After Gemini sparked controversy due to its problematic text and image responses, Google CEO Sundar Pichai has responded, admitting that the company made a mistake.
He stated on Tuesday that the situation was “completely unacceptable” and assured that efforts are underway to rectify it. Pichai conveyed in a memo, shared by Semafor, that teams have been diligently working to resolve the issues, with noticeable progress already being made on various prompts.
Gemini’s woke content
For the past week, Google’s AI chatbot Gemini has been embroiled in controversy. It began with the chatbot generating inaccurate historic images, such as depicting the Pope as a woman and Vikings as Black people.
This was followed by a viral query where Gemini failed to definitively answer whether Elon Musk posting memes or Adolf Hitler is worse.
In a communication to Google employees, CEO Sundar Pichai tackled the recent concerns regarding problematic text and image responses within the Gemini application (formerly known as Bard). He acknowledged that some of its responses have offended users and displayed bias, stating unequivocally that such occurrences are completely unacceptable and the company erred.
Yet, Pichai also stood by Gemini, noting that “no AI is flawless.” He emphasized that particularly in this nascent stage of the industry’s growth, perfection is elusive. However, he underscored Google’s commitment to striving for excellence, regardless of the time it may require. Pichai assured that the company will thoroughly review the situation and ensure comprehensive fixes are implemented.
Pichai resolves to solve the issue
Pichai pledged to take several steps to address the issue, including restructuring within the company, updating product guidelines, improving launch processes, conducting more thorough testing, and recommending adjustments to their technology. “We are examining all of these aspects and will enact the necessary modifications,” Pichai affirmed.
“Our commitment to organizing the world’s information and ensuring its universal accessibility and usefulness is unwavering. Throughout our history, we’ve aimed to provide users with helpful, accurate, and impartial information in our products, fostering trust. This principle must guide all our endeavors, including our burgeoning AI products,” Pichai stated in the memo.
While Pichai acknowledges the imperfection of AI, the crux of the issue lies not solely in the inaccuracies but in Google’s premature rollout of its LLM before adequate fine-tuning. As Pichai also conceded, the company must conduct more thorough testing of its products before public release.
