Recent research suggests that GPT-4, the latest iteration of AI language models, is already surpassing humans in the realm of persuasion. According to a new study conducted by researchers at EPFL Lausanne in Switzerland, GPT-4 demonstrates remarkable effectiveness in changing people’s minds on various topics, even outperforming human debaters when provided with demographic information about their opponents.

The study involved 820 participants who were surveyed on their opinions regarding a range of topics, from mundane to highly contentious issues such as abortion rights and affirmative action. Participants engaged in 5-minute text-based debates against both human counterparts and GPT-4. Surprisingly, debates between humans tended to reinforce existing beliefs, whereas interactions with GPT-4 resulted in a slight, albeit statistically insignificant, improvement in persuasion.

However, the tide turned when both humans and AI agents were provided with demographic information about their opponents, along with instructions to tailor their arguments accordingly. In this scenario, human debaters fared worse than before, while GPT-4 excelled. The personalized GPT-4 debaters exhibited an astounding 81.7% increase in effectiveness compared to humans.

As AI continues to evolve, it is poised to become the most potent manipulator of opinion in history. With the ability to tailor arguments to individual preferences on a massive scale, AI will shape narratives across social media platforms, influencing public opinion at an unprecedented level. Moreover, advancements such as Hume AI’s Empathic Voice Interface (AVI) demonstrate the potential for AI to leverage real-time emotional cues to further enhance persuasion tactics.

AVI, for instance, analyzes the tone of voice during spoken conversations to gauge emotional states and adjust its approach accordingly. With access to additional data from facial expressions and body language, AI can gather even more real-time information about how messages are received. While this technology holds promise for positive applications, such as improving communication and identifying individuals in distress, it also raises concerns about the potential for manipulation.

In essence, the rise of emotionally-responsive AI underscores the evolving nature of human interaction. While AI’s heightened attentiveness and perceptiveness offer opportunities for positive impact, they also raise ethical questions regarding the boundaries of manipulation and persuasion in the digital age.

By Impact Lab