Google Updates Conversational AI Model Gemini with Larger Context and Multimodal Input

QUICK ZAPS

Google has announced an updated version of its conversational AI model called Gemini 1.5. The new model has some major improvements over the previous version. Gemini 1.5 can now process up to 1 million tokens of context, which is a huge amount of text, like an entire book or movie script. This large context window will allow people to have more natural conversations with the AI by providing more relevant personal context.

In addition, Gemini 1.5 uses a technique called "Mixture of Experts" which makes the model more efficient. It can selectively activate only the most relevant parts of the network for each question, saving computing resources. This also makes the AI faster to respond. The model also now incorporates multiple modes of input, so it can understand images, text and other media types as part of conversations.

Google believes these capabilities will allow the AI to be more helpful for complex tasks like assisting developers. However, the company acknowledges Gemini 1.5 is still in early research. It will initially only be available for businesses and developers to experiment with.

While impressive, the model has limitations like speed which Google is working to improve. This updated version of Gemini shows progress in conversational AI, but also highlights the challenges still remaining for these large language models.