Alphabet's Google is set to relaunch its Gemini AI image generation tool in the coming weeks, following a temporary pause due to inaccuracies in historical depictions generated by the tool.
This move comes as Google DeepMind CEO Demis Hassabis announced plans to reintroduce the Gemini AI image generation tool after addressing the inaccuracies that led to its temporary suspension. The inaccuracies were notably flagged by users on social media, prompting Google to take corrective action.
Recently, the AI tool's results on Prime Minister Narendra Modi came under fire for being “too woke” and “racist.” Google CEO Sundar Pichai also rendered an apology for the “unreliable” AI tool, acknowledging that the company “got it wrong.”
Also Read: What really happened with Google's image generator?
Meanwhile, AI companies are keenly watching Google to set a precedent. Lucas Bonatto, Director of AI and ML at Semantix, a development firm that focuses on big data and artificial intelligence, said, “The different types of biases present in machine learning models represent a serious issue for artificial intelligence applications, especially those operating at scale. In this case, it appears the biases did not cause major issues beyond discomfort and the generation of memes on social networks.”
Google initially offered image generation through its Gemini AI models earlier this month, but some users highlighted that it generated historical images which were sometimes inaccurate. This led to the generative AI tool being described as not “working the way we intended” by Google representatives.
The inaccuracies and subsequent pause in the tool's usage underscore the challenges faced by tech giants in developing AI software that is both innovative and accurate.
The relaunch of the Gemini AI image generation tool is part of Google's broader efforts to compete with industry rivals, particularly OpenAI's ChatGPT.
Also Read: Amidst Google Gemini AI fiasco, IT Minister says saying sorry “not
Read more on tech.hindustantimes.com