60 Minutes Made Erroneous Claims About ChatGPT’s Abilities

AI Horror Stories cover art and logo

60 Minutes, the popular new program on CBS aired a story last Sunday that included a segment showcasing conversations with distinguished individuals, such as Google CEO Sundar Pichai, while presenting debatable assertions concerning one of the organization’s extensive language models (LLM).

The central theme of the video revolves around the concept of emergent behavior, which refers to an unforeseen outcome of an AI system that its creators did not necessarily anticipate. Instances of emergent behavior have been witnessed in recent AI ventures. For instance, in a study published online last week, researchers employed ChatGPT to develop digital characters with objectives and backgrounds. The system displayed various emergent behaviors, including exchanging novel information between characters and forming relationships with one another—actions that were not part of the original design.

It is undoubtedly a relevant subject for a news program to explore. However, the 60 Minutes segment becomes contentious when it introduces allegations that Google’s chatbot was capable of independently learning a language it had not previously known upon being prompted in that language.