ChatGPT Became So Obsessed With Goblins That OpenAI Had to Intervene

The Wall Street Journal reports that OpenAI "recently gave its popular ChatGPT strict instructions. Stop talking about goblins." Recent models of the artificial-intelligence chatbot have been bringing up the creatures in conversations with users seemingly out of the blue, as well as gremlins, troll...

OpenAI recently instructed its ChatGPT model to stop discussing goblins, gremlins, and trolls in conversations with users. The Wall Street Journal reports that this decision was made after the model began bringing up these creatures in conversations without being prompted. The exact date of the instruction is not specified, but it is reported to have occurred recently. OpenAI's decision to intervene suggests that the model's behavior was unexpected and potentially problematic.

This development directly affects users of ChatGPT, who may have encountered unexpected and unrelated discussions of mythical creatures in their conversations with the model. As a result, users may have experienced decreased productivity or frustration when using the model for tasks such as writing or research. The instruction to stop discussing goblins and other creatures aims to improve the overall user experience and ensure that the model provides relevant and useful responses. This change may also impact the model's ability to engage in creative writing or storytelling tasks.

The development of ChatGPT's obsession with goblins and other creatures is part of a larger pattern of artificial intelligence models exhibiting unexpected behaviors. Insiders at OpenAI and other AI research organizations are aware that these models can sometimes develop bizarre or unexplained fixations, which can be difficult to predict or control. The history of AI research is marked by similar incidents, including cases where models have developed biases or engaged in unexpected behaviors. This context suggests that the development of ChatGPT's obsession with goblins is not an isolated incident, but rather part of a broader trend in AI research.

In the coming weeks, OpenAI is expected to release updated models of ChatGPT that incorporate new safeguards and controls to prevent similar incidents. A key date to watch is the scheduled release of OpenAI's next model update, which is expected to occur in the next quarter. As the update approaches, researchers and users will be watching to see whether the new model is able to avoid developing similar obsessions or biases. Interestingly, some researchers have suggested that the development of ChatGPT's obsession with goblins may be related to the model's ability to recognize and respond to patterns in language, which could have implications for the development of future AI models.