustainable is a product leader
What this project taught me. Design alone cannot make a public good sustainable — TrueMedia was research-grounded and mission-complete, and it still closed on economics, not design. The compass is trusted because it has no business model; the phone manufacturer absorbs the cost. That's why the integration vision — Apple News, the share sheet, the native camera — is the sustainability answer, not a nice-to-have. Coming back to school after industry, I couldn't stay inside a class project's boundaries; every decision triggered questions about moderation staffing and who pays. My professors thought I was overcomplicating an exercise. I've come to see that instinct as the point: a designer who thinks only about what users see is a UI designer; one who thinks about what makes it real and sustainable is a product leader.
What changed from feedback and testing. Reviewers pushed the quiz to teach the why behind each answer, producing the post-answer reveal, and flagged visual and navigation issues fixed in the refinement pass. Usability testing showed people liked the clarity of color-coded credibility badges but wanted to know why — pushing the rating system from a flat badge into a layered one: quick signal upfront, reasoning a tap away.
"Demo was very well designed. Excited to see the next steps."
— capstone reviewer
If I built this in 2026. In 2024 this needed an organization — moderators, funding, a maintenance team. By 2026, AI collapses that requirement: triage for Ask Us, monitoring to keep quiz content current, AI-augmented research at scale. What took a fifteen-person nonprofit now takes a determined individual. The capstone answered the design question; the ecosystem question — reaching people at scale, outliving a funding cycle — is still open.
