The State of Automated Factchecking
The State of Automated Factchecking is an in-depth report looking at where we are with automated factchecking globally, and where we could get to with the necessary funding.
It sets out Full Fact's roadmap for our own work on automated factchecking, and our design principles for the tools we are building.
We propose principles of collaboration for factchecking organisations, researchers and computer scientists around the world.
We hope that it will be the beginning of many fruitful conversations.
It's split into two parts:
- Part One: A roadmap for automated factchecking
- Part Two: What we can do now and what remains to be done
- We can scale up and speed up factchecking dramatically using technology that exists now.
- We are months—and relatively small amounts of money—away from handing practical automated tools to factcheckers and journalists. This is not the horizon of artificial intelligence; it is simply the application of existing technology to factchecking.
- Automated factchecking projects are taking place across the world, but they are fragmented. This means factcheckers and researchers are wasting time and money reinventing the wheel.
- We propose open standards. Automated factchecking will come to fruition in a more coherent and efficient way if key players think in terms of similar questions and design principles, learn from existing language processing tasks, and build shared infrastructure.
- International collaboration is vital so that the system works in several languages and countries.
- Research into machine learning must continue, but we can make serious progress harnessing other technologies in the meantime.
Read The State of Automated Factchecking (pdf, 6Mb) and sign up below to keep up with the latest.