• What it is: An AI-powered media framing & sentiment analysis tool
• What it does: Compares how conservative and liberal outlets use tone, metaphor, and polarizing (splitting) language for a meta analysis of the news
• AI used for: Sentiment analysis, metaphor detection, tone classification, language pattern extraction
• Built as: Interactive prototype (Figma Make)
• Tools: ChatGPT / Claude / NotebookLM / Figma Make
• Why it matters: Makes media bias and narrative framing visible without prescribing conclusions

This prototype builds on my experience leading AI-powered media products at Dotdash Meredith and The Wall Street Journal. Below is a deeper look at the thinking, research, and design decisions behind NewsWise.

Context

As AI becomes embedded in news production, distribution, and personalization, the challenge is no longer just misinformation, but how AI-mediated language shapes meaning, emotion, and belief. Many AI tools optimize for engagement or credibility scoring, yet few help readers understand how news narratives themselves are constructed.

Problem

Readers increasingly struggle to interpret politically charged media ecosystems shaped by algorithmic amplification. While bias is widely discussed, the linguistic mechanisms—framing metaphors, emotional tone, and polarizing language—remain largely invisible. Existing AI-powered media products tend to judge or rank content rather than reveal how meaning is produced.

Insight

Drawing from my interest and background in anthropology and linguistics, I approached news as a cultural artifact rather than a neutral transmission of facts. My research suggested that readers are often influenced less by explicit claims and more by framing devices: metaphors that imply threat or decay, emotionally charged verbs, and “splitting language” that reinforces in-group versus out-group identity. Making these patterns visible could help readers develop greater media literacy without prescribing ideological conclusions.

Prototype

I designed NewsWise, a meta-analysis dashboard that compares conservative and liberal media sources side by side. The prototype uses AI-assisted analysis to surface:

• Sentiment and emotional intensity
• Framing metaphors used across topics
• Tonal patterns (e.g., alarmist, moralizing, dismissive)
• Language that reinforces polarization

The interface emphasizes comparison and pattern recognition rather than scores or judgments, allowing users to explore how stories are framed rather than being told which source to trust.

Outcome & Implications

NewsWise reframes AI’s role in media from arbiter to interpreter—supporting reflective reading instead of passive consumption. The project suggests a broader design opportunity for AI in journalism: tools that increase transparency, reveal narrative structure, and strengthen human judgment rather than replacing it. This work directly informed my ongoing exploration of trust, explainability, and bias-aware design in AI-powered media products.