Home / Technology / Pichai: AI like black boxes, don't trust blindly
Pichai: AI like black boxes, don't trust blindly
2 Feb
Summary
- AI can translate Bengali without specific training, Pichai noted.
- Some AI system processes are opaque even to engineers.
- Users should not blindly trust AI, diverse sources are needed.
Google CEO Sundar Pichai has recently drawn attention to the advancing capabilities of artificial intelligence, while simultaneously cautioning users about its inherent limitations. In a resurfaced 2023 interview, Pichai shared an example of an AI model accurately translating Bengali with minimal input, despite not being specifically trained for it. This underscored how AI can develop skills beyond direct programming.
He further explained that some complex AI systems operate as "black boxes," meaning their internal workings are not fully understood, even by their developers. Pichai reiterated in a November 2025 BBC interview that AI models are prone to errors. He stressed the importance of not solely relying on AI and recommended using it alongside other tools and diverse information sources.
