Google is set to unveil significant changes to its AI chatbot, previously known as Bard, with a leaked changelog revealing its rebranding as "Google Gemini." This move signifies Google's commitment to integrating its Large Language Model (LLM) Gemini across its range of products and services.
Google Gemini: Unveiling the Evolution
Google Gemini, powered by LLM, is a versatile chatbot capable of generating images and code based on text prompts. Launched last year, Gemini boasts prowess in complex tasks like logical reasoning, coding, nuanced instruction following, and creative collaboration. The three versions—Gemini Nano, Gemini Pro, and Gemini Ultra—offer distinct capabilities.
New Features and Plans for Google Gemini
The leaked changelog hints at upcoming updates scheduled for February 7:
Rebranding to Gemini: Aligning with Google's LLM strategy, the chatbot transitions from Bard to Gemini.
Advanced Tier: A paid 'Advanced' tier, powered by Gemini Ultra, promises enhanced multi-modal capabilities, improved coding support, and deeper exploration and analysis of files and documents.
Canadian Expansion: Gemini will extend its availability to Canada, marking the first country outside the US to access the chatbot since its launch.
Dedicated App: Smartphone users can anticipate a dedicated Gemini app, enabling Google AI usage for learning, writing, and planning. Compatibility with Gmail, Maps, and YouTube is expected, with Android users receiving a separate app and iOS users potentially accessing it through the Google app.
Language Support: Gemini aims to support more languages, including Japanese, Korean, and English worldwide, excluding select European countries. Google plans to expand its language offerings and global reach soon.
These updates, though unconfirmed by Google, reveal the ambitious trajectory of Google Gemini, intending to democratize access to Google AI.
Google Gemini's Image and Code Generation
A multimodal AI model, Google Gemini employs a unique architecture featuring a multimodal encoder and decoder. The encoder transforms diverse data types—text, images, video, audio, and code—into a common language for the decoder. This facilitates the generation of outputs in various modalities based on the encoded inputs and the specific task.
For instance, a text prompt like "a cat wearing a hat" results in Gemini encoding the text and generating an image output. Similarly, coding tasks, such as "write a function to reverse a string in Python," lead to code output generation. Gemini seamlessly navigates between text, images, and code, producing outputs across different modalities.
Gemini stands out by producing images natively, without relying on an intermediate natural language description. This enhances the model's ability to express images and allows it to generate images efficiently with interleaved image and text sequences. The chatbot also excels in coding tasks across multiple programming languages like Python, Java, C++, and HTML, showcasing its prowess in logical reasoning, nuanced instruction following, and creative collaboration.
Accessible through Google Bard, Gemini heralds a new era in AI chatbots, demanding users to have a Google Workspace account with Bard access enabled and be at least 18 years old. The transformative potential of Google Gemini leaves us eager to witness its evolution in the coming weeks.
Top comments (0)