In June, Google introduced Smart Fill, an AI-powered Google Sheets feature that automatically detects patterns, generates formulas, and automatically completes columns. After a month-long beta, Smart Fill is starting general availability for all G Suite users today.
Smart Fill will be introduced after the start of Smart Cleanup as Google wants to equip Google Workspace with more AI functions. The company recently added a tool that allows users to ask natural language questions about data in spreadsheets, such as: B. “Which person has the highest score?”. and “What is the sum of the price per seller?” Google Meet received adaptive noise cancellation earlier this year. Two years ago, Google launched Quick Access, a machine learning product that suggests files relevant to documents that users are editing, including sheets, documents, and slides.
“The idea with Smart Fill is that your table has a number of columns that are filled with names, for example, and that you want to add one or more additional columns,” Weber said. “Smart Fill will look at what you want to do and suggest extracting the first name and filling in the rest of the values. If you agree to the suggestion, Smart Fill will automatically generate a formula that you can use to fill in the entire new column. “
Similar to Smart Compose, Google’s feature that automatically tries to end sentences in Gmail, Smart Fill is designed to detect text transformations and perform tasks like subtracting parts of addresses and getting first and last names. Smart Fill relies on both G Suite’s personal directories and the Knowledge Graph, Google’s knowledge base that contains over a billion facts. For example, it can understand state and state capitals and enter the addresses and phone numbers of contacts.
Smart Fill isn’t the only feature of its kind. Microsoft Excel can also automatically create new columns based on an understanding of what tables are about. However, Weber says that the AI models underlying Smart Fill, trained on large amounts of data from sheets of anonymized and aggregated information, will continue to improve over time as people interact with Smart Fill and either accept or reject changes . In addition, these models are only triggered when they reach a certain confidence threshold, mainly to prevent unwanted or erroneous recommendations from showing up in users’ views.
“We [as a team] Evaluate if we are confident that we can promote discoverability in Sheets in useful ways so that we can actually speed up our users and make them faster and more efficient, ”said Weber. “What we don’t want is a series of proposals constantly getting in the way … It’s not a great result.”