In a day plagued with announcements about Artificial Intelligence, Google has confirmed the launch of test versions of three new Artificial Intelligence features. They are the writing supported by her in different elements of Google Workspace, prototyping and writing code with generative Artificial Intelligence through Palm API and MakerSuiteand several validated models for its Vertex AI service.
The New features for Google Workspace announced are, for now, focused on Docs and Gmail. With them, users will be able to use Google’s Artificial Intelligence to rewrite emails so that they have a specific tone or style. They will also be able to summarize emails, create drafts for messages, reply to those they receive or assign priorities to emails. The Artificial Intelligence that will be integrated into Gmail will be able to detect the context of what you want to write based on a list of highlighted points.
In Docs, they will be able to use these easily accessible AI features to redact documents along various outlines, rewrite documents, correct writing, or use the tool to participate in brainstorming sessions. You can also make summaries of a text.
In addition, Docs can be asked to generate basic documents just by specifying their main subject. For example, you may be asked to write a document to sell a product to your potential buyers, to obtain a basic text about said sale, with its general lines, which can be expanded or modified to be more complete or exact.
The company has confirmed that AI will later be added to other Workspace applications: Sheets, Slides, Meet and Chat. Thus, for example, Slides will have the possibility of automatically generating images, audio and video on the fly, while users are creating their presentations. In this way, the results of the same will be better, with less effort for those who create them.
Sheets will have the ability to go from raw data to insights and analysis with tools for auto-completion, formula building, and contextual categorization. As for Meet, it will allow you to generate new wallpapers and capture notes. Chat will have workflows to streamline tasks.
Of course, for now, the functions will focus, as we have said, on Gmail and Docs, to gradually expand to the rest. Initially they will only be available in closed trial form for the United States, where they will be provided first to trusted testers of the company. Later it will be opened to the rest of the users of the suite, as well as to other countries, so that before the end of 2023 it will be available to all Workspace users.
Palm API and MakerSuite
The rest of Google’s ads are more focused on developers and business users. The first is Palm APIan application development interface designed as a simple entry point for developers to use Google’s larger language models for whatever they want at any given time: from developing content generation systems and chat tools to for bring general purpose analysis systems to life.
For now, Google has made available to developers, through the PaLM API, an efficient model in terms of size and capabilities. But from the company they have assured that other models of various sizes will soon be available.
This API works in conjunction with a tool that Google has called MakerSuite, which can be used to create test code for developers. Basically, it is a tool for prototyping ideas for prompt engineering, a concept applicable to development with AI, which can be translated as rapid engineering or Artificial Intelligence training.
The web-based MakerSuite app will offer easy controls for adding data, iterating on the prompt, and custom tuning the model to fit the specific needs of the ongoing project. Thus, once the model has been adapted to the developer’s preferences, the work with the PaLM API can be integrated into a project. Among them they can be perfectly «applications developed with TensorFlow, Keras, JAX and other open source libraries«.
Both the API and the suite will, in practice, lead to more developers being able to experiment with integrating AI-powered features into their apps, and could lead to a wave of AI-powered experimental projects. For now, both are in the private testing phase, but Google has already assured that it will open a waiting list to use them in the near future.
The price of accessing the PaLM API, as well as MakerSuite, is still unknown, although what Google is looking for with both is that developers can create apps successfully and, in the future, access the VertexAI service in Google Cloud, which It has also managed to have compatibility with generative Artificial Intelligence from now on.
What’s new for Vertex AI
As for the new functions for Vertex AI announced are designed so that companies can use Google technology to create and customize their applications with Artificial Intelligence. They are collected in Generative AI App Builder, a generative AI app development tool that allows its users to preset templates to automatically generate bots, search engines, and even digital assistants. In addition, they will be ready for these elements to be generated in an agile and fast way.
Support for Generative Artificial Intelligence will allow companies to develop and customize these elements in a granular way. As is often the case with these tools, for the moment they are available in a private test phase for some Google customers, although throughout this year they will be made available to everyone who wants to work with them.