Google has announced a new AI project called Gemini, which it claims will eclipse OpenAI’s GPT-4. The project is being led by Jeff Dean, the head of Google Brain, and Demis Hassabis, the founder of DeepMind.
Gemini is said to be a multimodal AI model, meaning that it can understand and respond to information from different sources, such as text, images, and audio. It is also said to be more powerful and efficient than GPT-4.
Gemini is a set of large language models (LLMs) that combine GPT-4 with training techniques taken from AlphaGo, such as reinforcement learning and tree search, which has the potential to unseat ChatGPT as the most dominant generative AI solution on the planet.
Google and Alphabet CEO Sundar Pichai in a blog post described Gemini as a versatile and efficient model designed for multiple uses, including advanced features like memory and planning. Pichai also mentioned that once it’s made safe through testing, Gemini will come in different sizes and capabilities, similar to PaLM 2.
Experts are skeptical of Google’s claims, and they argue that Google has a history of making bold promises about AI, only to fall short. They also argue that OpenAI is still the leader in the generative AI space.
The sources for this piece include an article in AnalyticsIndiaMag.