We all know that AI isn’t the future. It’s the present. And there will be a moment when it organizes our entire world. That day could be happy… or it could be the end of humanity as we know it.
And at Google, they are preparing for that day with an AI model that aims to engage in conversation and respond to your questions more efficiently and realistically than even ChatGPT itself with its GPT-4 model.
Google’s goal is to defeat GPT-4
Google has officially unveiled its most capable language model to date, Gemini. CEO Sundar Pichai stated that it’s the first of “a new generation of AI models inspired by how people understand and interact with the world.”
Naturally, it’s all quite complex, but Google’s multibillion-dollar investment in AI has birthed a model more flexible than any before it. Let’s break it down.
The system has been developed from scratch as an integrated multimodal AI. Gemini is trained to seamlessly understand and reason across various inputs, which should make it highly capable in handling complex coding requests and even physics-related problems.
Gemini is being developed in three sizes: Nano, Pro, and Ultra. Nano exists on-device, while Pro will be integrated into Google’s chatbot, Bard. The enhanced Bard chatbot will be available in the same 170 countries and territories as the current service.
Reportedly, Gemini Pro outperformed the previous model, initially powering ChatGPT, known as GPT-3.5, in six out of eight benchmark AI tests. However, there haven’t been comparisons yet between OpenAI’s leading chatbot running on GPT-4 and this new contender.
On its part, Gemini Ultra, which won’t be available until at least 2024, achieved better results than any other model, including GPT-4, in certain performance tests. However, this Ultra version requires additional testing before being authorized for release to “select customers, developers, partners, and experts in security and responsibility.”