Misinformation spreads rapidly online, in the news, and around the world, with real consequences for us all. VeritablAI combats this by analyzing spoken statements and checking them against well-verified and well-cited Wikipedia data. In real time, it displays a continuously updating truthfulness score, allowing audiences to instantly see whether claims are accurate or misleading. Deployed in media or public forums, VeritablAI empowers viewers to discern fact from distortion, promoting accountability and informed decision-making.
The core of our project is written in Python, with minor use of SQL and JavaScript where needed. We use FastAPI to serve our web application, Vultr for hosting and running continuous processes, Snowflake to store and query structured data efficiently, and Google Gemini to fact-check transcribed audio against relevant Wikipedia content for accuracy.
This website is using an HTML5 UP template that was
released for free under the Creative Commons.
[He/Him] A junior studying computer science with a minor in mathematics. He is highly experienced in programming in several languages, as well as extensive debugging and contributing to GitHub repositories. He is fascinated by the capabilities of artificial intelligence and was up to the challenge of combining various frameworks to build the realtime misinformation detector.
[They/Them] A junior studying computer science, astrophysics, and physics. They are highly passionate about black holes and providing a pathway of knowledge to beginner astronomy enthusiasts. They are highly enthusiastic about trying new web frameworks and expanding their skillset as they partake in each new project.
We have participated in four UMass hackathons on the same team thus far, designing projects ranging from RPG games to innovations of useful AI tools that transform the world around us. We have combined our knowledge of programming and physics, as well as our willingness to learn new frameworks to efficiently design projects within approximately 30 hours. Two of our selected projects are shown below:
Our previous project that used machine-learning algorithms that analyze spectroscopy data for detecting the presence of black holes. It won two awards at Hack(h)er 413 last year.
Not an AI tool, but it was a helpful concept of a robotic companion that can help hospital patients with mobility challenges. It would move around with an internal gyroscope, supporting sufficient weight for small objects.
What questions do you have about our AI products? We will happily provide more information about them, and we are happy about your interest in VeritablAI and our other projects.