Google has been making headlines in the artificial intelligence space recently following the launch of its ChatGPT competitor Bard back in February. While the bot’s initial demo didn’t go quite as planned, the global tech giant has remained hard at work developing further innovations leveraging AI tech.
Google Workspace AI
Many of us might be familiar with the sentence-finishing suggestions that appear while composing a draft in Gmail or the quick reply messages offered to us upon receiving an email. In fact, more than 3 billion people have made use of Google Workspace’s AI-powered features, Thomas Kurian, CEO of Google Cloud, said in a company blog post.
Including features like Smart Compose and Smart Reply, Google’s productivity suite has come to provide a range of AI-based functionalities. Users can generate summaries for Google Docs, improve video quality in meetings and automatically block email scams thanks to Google’s AI innovations.
Now, Google wants to further its offerings by augmenting its existing features with more generative AI tools across all its Workspace apps. The product suite consists of Gmail, Docs, Slides, Sheets, Meet and Chat. To start, it’s introducing the first set of AI-powered writing features in Docs and Gmail this month.
Google AI will help users draft, reply, summarize and prioritize emails in Gmail. On Docs, AI will help people get started writing on a topic of their interest. This Docs functionality will also help users proofread, write and rewrite documents. These features will launch with U.S.-based testers working in English, according to the company.
More capabilities in Google’s pipeline include auto-generating images, audio and video in Slides presentations, auto-completing spreadsheet entries and generating formulas in Sheets, taking notes in Meet and using productive workflows in Chat.
Throughout the year, these remaining features will be released to testers on a rolling basis before being refined and made publicly available in more countries and languages, Google said in a statement.
Google PaLM API
Alongside its array of AI products coming to Google’s Workspace apps, the company is also ramping up AI tools for its Cloud products.
The tech giant is putting its recent research on large language models, or LLMs, toward helping developers incorporate AI features into their products. LLMs, like Google’s dialogue-trained conversation model LaMDA, are AI tools that can understand and generate language.
PaLM API aims to help developers experiment with Google’s LLMs in a safe, easy and scalable way. Currently available in a private preview program, PaLM lets developers access models designed for use cases like content generation and chat forums. Additionally, developers can access general-purpose models that are optimized for tasks like summarizing and classifying information.
Google announced a version of PaLM geared toward the medical domain last year. The Med-PaLM model is trained to answer medical licensing-style questions. The company’s recent iteration of this tool, Med-PaLM 2, answered questions with 18 percent higher accuracy than its predecessor, Google said in a statement.
While Google is pursuing an abundance of new research in the health AI sector, the company said its PaLM API offering aims to impact various industries. The offering is available through Google Cloud alongside Google’s other AI models. PaLM API also features a tool called MakerSuite, a prototyping environment for generative AI applications.
MakerSuite works to streamline generative AI development workflows. The tool allows developers to test and improve command prompts, as well as export their prompts as code in languages they’re familiar with, such as Python and Node.js.
The offering also provides tooling to tune custom AI models and generate additional data to better inform developers’ AI systems, Google explained in a blog post. The platform is also supported by Google’s suite of safety tools.