Google’s new feature, AI Overviews, has been generating wildly incorrect answers to search queries, leading to some strange and even dangerous suggestions like using glue on pizza. According to an interview with Google CEO Sundar Pichai, these errors are an inherent feature of AI large language models (LLMs) and are still an unsolved problem. Despite efforts to fix these bugs, more incorrect results are expected.
Pichai acknowledges the errors but believes that AI Overviews can still be useful, emphasizing the progress that has been made in improving the technology. However, online observers have criticized the feature for generating incorrect information and tarnishing Google’s reputation, which has already faced criticism for providing low-quality search results.
AI consultant Britney Muller highlights the importance of accuracy in AI and suggests that Google should focus on developing AI for more valuable use cases beyond search. Pichai recognizes the challenges ahead for Google and the internet as a whole.
Overall, while AI Overviews have faced criticism for their inaccuracies, Google remains committed to improving the technology and addressing the challenges it presents. The future of AI and its impact on search and other applications continue to evolve, presenting both opportunities and obstacles for companies like Google.
Article Source
https://futurism.com/the-byte/ceo-google-ai-hallucinations
