Artificial intelligence: GPT-4 impresses better than people in debates!

Artificial intelligence: GPT-4 impresses better than people in debates!

Lausanne, Schweiz - A current study has shown that the AI ​​model GPT-4 in online debates is above average convincing. According to researchers from the Swiss Federal Technical University of Lausanne (EPFL), it was found that GPT-4 was more successful in 64.4 percent of cases as a human participant if it had access to personalized data such as age, level of education or political attitude. This enables the AI ​​to adapt their argument in a targeted manner and thus increase persuasiveness. Without such information, however, the effectiveness of the AI ​​was comparable to that of human discussion partners. These results were published in the renowned journal "Nature Human Behavior" (Bluewin] (https://www.bluewin.ch/de/news/international/gpt-in-in-online-debatten-teuenderer---270375.html).

The study comprised around 900 participants who debated intelligence in structured online debates on various sociopolitical topics such as the ban on fossil fuels and the effects of social media. The research team leader Francesco Salvi emphasizes the importance of personalized arguments and urges online platforms and social media to take the risks resulting seriously through AI-driven persuasive strategies and to take appropriate measures to contain. This is particularly relevant because the results are not related to a controversial study by the University of Zurich, which caused turmoil in the Reddit community Krone.

artificial intelligence and their effects on social media

artificial intelligence has fundamentally changed the way we interact online and consume content. AI applications are not only found in art and text generation, but also influence how social media are operated. The personalization of content is a great advantage because it offers users relevant information and increases efficiency. For example, platforms such as Twitter and Instagram use AI tools for content moderation and targeted advertising [Computer Weekly] (https://www.coverteweekly.com/de/feate/die- undeafache-kuenstlicher-intelligenz-auf-zial-Medien).

The use of AI on social media not only brings advantages, but also risks. The possibility of bias in decisions, the risk of the formation of echo chambers and an inadequate moderation of content are critical points that need to be considered. In addition, active data collection via users is often classified as problematic. Clear measures and responsible use of AI in these platforms are therefore urgently required to protect the integrity of the online discussions.