AI systems are capable of coding in every language, but they often make up stuff, a phenomenon known as hallucinations. This issue is prevalent in language, audio and visual models and has been a long-standing topic of discussion among AI researchers.
While some researchers are frustrated with AI hallucinations, others see their creative potential, suggesting these models could serve as valuable 'co-creative partners'. However, there are concerns that as chatbots become more reliable, people might start trusting them excessively, which could be problematic.