Seeing Is No Longer Believing: How AI Is Rewriting Trust Online [2026.15]
Our mission is simple: to keep the UVA community informed, engaged, and inspired as we navigate this transformation together.
Teaching Spotlight: AI, Images, and Trust at UVA
👉 Main idea: AI is making fake images easier to produce, but the bigger challenge is protecting trust and judgment in a world where seeing is no longer believing.
Mona Kasra, Associate Professor of Digital Media Design, brings together visual culture, creative practice, and democratic life to examine how AI is reshaping the way we see, interpret, and trust images.
“For the longest time, we could trust photography in general… and now we are not at that place anymore”. - Prof. Kasra
🔑 Key Insights from the Podcast
The real problem is not just fake images, it is collapsing trust.
As AI visuals get harder to detect, they create broader skepticism around what counts as evidence, truth, and reality online.AI literacy now has to include visual literacy.
People need to understand how image systems work, how bias enters outputs, and how provenance and metadata help trace where images come from.This is a civic problem, not just a technical one.
The conversation frames synthetic media as a challenge for democracy itself, because manipulated or subtly altered images can shape public reaction, political meaning, and collective judgment.
The next step is building visual AI literacy strong enough to protect trust without giving up on truth.
Mentioned in Podcast:
Digital Technology for Democracy Lab — Pioneering transdisciplinary research to explore democracy in the digital age
C2PA (Coalition for Content Provenance and Authenticity) — provenance standard for tracking edits and authorship of digital media
Insights from Boris Cherny (Claude) and Bryan Catanzaro (Nvidia)
I was at Capital One’s headquarters this week for an event featuring Boris Cherny, who created Claude Code, and Bryan Catanzaro from NVIDIA, who leads much of their AI reasoning work. Both gave talks that were accessible to a non-technical audience. Some takeaways:
Code review is no longer a bottleneck. We’ve known for a while that AI writes production code. But I hadn’t registered the review piece. Code review has always been the chokepoint: senior engineers are expensive and busy.
Takeaway: AI agents now handle that review process, which means code ships faster and senior engineers spend their time on harder problems.
Non-engineers are writing code. Product managers, Designers, Marketers, Finance people, and so on. They're building with AI tools and shipping code without engineering backgrounds. Most org charts still separate "technical" and "non-technical" roles. Anthropic skipped that entirely. Everyone there is a "Member of Technical Staff," the same title series Bell Labs used in the 1960s.
Takeaway: Job roles are fading. I teach in a business school, and this changes what we should be preparing students to do.
Engineers are becoming conductors. Instead of writing every line or reviewing every pull request, engineers are managing fleets of coding agents, sometimes hundreds at a time. They set direction, define constraints, review what the agents produce, and step in when something breaks.
Smaller models, tighter outputs. NVIDIA is building specialized models that return focused answers instead of walls of text. Fewer tokens, lower cost, same quality. Anyone paying API bills should pay attention.
A fourth scaling law. This one I’m still chewing on. The three existing scaling laws in AI (pre-training, post-training, and reasoning) each drove model intelligence forward but hit compute limits. (NVIDIA’s blog explains the progression well.) Last week, a fourth was proposed: agentic AI. Every scaling law demands more compute, and compute is finite.
Takeaway: NVIDIA said they’ve hit cost-of-compute limits on every scaling law so far. These are the people who make the hardware.
Research: For Business Student ChatGPT narrows the gap by helping weaker students and dulling stronger ones.
Carsten Bergenholtz, Oana Vuculescu, Franziska Günzel-Jensen, & Lars Frederiksen (2026). Leveling Up or Leveling Down? The Impact of Generative AI on Student Performance in Business Schools. Academy of Management Learning & Education. DOI
💡 The Big Idea
Giving business students access to GPT-4 improves performance overall, but unevenly. Lower-performing students gain the most, while higher-performing students lose some of their advantage, narrowing the gap.
🧪 How they did it
Randomized study of 141 business-school students.
Students completed two short case-analysis tasks.
In the second task, one group could use GPT-4; the control group could use the internet, but not chatbots.
Responses were graded on a 1–10 scale, and researchers also interviewed students about how they used the tool.
📈 Key findings
GPT-4 users scored 5.48 on average in Task 2, versus 4.32 for the control group.
Chatbot access was linked to a 0.82-point gain overall.
Lower-performing students saw the biggest benefit, with an added 1.79-point boost.
Higher-performing students did not benefit in the same way and in some cases did worse.
Better students tended to use prompting more actively, but most students used the tool only lightly.
🚀 State of AI in the Commonwealth Report
Around UVA
Now available from UVA Research Computing is UVA RC GenAI, a high‑powered generative AI platform running on Kimi K2.5, an open-source, 1-trillion parameter multimodal model well suited for visual analysis, agentic applications, and code‑based tasks. Open to Research Computing users, the tool runs on eight NVIDIA H200 graphics processing units (GPUs) and allows researchers to submit up to 60 queries per minute at no charge.
AI Events @ UVA
KNOWLEDGE CONTINUUM — THIS FRIDAY!
🗓 May 1, 2026| Charlottesville, VA
Join us for an exciting day of learning and networking at the Knowledge ∞ Continuum—the signature event of the Center for the Management of IT (CMIT).
Bringing together world-class speakers, practitioners, alumni, faculty, and students, the program explores cutting-edge topics at the intersection of technology and business through keynote insights, expert panel discussions, and a showcase of emerging ideas from UVA student hackathon teams.
Upcoming Events at AI Research Initiative @ UVA
Agentic Workflows in Practice
📅 June 5 | 12:00–2:30 PM
Instructor: Reza Mousavi, AI Guide and Associate Professor of Commerce
Your First API Call: Scaling AI on UVA HPC
📅 June 10 | 12:00–1:00 PM
Instructor: Josh Baller, Research Computing
Smarter Literature Reviews with AI-Powered Tools
📅 June 17 | 12:00–1:00 PM
Instructor: Stephen Turner (Data Science),See Stephen's Substack
→ More info / Register here
2nd Annual Cosmic Horizons Conference
🗓 July 13 - 26, 2026| Charlottesville, VA
The NSF-Simons AI Institute for Cosmic Origins (CosmicAI) is excited to announce the 2nd Annual Cosmic Horizons Conference hosted by the NSF National Radio Astronomy Observaotry (NRAO).
The recent revolution in AI is fundamentally changing how astronomers observe, explore, analyze, and model astronomical data. The Cosmic Horizons Conference aims to bring together researchers who are actively developing and applying AI/ML methods in astronomy.
UVA AI Resources
AI in the Curriculum Playbook: A practical framework for intentionally embedding AI capabilities into any program or course.
AI for Academic Excellence - Student Toolkit: A comprehensive guide for students on the best uses of AI.
AI Agents in Economic Research (Anton Korinek): A guide for researchers on the use of AI agents.
UVA Claude Builders Student Club: A 250+ strong group for those interested in development via Claude.
UVa AIML Seminar - Seminar featuring artificial intelligence, machine learning, and their applications.
UVA Podcasts We Listen to
Co-Opting AI: Public Conversations About Artificial Intelligence and Society:
Prof. Mona Sloane’s series Co-Opting AI is a virtual public speaker series that interrogates and demystifies AI.UVA Data Points: Podcast from the School of Data Science.
HOOS in STEM: From Prof. Ken Ono this series showcases the marvelous cornucopia of STEM at UVA, from the latest innovations to growth inside and outside the classroom.
Thanks for reading AI @ UVA Substack! Subscribe for new posts and podcasts.