With the rise of artificial intelligence models such as Grok and ChatGPT, many internet users’ instinct is to use them as arbiters to settle debates on social media. However, these tools are not infallible. Despite the apparent reliability of Perplexity or ChatGPT’s ability to perform real-time searches, these AIs are prone to hallucinations and can provide fabricated information. Using these models to verify facts raises issues, making them unreliable when it comes to confirming or denying claims.With the rise of artificial intelligence technologies, more and more internet users are turning to models like Grok and ChatGPT to verify information shared on social media. While these tools may seem promising, they present significant risks due to their intrinsic limitations and their potential tendency to generate « hallucinations. » The following article examines these dangers in detail. The Reflex of Using AI for Verification In today’s digital climate, it’s common to turn to automated solutions to resolve online debates. Grok, developed by the parent company of X (formerly Twitter), is frequently invoked in discussions when information seems questionable. The application has become widespread among internet users, as has ChatGPT, a popular tool for confirming or denying claims. AI Hallucinations A major problem associated with the use of Grok and ChatGPT for information verification lies in what are called « hallucinations » of the AI models.These models are not specifically designed for fact-checking and can fabricate information. Even ChatGPT, often perceived as more reliable thanks to its real-time search function, is not immune to these abuses. The Inadequacy of Models for Verification Artificial intelligence tools like Grok and ChatGPT are not developed for fact-checking. They are designed to generate text based on available data. Consequently, they lack the ability to critically analyze and cannot distinguish between a verified fact and an unfounded assumption, thus generating potentially erroneous results.
The Craze for AI in Public AffairsA striking example of the use of AI in public affairs is the public stalking of Xavier Dupont de Ligonnès, orchestrated on Instagram. The participants and the administrator used AI tools like ChatGPT to explore « leads, » illustrating the craze for these technologies. However, the hope for an informed resolution through AI has run into the reliability limitations of these models.Conclusion: Precautions to Take In conclusion, although Grok and ChatGPT offer impressive capabilities in various fields, their application to information verification remains dangerous without rigorous validation by reliable sources. It is essential to keep these limitations in mind to avoid the spread of potentially harmful misinformation in the public sphere.
