Preeti
Google CEO Sundar Pichai admitted that the company made a mistake when racially biased image results were generated by Google Gemini earlier this year. The feature was quickly paused after social media users complained about inaccurate depictions, such as including people of color in images of Nazis and US Founding Fathers. Pichai called the AI tool’s responses “completely unacceptable” and ordered a complete rebuild of the models. He stated that they are retraining the models from the ground up to improve the product before re-releasing it in a few weeks.
Pichai previously defended Gemini’s answers, saying they align with the information found on Google Search and aim to provide trustworthy and high-quality information. He emphasized the importance of surfacing a range of opinions on the web, not just providing a single answer. The company acknowledged that the issue was due to limitations in the training data used for Gemini and even Google co-founder Sergey Brin admitted that the AI chatbot “definitely messed up.”
Overall, Google is working to address the biases in its image generation capabilities and ensure that the product reflects diverse and accurate information. Pichai emphasized the importance of providing a range of opinions and information to users, rather than just a single answer.
Source link
Source link: https://timesofindia.indiatimes.com/technology/tech-news/we-got-it-wrong-google-ceo-sundar-pichai-on-gemini-ai-chatbots-inaccurate-image-generation/amp_articleshow/109958685.cms
GIPHY App Key not set. Please check settings