What is disinformation according to AI?
What is disinformation according to AI?

Summary/Abstract: In recent years, the EU has identified disinformation as one of the most significant risks to social health and society. The development of artificial intelligence (AI) has further facilitated the creation and dissemination of false and/or manipulated information. But does AI understand what disinformation is, and can it identify it independently? Are there differences in how the five most popular generative models define disinformation? Which model – ChatGPT or BgGPT – has more insights on this topic? This article aims to explore and systematise the definitions of disinformation provided by the most widely used language models. The main hypothesis is that commonly available language models can define basic concepts like "disinformation" based on both the information available online and their own ethical standards, if they have any. However, it is acknowledged that there may be some variation in responses across different models. The paper will present the responses in a comparative table, utilizing scientific methods such as comparison, analysis, and synthesis, along with cyberethnographic and autoethnographic research methods. In the discussion, the article will examine both the risks and the opportunities posed by the intersection of disinformation and AI.

• Details
• Contents