The accuracy issue no one’s escaping
Google CEO Sundar Pichai says people shouldn’t fully rely on AI tools for accurate information.
Speaking to the BBC, he warned that today’s AI models still make mistakes and should be used with other sources, not on their own.
He said this is why products like Google Search still matter, as they’re built to give more reliable answers.
But some experts argue that tech companies shouldn’t expect users to fact-check AI; they should focus on making the systems more dependable.
Google has faced its own issues, especially when AI Overviews first launched and produced several inaccurate responses.
Researchers also found that major chatbots, including Gemini, ChatGPT, Copilot and Perplexity, still misrepresent news around 45% of the time.
The interview comes as Google launches Gemini 3.0, its latest consumer model designed to improve reasoning, search and multimodal input.
What to take away:
AI tools still make mistakes, so they shouldn’t be your only source.
Gemini 3.0 is Google’s push to rebuild trust and compete with other AI models.
Research shows chatbots still struggle with factual accuracy.
Google’s ‘please double-check us’ moment
Google is calling this a “new phase” in its AI shift, and a move to stay competitive with ChatGPT.
Pichai said the company is trying to balance speed with safety, and has increased investment in AI security.
This includes open-sourced tools to detect AI-generated images.
He also commented on concerns raised by Elon Musk years ago, saying no single company should control all AI, and that the current market has many players.
Gemini trying to win back the crowd like a politician after a scandal.- MG