A former Google employee stated that the company used ChatGPT data taken from the ShareGPT website to train its Bard conversational AI. This is written by The Information.
According to the publication, former Google artificial intelligence engineer Jacob Devlin strongly opposed this approach. After trying to dissuade management, he left the company and joined OpenAI.
Devlin believes that using ShareGPT violates the license agreement. This approach to learning would also make the responses of Bard and ChatGPT too similar, the engineer added.
Another interlocutor of the publication said that Google stopped using the specified data after his warnings. Perhaps the company excluded this part of the training.
In a comment to The Verge, Google spokesman Chris Pappas denied the publication.
Google introduced its version of conversational AI in February 2023. Apparently, the company was in a hurry with the announcement – the presentation took place the day before the release of the “new Bing” from Microsoft.
At the same time, Google did not release Bard to the public. Preliminary testing of the service among a limited number of participants took place a month and a half after the announcement.
Employees criticized Google for a hasty presentation. The company also did not notice that the chatbot made a mistake in advertising materials, which cost the parent holding $ 100 billion in capitalization.
In February, Google management asked employees to test Bard and fix bad answers.
Soon internal chats were filled with memes and AI conversations about potential cuts.
Source: bitcoinlinux.com
