Google's Artificial Intelligence (AI) research labs, DeepMind, is working on a new chatbot that will rival ChatGPT, says the company's CEO Demis Hassabis. In an interview with Wired, Hassabis says that the company is applying techniques it pioneered for AlphaGo, a programme that made headlines for defeating world champions in the board game Go.
Also read | Booking.com to launch trip planner partially powered by tech behind ChatGPT
The team is working on a new Large Language Model (LLM) codenamed Gemini and it will use techniques such as reinforcement learning, where the AI will repeatedly attempt to solve problems and learn from receiving feedback on each attempt.
If that sounds familiar, it's because the technique played a large part in training LLMs that power chatbots such as ChatGPT. It's a technique DeepMind pioneered for AlphaGo.
Also read | NASA working on ChatGPT-like assistant for astronauts
If things go well, the team's experience with reinforcement learning and building systems such as AlphaGo will help them create a model that can potentially be capable of planning and solving problems.
“At a high level you can think of Gemini as combining some of the strengths of AlphaGo-type systems with the amazing language capabilities of the large models,” said Hassabis.
“We also have some new innovations that are going to be pretty interesting.”
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
