Overview

Contending for a healthier information landscape in the future requires many actors, from academia to policymakers and more. Fact-checking is a key part of this project, and the use of AI to accelerate and better enable fact-checkers is still rapidly developing.

When asked what would be most beneficial to have AI tools help with, factcheckers identify three major areas where automated processes can help: the filtering of claims that may be fact-checked to a manageable volume, retrieval of evidence for a human to evaluate, and triaging of tip lines. This project targets the first filtering aspect through two avenues. It proposes to develop new AI to further the valuable work that fact-checkers already perform.

The two avenues that will comprise the research agenda are to surface potentially false or misleading content from politicians or from individuals on social media: the former will involve an analysis of parliamentary debates recorded in Hansard and news media, and the latter a system to monitor visual disinformation from social media sites such as Instagram and TikTok.

Both avenues will work towards a similar output structure: a regularly updateable shortlist of claims (images, videos, or text) that are identified through AI as fact-checkable and check-worthy. These can then be used by fact-checkers or academia. Answers to all the foregoing questions impact how government can be held accountable, and possibly how policy can be structured to better handle these issues.