MIT Technology Review September 22, 2022
Melissa Heikkilä

The lab trained a chatbot to learn from human feedback and search the internet for information to support its claims.

The trick to making a good AI-powered chatbot might be to have humans tell it how to behave—and force the model to back up its claims using the internet, according to a new paper by Alphabet-owned AI lab DeepMind.

In a new non-peer-reviewed paper out today, the team unveils Sparrow, an AI chatbot that is trained on DeepMind’s large language model Chinchilla.

Sparrow is designed to talk with humans and answer questions, using a live Google search or information to inform those answers. Based on how useful people find those answers, it’s then...

Today's Sponsors

LEK
ZeOmega

Today's Sponsor

LEK

 
Topics: AI (Artificial Intelligence), Technology
5 tips on AI adoption from a frontline physician champion
US federal agencies must have a chief AI officer under new order
Physicians say AI documentation is winning them over
AI expertise gets more expensive as hospitals bulk up
Prompt Engineering For Healthcare: 11 Tips To Craft Great ChatGPT Prompts - March 2024

Share This Article