This paper investigates the potential of GPT-4V in analyzing social media content. The study involves five tasks: sentiment analysis, hate speech detection, fake news identification, demographic inference, and political ideology detection. The results show GPT-4V’s remarkable efficacy in these tasks, demonstrating strengths like understanding image-text pairs, cultural awareness, and extensive commonsense knowledge. However, GPT-4V faces challenges in multilingual social multimedia comprehension and generalizing latest social media trends. It also tends to generate erroneous information about evolving celebrity and politician knowledge.
Publication date: 13 Nov 2023
Project Page: https://github.com/VIStA-H/GPT-4V_Social_Media
Paper: https://arxiv.org/pdf/2311.07547
