HTLS 2025: Can Doctors and Scientists Trust AI?

At the Hindustan Times Leadership Summit (HTLS) 2025, Pushmeet Kohli from Google DeepMind talked about whether doctors and scientists can trust AI.

He said AI is powerful but can make mistakes.

For example, Google's AlphaFold AI can predict protein structures but sometimes gets it wrong.

Kohli emphasized the importance of knowing when AI is accurate and when it's not.

He also mentioned that AI can show uncertainty, like when it's not sure about its predictions.

Kohli assured that AI is improving, but errors can still happen.

He highlighted the need for responsible AI use and introduced SynthID, a tool to detect AI-generated content.

AI is also becoming important in healthcare, helping to improve access and efficiency.