The popular question-and-answer site for coders and programmers, Stack Overflow, has temporarily prohibited users from sharing responses generated by Open AI’s chatbot ChatGPT.
The ban was imposed because ChatGPT’s average rate of correct answers is too low, as it did not adequately deliver the simple prompts it was given. The tool generated blogs and even songs in some cases. Open AI also observes that ChatGPT occasionally writes “plausible-sounding but incorrect or nonsensical answers.”
ChatGPT, which is based on OpenAI’s GPT-3 language model, responds to prompts in a human-like manner but provides flawed answers such as incorrect biographical data, incorrect codes, and so on.
Stack Overflow stated that the ban was only temporary and that a final decision would be made after consulting with its community, but that it was taking action now because it was concerned that ChatGPT could be significantly harmful to both the organization and its users.
ChatGPT has over one million users since its launch last Wednesday.
The sources for this piece include an article in TheRegister.