Google's Bard AI has once again found itself at the center of controversy, this time over allegations that Bing's rival was trained on data pulled from OpenAI's ChatGPT.
As you may know, ChatGPT is the power behind the throne of Bing AI, and the accusation of nefarious activity behind the scenes comes from a report by The Information (opens in a new tab).
We are told that Jacob Devlin, a software engineer at Google, a former engineer, we might add, who left the company over this matter, claims that Google used data from ChatGPT (pulled from the ShareGPT website, apparently) to expand Bard.
Devlin notes that he warned Google against this, as it clearly violated the OpenAI terms of service.
According to the report, Google stopped using the aforementioned data after warnings from Devlin (who left Google to join OpenAI, we're told).
However, Google denies all this. A company spokesman, Chris Pappas, told The Verge (opens in a new tab): "Bard is not trained on any ShareGPT or ChatGPT data."
Analysis: denial in despair
There you have it, then: a clear denial from Google in no uncertain terms that anything sneaky was going on data-wise with Bard. And to be fair, there's certainly no evidence that Bard's answers remotely resemble those provided by ChatGPT. (Devlin had further warned that the alleged data theft could mean just that, and it would be pretty obvious what happened as a result.)
We guess the problem with this episode is that it really does seem like Google was quick to remove Bard, dropping noises in the process, as it was forced to catch up with Microsoft's Bing AI. With the latter now successfully driving search engine adoption towards Bing already at this early stage, all of this could make it quite easy for some to believe that Google might be getting a bit desperate with behind-the-scenes tactics.
Whether or not the stolen data story is true, we'll take Google's word that it isn't, the report still interestingly reveals that Google's Brain AI group is now working with artificial intelligence firm DeepMind ( both exist under the parent company Alphabet umbrella).
Apparently DeepMind was thrown into the mix to quickly hone and power up Bard, and that's notable because the two AI teams are huge rivals and very much forced to collaborate on this.
This again paints a picture of a rather desperate fight to make Bard more stable, as Microsoft's Bing AI continues to update with new features at a fairly old rate of nodes. (Although the new rumors about one of the potential next "features" of the Bing chatbot have us very concerned, it must be said.)
You may also recall that alarm bells sounded on the privacy front when Bard himself apparently revealed that he was using internal Gmail data for training, again prompting Google to tell us that this was not the case and that the bot was wrong. . Of course, getting Bard wrong is part of a bigger problem.