Research

The research called “Retraining fact-checkers: The emergence of ChatGPT in informative verification”, carried out by Roger Cuartielles, Xavier Ramon-Vegas and Carles Pont-Sorribes of the Pompeu Fabra University, analyzes the impact of ChatGPT on the work routines of the main fact-checking platforms in the country. The study is based on interviews with verifiers from Maldita.es, Newtral, EFE Verifica, AFP Factual España, Verificat and VerificaRTVE.

The research concludes that ChatGPT’s impact on verification is ambivalent. On the one hand, it is perceived as a tool that can facilitate the detection of uninformative patterns and accelerate the collection of context information. Verifiers agree that it can be an auxiliary tool in the phase of detection of falsehoods and the creation of more effective denials.

On the other hand, however, a number of problems and risks are also identified. One of the main drawbacks is the lack of transparency in the sources, which makes it difficult to verify the contents it generates. In addition, its ability to generate coherent, but not necessarily true, texts poses the danger of a greater spread of disinformation. Experts warn that its accessibility would allow the creation of false information in a more sophisticated way and in different languages.

This duality makes some verifiers prudent. “Their usefulness will depend on how it evolves and whether its functionalities are improved, especially in the identification of reliable sources,” says Pablo Hernández, from Maldita.es.

The consulted verifiers agree that the emergence of ChatGPT also redefines its profession. The appearance of AI reinforces the role of fact-checkers as context agents, responsible not only for denying false information, but also for rigorously explaining and contextualizing news. In addition, they believe that artificial intelligence forces them to adapt and collaborate even more between platforms to combat disinformation. In this sense, they claim the need to accelerate media education initiatives to teach citizens to use these tools responsibly. 

The study highlights that ChatGPT can become both a support tool and a challenge for fact-checkers. Its ability to gather information quickly is an advantage, but its lack of reliability in sources and the possibility of generating falsehoods make it a tool for limited use in fact-checking.

The authors of the research stress that more transparency, supervision and technological development will be needed in order for these tools to be effectively integrated into the fight against disinformation. For now, fact-checkers are still irreplaceable to guarantee truthfulness in an increasingly complex media environment.

To read the full article and access the detailed results, check it here.