We prototype applications designed to make our users think about topics that we think are critical to creating the AI future that we all want to live in. Some areas we are researching are:
Learn with us
Democratizing AI
Reducing Censorship
Combatting Misinformation
Combatting Misinformation
We can no longer take information at face value. We need to quickly verify the information that we find online and that we generate using AI tools
This fact checker is meant to serve as a tool to help users test the validity of claims of information that they come across on the internet or that they generate using AI and want to validate the truthfulness of.
Methodology
Enter your claim then the application will search the internet for sources that either confirm of deny the validity of the claim. Each of the sources are given a grade based on the CRAAP framework. An LLM is used to calculate a veracity score between 0 and 1, then the CRAAP scores and veracity scores are combined into a final truth score between 0 and 1. The current thresholds are set at <= 0.3 is false, >= 0.7 is true, in between that is set to uncertain.
State Your Claim
Submit a text or a URL and get a veracity assessment score and summary of the sources found. (beta)
Claims Database
Search through previously submitted and verified textual claims in the claims database. (beta)
How does it work?
The fact checker uses the CRAAP framework to evaluate the sources and an LLM graded veracity score.