G000168 "Darwin AI"
Sector: Technology
Primary role:
Secondary role:
Primary subject:
Secondary subject:
Has volunteers:
Region: North America
Country: Canada
Twitter handle:
Summary: The Darwin AI team has developed an “Automatic Disinformation Assessment” engine. It is a large scale, bidirectional language model that can detect the stance of an article. The project had an accuracy of 90.01%, which surpasses general benchmarks in the industry. The tool was able to determine whether or not a document agrees, disagrees, or takes no stance towards a specific claim.