Google LLC is continuing its ongoing journey with generative artificial by bringing the benefits of the emerging technology to its products through Google Cloud for developers and to Workspace for business users.

Generative AI has become a flashpoint due to its capability of producing new content based on simple text descriptions which has led to chatbots that can hold conversations, draft emails, generate code and create artwork. This differs from how traditional AI works, which can only analyze and categorize data.

Today the Google Cloud team unveiled a number of generative AI tools to help developers experimenting with the new technology including support for Vertex AI, the company’s managed machine learning platform, and a generative AI app builder that puts the power of AI in the hands of business users. “This is Vertex AI’s biggest update yet,” Nenshad Bardoliwalla, director of Vertex AI, told reporters in a briefing.

With full support for generative AI in Vertex AI, data science and machine learning teams will have access to build and deploy AI applications at scale using Google’s foundational models. The company said that these models will be initially available for generating text and images, but will be expanded over time for audio and video. Using the system, developers will be able to explore models, fine-tune them with their own data, modify prompts and implement them in their own apps.

The new Generative AI App Builder will allow business users and developers to build new AI experiences rapidly, including bots, chat interfaces, search engines, smart assistants and more. Using the app builder, users will also get direct application programming interfaces to Google’s own models and can use it to jumpstart the creation of their own apps using prebuilt templates.

Google said that starting today it’s opening up access to the Generative AI App Builder and support in Vertex AI to a limited number of testers, which include Toyota, HCA Healthcare, Mayo Clinic, Deutsche Bank, Automation Anywhere and others. Those interested in getting early access can join the Google Cloud Innovators technical community.

In order to make it easy to experiment with Google’s generative AI large language models, Google Cloud is offering the PaLM API, which is designed to give developers quick access for building text generation quickly such as chat, summarization and classification. PaLM, or the Pathways Language Model, is a highly efficient AI for understanding and generating language at a very large scale for a multitude of different tasks.

Alongside the release of PaLM API, the Google Cloud team also announced the release of MakerSuite, a tool that helps developers fine-tune custom models, examine data and build appropriate prompts for creating the proper responses from the AI. Using MakerSuite, developers can adjust the text that they feed the AI initially until they get the response they are looking for and export it as code.

The Google Cloud team has been exploring AI workflows for many years, including building applications with the large language model LaMDA in the AI Test Kitchen and the MUM to Search model. The team said it put what it learned about fragmented workflow development into practice with MakerSuite. Developers can use the tool to fine-tune their models right in the browser, saving lots of time and energy.

MakerSuite is currently in a private preview for select developers and a waitlist is coming soon.

In addition to all of this news, Google is opening up its ecosystem partnership program to more companies looking to use its AI foundational models to get access to Google Cloud. Generative AI platforms such as AI21 Labs Ltd., a startup that builds text generation and comprehension features using AI; Osmo, a startup focused on helping computers detect scents through AI models; and Midjourney Inc., a community-supported AI research lab that produces the popular AI-powered artwork generator are all joining Google Cloud.

Google Cloud Chief Executive Thomas Kurian said in a press briefing that pricing for the various services would be announced soon but didn’t provide a precise timetable.

Generative AI for business users in Google Workspace

Google is also putting the power of generative AI into the hands of Workspace users by introducing the first set of AI-powered writing features for Google Docs and Gmail to a limited number of users.

Many users have already experienced some form of AI in Gmail with Smart Compose, which provides autocomplete suggestions for written phrases and Smart Reply, which offers suggested replies to emails. However, now for some Workspace testers, the full power of generative AI will be rolling out throughout the year.

Using generative AI capabilities, Workspace users will soon be able to draft emails or text by simply typing what they want in a simple sentence and have the AI generate a potential draft based on their style of writing and then fine-tune it based on tone. The same can be done in Docs for proofreading, writing and rewriting in Docs.

Examples of how this technology could be put to use include writing up a work email to a vendor or a supplier rapidly for work and then asking the AI to help formalize the tone in order to make sure that it isn’t too personal-sounding. It could also quickly flesh out bullet points written down quickly during a meeting into a full-fledged summary that uses internal business documents to capture insights for the team.

Future features Google plans to roll out include auto-generated images, audio and video for Slides, producing insights and analysis via auto-completion and formulas generation in Sheets, generating capture notes in Meet, and enabling workflows for getting things done in Chat.

Kurian emphasized data privacy for the new services, saying data won’t be used in training its models and customer data will be isolated to ensure that. “That isolation is what’s giving customers confidence to use this technology while maintaining control of their data,” he said.

With reporting from Robert Hof

Image: Pixabay