Google’s Gemini Gets an Upgrade
The launch of ChatGPT sparked an innovation race within the AI sector, leading giants like Google and Microsoft to unveil their AI models. These developments highlighted AI’s versatility, showcasing applications in diverse fields such as content creation and even website development for Bitcoin casinos. Less than a year after the debut of Gemini, Google’s premier AI is already set for an enhancement.
What is Gemini?
Created by Google DeepMind, Gemini is a groundbreaking AI model that stands as a testament to extensive research and teamwork. As a multimodal AI, Gemini can effortlessly process, interact with, and synthesize various types of data. Like with the releases from other AI companies like OpenAI, Gemini was designed to perform across a broad spectrum of devices, from smartphones to data centers. Gemini’s first iteration is the Gemini 1.0, but it’s now available in three distinct versions:
- Gemini Nano: The most effective model for tasks to be conducted on one device
- Gemini Pro: For users looking for something that can handle a wide range of tasks
- Gemini Ultra: Best for handling much more complex tasks. When this model was introduced and tested, it outperformed human experts on Massive Multitask Language Understanding (MMLU)
Features of Gemini
When Gemini was introduced, it offered a few features that made it better than AI models of the time. The features include:
- Sophisticated reasoning: The multimodal reasoning capabilities of Gemini make it capable of extracting insights from large amounts of data. So, this AI can read thousands of documents and provide helpful information for users in various industries.
- Advanced coding: Google ensured their AI model could generate high-quality code in popular programming languages like Go, Java, Python, and C++.
- Understanding audio, text, and images: This AI model can recognize audio, text, and image data, which makes it very useful in understanding complex subjects like physics and math.
What Is the Upgrade to Google’s Gemini?
The upgrade to Gemini is only coming to one version of this AI model—Gemini Pro. Gemini Pro 1.5 launched on February 15, 2024, with increased capabilities in processing audio, image, and video data.
Demis Hassabis, the CEO of Google Deepmind, compares Gemini Pro 1.5 input capacity to a person’s working memory. During the release of this upgrade, a demo depicted this AI model analyzing a 402-page transcript PDF of the Apollo II communications. It accurately answered all questions, including detecting humorous moments during the communications.
Gemini Pro 1.5 can process an hour of video, 30,000 lines of code, 11 hours of audio, and 700,000 words. This makes it better than the current AI models, including Open AI’s GPT 4.
While Google has kept a lid on how they upgraded their AI model so much, they’ve proposed some uses for it. One of them is using Gemini Pro 1.5 to compile important takeaways from Discord discussions, which can number in the thousands.
Conclusion
The Gemini Pro 1.5 upgrade has a bigger memory for processing all text, video, and audio information. This innovation uses a unique Google research technique called a ‘mixture of experts’ to offer better performance without compromising computing power. Gemini Pro 1.5 will be provided to developers through AI Studio and Google’s Vertex AI cloud platform API. Google hopes the new upgrade will make it easier for developers to build apps on Gemini.
