ChatGPT has caught the attention of the world since its public release, and while the number of AI chatbots has increased in recent months, none of them have reached the popularity of ChatGPT. DeepMind, the Google-owned research lab, says its next big language model will rival OpenAI’s model, and could even outperform it.
According to single crunch, DeepMind is busy using alphago techniques. AlphaGo is DeepMind’s artificial intelligence system that was able to beat a professional player in the board game Go for the first time. Alphago has been used to create a chatbot called Gemini.
Demis Hassabis, CEO of DeepMind, says that if all goes according to plan, Gemini will be able to program or solve problems and move on to text analysis. According to Hassabis, Gemini combines some of the capabilities of AlphaGo-type systems with the capabilities of large language models: “We also have some new innovations that are going to be very interesting.”
According to DeepMind’s CEO, Gemini uses Reinforcement Learning technology to easily process tasks that today’s language models struggle to do.
Reinforcement learning has led to advances in the field of language models and plays a key role in how systems like ChatGPT respond. DeepMind, which has a long-term experience in the field of reinforcement learning, wants to inject what it has learned into the field of generative artificial intelligence.