Google claims that ChatGPT didn’t train Bard
After some allegations, Google says it never used ChatGPT records to train AI system, Bard
Have you, by any chance, noticed that Google’s AI system called Bard seems to be producing similar results to ChatGPT? Perhaps that’s just how AI works these days, and they’ll all be alike. Or not. According to The Information and a resigned Google AI researcher, the company was using ChatGPT via ShareGPT to train its new system.
As you may know by now, the beginning of Bard didn’t go so well. Now, it’s keen to rope in the division known as DeepMind to do everything it can to beat its competitor. Besides its new Perspectives carousel, it wants to make sure you still search for topics on Chrome instead of ChatGPT.
Thanks to new details from The Information, it seems like Google tried to take a shortcut initially. The sources say that it used training data scraped by ShareGPT from ChatGPT for Bard. That’s why it was able to appear so quickly with so many answers.
Even the former Google AI researcher that resigned, Jacob Devlin, had warned the company that what they were doing was wrong and against ChatGPT terms and conditions. From what anyone can tell, they then scrapped that training data the moment Devlin moved over to OpenAI to assist them with ChatGPT.
NEW: Prominent Google AI researcher resigned after warning Alphabet CEO Sundar Pichai and other senior execs that Bard—Google’s rival to ChatGPT—was *using data from ChatGPT*.
— Amir Efrati (@amir) March 29, 2023
However, Google is vehemently stating that Bard isn’t trained on ShareGPT or ChatGPT. They’re not saying it wasn’t done before, only that it isn’t the case at the moment. It’s still possible that they initially did it, Devlin warned them and resigned, and then they changed the system.
All we have is the word of the sources. For now, Google and DeepMind are going to work like mad to make Bard the best thing since sliced AI, or at least a top competitor of ChatGPT.Advertisement