Google Workspace: AI, features, access, demo
After Microsoft, it's Google's turn to integrate AI in its office suite. Workspace users will be able to use intelligent assistant to automatically generate documents, including in Gmail.
Ever since ChatGPT, OpenAI's revolutionary conversational bot, hit the Internet in late 2022, Microsoft and Google have been in a frantic race to release their own AI software. A few days ago, the startup announced the release of GPT-4, the new version of the language model that powers ChatGPT - and which is already built into Bing's chatbot. Google immediately responded by announcing on its blog the integration of artificial intelligence into the services of its Google Workspace suite, including Gmail, Docs, Slides, Sheets, Chat and Meet. "We're now at a pivotal moment in our AI journey. Breakthroughs in generative AI are fundamentally changing how people interact with technology," the Internet giant explained in a post. As a result, users will be able to delegate certain writing or content creation tasks to artificial intelligence.
Google Workspace: what is an AI assistant?
The integration of AI with Google's services should save users valuable time in performing administrative tasks. For example, they will simply have to type in a subject or instruction for the AI to generate a draft, which they can modify and edit as they wish - to write a welcome email to a new employee, for example. Similarly, it will be able to "write, reply, summarize and prioritize" emails in Gmail. It will also be able to summarize a conversation with multiple emails, suggest an appropriate response, or convert notes taken as bullet points into presentable text.
In Sheets - the equivalent of Microsoft Excel - the AI will be able to go "from raw data to insights and analysis through auto-completion, formula generation and contextual categorization." In addition, it will be able to automatically generate images, audio and video in Slides. Finally, it will be able to generate new backgrounds and take notes in Meet, as well as "enable workflows to get things done in Chat." Sounds pretty good to us!
Google Workspace: What's at stake in the AI race against Microsoft?
Not to be outdone, just after Google's announcement, Microsoft unveiled Copilot, an AI that can be used as an assistant to create PowerPoint presentations, write Word documents, summarize Excel tables, and that promises to be truly revolutionary, and will be integrated into its Microsoft 365 suite. It's now war between the two digital giants. However, the speed at which these two AI models were released does raise some concerns, especially regarding the security of users' professional data and the reliability of the technology.
So what's at stake? Mistakes, drifts and "hallucinations" - when AIs start to generate nonsense, but with a lot of confidence -, especially as Microsoft puts pressure on Google, leading it to rush a product to market before it's ready. This is what happened with Bard, which was a real fiasco and lost the company a lot of money. And while errors like hallucinations can be funny when you're testing a chatbot, it less funny in the middle of a serious project, with an unreliable artificial intelligence that can new customers or new products, attribute quotes or ideas quoted in meetings to the wrong people, or make mistakes when summarizing a Sheets table with hundreds of data points. "Sometimes artificial intelligence gets it wrong. Sometimes it regales you with something quirky, and often it needs guidance," Google admits.
In view of the controversies with LaMDA, (Google's language model which has allegedly claimed to be conscious, according to one of its former collaborators) the unreliability of Google Bard, the erratic behaviour of Bing's chatbot or even the drifts of Snapchat's My AI, it's not unreasonable for us to be concerned about this AI race in which tech companies, not wanting to fall behind, release their technologies too quickly, without having put the necessary safeguards. Google is reportedly taking great care with its AI with "thoughtful experimentation and many iterations based on user feedback," in order to put in place "safeguards against abuse, protecting the privacy of user data and respecting customer controls on data governance." As such, the new AI built into the Workspace suite will be tested first by trusted U.S. programmers. No public availability date or price has been released - so will this AI be offered via a paid option or added for free to Google's tools? More information is expected to be released at the firm's annual Google I/O conference on May 10.