In the ongoing AI race of 2023, the landscape is evolving rapidly, challenging the longstanding dominance of ChatGPT. Google has now entered the fray with the launch of Gemini, its latest, most substantial, and most versatile AI model to date.
According to a recent blog post by Google, Gemini AI has been meticulously crafted to be inherently multimodal, a departure from conventional models. The AI is not only pre-trained for various modalities but has undergone further optimization and training to seamlessly process and generate outputs across different mediums. This means that Gemini comprehensively understands and responds to text, images, videos, audio, and even code.
A demonstration of Gemini’s capabilities was evident when Google fed the model a simple video featuring two yarns, prompting the AI to suggest potential creations. Similarly, Gemini efficiently generated code for a dot matrix pattern presented to it. The model’s ability to interpret and translate video content into actionable outputs is particularly noteworthy.
Google didn’t just showcase Gemini’s multimodal prowess but also presented benchmark figures comparing it to OpenAI’s GPT-4 model. Across various areas, including reasoning, mathematics, and coding, Gemini outperformed GPT-4 in the Massive Multitask Language Understanding (MMLU) test, a widely recognized benchmark. This achievement is significant, especially considering ChatGPT’s reputation as a leading GPT model in the AI landscape.
Gemini AI will be available in three distinct models: Ultra, designed for highly complex tasks; Pro, optimized for scalability; and Nano, a compact version. The Ultra model is scheduled for release next year, while the Pro model will power Google Bard, among other services. The Nano model will find its place in the Pixel 8 phone, assisting with reply suggestions and summarizing audio recordings.
Gemini AI has been trained in the English language and is set to launch in over 170 countries. While the ability to analyze images and sounds will be gradually rolled out for Google Bard, Gemini’s arrival signals a new chapter in the competitive landscape of AI.
