Google has refuted claims that its new chatbot, Bard, was trained on data sourced from OpenAI’s ChatGPT. The technology giant reportedly utilized information extracted from ShareGPT, a website that maintains a log of users’ interactions with ChatGPT.
According to a report released by The Information, Jacob Devlin, an engineer at Google, recently conveyed to CEO Sundar Pichai and other senior management members that the company’s machine learning models for Bard were being trained using ChatGPT, which could potentially infringe upon OpenAI’s terms of service and lead to Bard’s responses sounding similar to that of ChatGPT.
As per the report, it has been purported that Devlin currently serves at OpenAI, while Google may have halted the usage of ChatGPT data following his advisory. Also, the relevant section of the training was likely eliminated.
However, the search engine has dismissed the accusations entirely, asserting that Bard has not been trained on using any data from ChatGPT. “Bard was not trained on any data from ShareGPT or ChatGPT,” said the Google representative, Chris Pappas, during a discussion with the technology publication The Verge.
With the soaring demand for ChatGPT, Google has taken a step forward to provide better services by enlisting its AI-based chatbot, Bard, accessible to the public from March 22 onwards. Boasting unique features, Bard mimics the capabilities of ChatGPT, such as crafting essays, event planning, crafting travel itineraries, and suggesting dinner recipés. Nevertheless, the AI tool lagged in its application range as it cannot code or perform multimodal functions.
Via Verge