The “put generative AI in everything” era is kicking off at Google. In addition to the ChatGPT-style features that are eventually coming to Google Search, today Google announced a round of generative AI features for Google Docs and Gmail. Basically, Google plans to eventually, someday, release a text bot that will do all the writing for you. It can reply to emails and make presentations with just a text prompt. As usual for Google and AI, this is not out yet, and the company says it is only “sharing our broader vision” with this blog post.
Just like the rise of Facebook and Google’s hyper-aggressive response with Google+, Google is in a total panic over the rise of ChatGPT and AI-powered text. Just like how Google put social features into every product back in the G+ days, the plan going forward is to build ChatGPT-style generative text into every Google product. Google’s blog post backs this up by framing this announcement as part of a larger plan, saying, “To start, we’re introducing a first set of AI-powered writing features in Docs and Gmail.” So far, the company has also promised to put AI into its health care offerings and opened up API access to a language model, but we’ve yet to see a real consumer product launch.
The company says it’s building a “collaborative AI partner” into Workspace and has a breakdown of the future features:
- draft, reply, summarize, and prioritize your Gmail
- brainstorm, proofread, write, and rewrite in Docs
- bring your creative vision to life with auto-generated images, audio, and video in Slides
- go from raw data to insights and analysis via auto completion, formula generation, and contextual categorization in Sheets
- generate new backgrounds and capture notes in Meet
- enable workflows for getting things done in Chat
Google has a video detailing how this is all going to work and has the bot basically take over the job of a manager, sending team emails and making presentations, all with just a few text prompts.
A “help me write” box in Gmail and Google Docs will let you type in the basic gist of what you want typed, and Google’s generative AI will spit out a block of text based on that prompt. Google’s example has a manager replying to a “sales campaign ideas” email chain by typing “I’m on it” into the “help me write” box. Google’s text bot then spits out an email in the parlance of corporate communication, starting with “Hi Team, We have a lot of great ideas to work with, and I’m excited to see what we can do with them.” Google even makes the managerial call to highlight a single person in the five-person email chain (“Raj” in the example), thanking them (and only them) individually for their contributions to the email. Gmail’s generated text pops up in the reply field, the user hits “send,” the email is done.
The user then tells Gmail to “write a brief based on these emails,” which launches a new Google Doc, titles it appropriately, and dumps all the ideas from the email chain into a nicely formatted Google doc, complete with small icons highlighting each major title. Another prompt says “build a presentation about the campaign,” and the AI launches Google Slides (Google’s version of PowerPoint) with a basic theme template and slides outlining the “goals,” “key deliverables,” and “milestones” of the project. A wild “generate images” sidebar in Slides shows that Google doesn’t just want to do generative text, but also wants to be able to whip up whatever clip art you need for your presentation, presumably using Google’s text-to-image diffusion model, Imagen.
Things really enter the “far off concept” realm when Google’s video kicks over to a copy of Google Meet. During a multi-person meeting, someone tells an AI bot to “capture notes for us,” and in the perfect world of an advertisement video, Google’s AI starts taking meeting notes. This would involve a voice-recognition engine perfectly listening to and deciphering your person-to-person communication (people speak to each other far differently than they do to voice-recognition systems), understanding the content of that conversation, and then generating notes based on the words it thinks it heard. I can’t even get the Google Assistant to reliably stop an alarm clock anymore, so I don’t know about this one. Google’s concept generates a meeting agenda and seemingly checks off items as they get discussed. Then it makes next steps for individual employees and has a rolling list of important meeting notes.
Another example has it generate mail-merge messages in Google Sheets for each client. You get the idea.