DeepMind, a Google research lab, is working on a chatbot that claims to be more advanced than ChatGPT. The project is called Gemini.

The research laboratory DeepMind, which belongs to Google, is developing a new one Large language model for a chatbot. The AI ​​project Gemini needs to get better than ChatGPT.

history class

Let’s go back to 2016, when DeepMind’s AI program AlphaGo defeated a human champion of the board game Go for the first time ever. DeepMind combines the techniques of this system with those of an LLM to further develop Gemini.

Learn while playing

“AlphaGo is based on a technique that DeepMind excels in,”reinforcement learningcalled. With this method, the software learns to make decisions, as is often necessary in a game like Go. Developers can then also give the system feedback on these decisions.

The ultimate intention is to give Gemini the ability to problem solve or plan things.

United Forces

Gemini is a joint venture by DeepMind, which since 2014 has been owned by Google and Brain, the tech giant’s main AI lab. Above all, the experiences of the former with reinforcement learning should give Google an advantage over competitors like ChatGPT.

Currently, Gemini is still in full development and no date has yet been set for when the chatbot can be fully operational.

Demis Hassabis, CEO of DeepMind, is one of the big names in AI who made a statement last month on the dangers of artificial intelligence. Even the godfather of the AI ​​is now convinced of this.