The world of social communication is undergoing a significant shift. Artificial intelligence is being used to manage how people interact online. This approach is based on the idea of behavioral economics, which looks at how people make decisions. This concept is not new. Daniel Kahneman, a famous psychologist, talked about two systems in the brain. System 1 is intuitive and quick. System 2 is more logical and slow. Now, a third system is being introduced. System 3 is about how people act based on their identity and society's norms. This system is influenced by what others think and how people see themselves.

    The goal is to use AI to handle social communications better. This way, people can focus on making their own choices and organizing themselves. AI can help create messages that support social change. These messages must respect the history, culture, and politics of the people they are meant for. The GPT-4 model is being used to generate these messages. An experiment was conducted to see how effective these messages are. Virtual people were used in this observational study.

    The experiment compared the impact of original messages and modified ones. The Claude 3. 5 Sonnet model was used for this comparison. The results showed that the modified messages, which were crafted to support social transformation, caused attitudes to become more negative. There was a 2. 5 units downward shift in the median. However, the intentions showed a slight positive increase, with a 0. 2 units upward change in the median.

    This raises an important question. If AI-generated messages make people more negative, is it worth using them? It is clear that AI has the power to influence people's behavior. But it is also clear that this influence can have unintended consequences. It is important to think critically about how AI is used in social communications. The potential for AI to support social transformation is great. But so is the potential for it to cause harm.

Social Media and the Power of AI

Actions