This commentary was published in commemoration of International Fact-Checking Day 2024, held April 2 each year to recognize the work of fact-checkers worldwide. Olivia Sohr is news initiatives director and Franco Piccato is executive director at Chequeado, a fact-checking organization based in Buenos Aires, Argentina
Generative artificial intelligence has produced tools capable of generating entirely fabricated images and videos, and that’s made fact-checkers’ work more difficult and more complicated.
But at the same time, AI has the potential to enhance the efficiency and speed of fact-checking efforts. If fact-checkers around the world could combine their efforts on these new tools, the impact on fact-checking in multiple languages and in multiple formats would be enormous.
Our work with AI at Chequeado has already revealed the potential of automation and experimentation in fact-checking. At the core of our approach lies the Chequeabot, a central AI application that has evolved since 2016 to facilitate faster and more accurate fact-checking.
This AI-powered tool enables Chequeado to identify checkable statements, detect previously verified data and transcribe real-time events:
- Identifying checkable statements: AI automatically identifies statements for fact-checking within any text.
- Detection of verified data: Given a statement to verify, we use AI to automatically detect if it has been previously verified and provide data to facilitate its verification.
- Real-time transcription: AI transcribes videos into text for live verification of speeches and events.
Thanks to Chequeabot, our fact-checking process has accelerated, resulting in hundreds of hours saved for our team and seven other organizations. Transcription time has been reduced by 90%, with one person now managing a task that previously required six. This automation not only expedites verification but also allows fact-checkers to focus on analyzing complex claims and narratives.
Additionally, we are exploring using AI for new tasks that would increase the impact of our work. Such tasks include:
- Text-to-image generation: Employing generative adversarial networks (GANs) or similar techniques to automatically create informative graphics to accompany fact-checked articles.
- Text-to-video production: Leveraging natural language and video synthesis to produce informative videos summarizing fact-checked information.
- Script generation from verified materials: Developing AI-powered tools to generate video scripts based on previously verified materials that accurately convey information while maintaining audience interest.
- User-generated content analysis: Using AI techniques to analyze user-generated content, such as comments, posts, and tweets, in order to identify potential disinformation.
Fact-checkers must navigate various challenges in integrating AI wisely. It requires cultivating proficiency in prompt engineering, embracing risk-taking, adopting a more tech-entrepreneurial mindset, and securing funding for large-scale technological development. Additionally, establishing guidelines to ensure ethical and transparent utilization of AI is imperative.
In addition to the challenges faced globally, fact-checkers from the Global South face specific hurdles including sourcing and hiring AI-specialized professionals, low levels of digital literacy, and the predominance of AI tools in English, exacerbating language barriers. These challenges compound existing gaps in knowledge, resource limitations, and infrastructural, legal, and political obstacles prevalent in these regions.
While the challenges are many, we’ve found it a necessity to cultivate collaborative networks, facilitating information sharing and coordination among fact-checking organizations globally. Numerous fact-checkers are spearheading innovation in several areas, such as Full Fact in the United Kingdom; Newtral and Maldita in Spain; Aos Fatos in Brazil; and OjoPúblico in Peru — the latter notably utilizing AI to produce fact checks in indigenous languages.
Looking ahead, the lessons learned from Chequeado’s experience offer valuable insights for organizations seeking to harness AI effectively in the fight against disinformation.
As the global landscape continues to evolve, it is imperative for fact-checkers to embrace experimentation and collaboration to uphold the integrity of information and safeguard democratic processes. Continuing the dialogue with fellow fact-checkers is vital for synchronizing efforts and maximizing impact.
