Google Has Entered The Chat
Google looks to become the ultimate AI powered search company
Created on May 12|Last edited on May 12
Comment
At Google's recent I/O event, the company highlighted its renewed focus on artificial intelligence, unveiling a suite of products aimed at streamlining the development process and improving user experiences. Google seems to be placing a strong emphasis on product cohesiveness and is leveraging its models to integrate tightly with existing services like google cloud and Appsheet.
AppSheet
AppSheet is a platform provided by Google Cloud that facilitates no-code development of mobile and web applications. The no-code approach is a significant shift that language models are enabling, and in practice, AppSheet allows users to connect to their data sources, such as Google Sheets, Excel spreadsheets, or SQL databases, and utilize this data to build applications.
Users can design the app's user interface, specify the app's behavior, and define workflows directly within the platform. This integration with proprietary data could be extremely useful for businesses looking to secure valuable data while also building customizable AI-powered tools.
Vertex
Google's new generative AI offerings in Vertex AI aim to facilitate the creation of customizable language models. Google has introduced three new foundation models: Codey, Imagen, and Chirp.
- Codey is a text-to-code model which aids developers by offering real-time code suggestions and generating code based on natural language prompts.
- Imagen, a text-to-image model, allows organizations to generate and customize high-quality images for various business needs.
- The third model, Chirp, is a speech-to-text model intended to foster deeper and more inclusive customer engagement through tasks such as voice control, captioning, and voice assistance.
To further enhance these models, Google introduced tools that enable applying Reinforcement Learning from Human Feedback (RLHF). The embeddings APIs for text and images are also part of the new offerings. Apex seems to be Google’s answer to the OpenAI API, and they’ve added the ability to add custom RLHF, which could be extremely useful for a more customized control of high end LLM’s.
Duet
Google Duet's Code Assistance and Chat Assistance aim to simplify and enhance the cloud development process. Code Assistance, an AI-driven feature, works alongside developers, providing real-time code recommendations, generating full functions and code blocks, and identifying and suggesting fixes for vulnerabilities and errors. It supports multiple programming languages including Go, Java, JavaScript, Python, and SQL. Chat Assistance, on the other hand, acts as a real-time guide for users, answering development or cloud-related queries in natural language. This feature can provide detailed implementation plans for cloud projects and offer advice on architectural or coding best practices, thus reducing the need for extensive research or consultation. It seems as though these assistants will be tightly integrated with many google cloud products, which could make building backend services much easier.
AI Focused
Whether you're a seasoned developer or a business user looking to create a simple app, Google is expanding its product offering to all audiences. In competition against Microsoft, although Google seems to be lacking in terms of model quality, their models are developed in house (whereas Microsoft seems to be outsourcing to OpenAI), and its likely that this could offer more ways for Google to integrate their models within their product offering. It will be interesting to see how the rest of Google’s product lineup evolves as Language models improve in their capacity.
sources:
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.