Google Bard continues its quiet evolution with the regular addition of improvements and features. This time, in addition to getting better at math and computer programming, Google’s chatbot manages to generate tables that can be directly integrated into Sheets.
Google is gradually making improvements to Bard, its artificial intelligence chatbot. After the integration of Adobe Firefly, a better summary of information, the display of images and the taking into account of the localization, a new functionality comes to make its appearance: the creation of tables which are integrated in one click in Sheets.
Google Bard can generate tables for use in Sheets
The main improvement brought by Bard from a usage point of view is its ability to create tables. In fact, from the start, the chatbot has been able to do this. Less than an improvement of the model, it is rather an improvement of its integration which was carried out. Asking Bard to create an array of data, the chatbot runs.
But where it goes further is that just below the table in question, there is a button labeled “Export to Sheets“. Clicking on it creates a Google Sheets document on their Google account. In one click, you can access it and see its dataset in the spreadsheet.
In addition to the spreadsheet, Google Bard adds textual explanations below. In the tested example, we asked the tool to create a table of expected technical characteristics for the Samsung Galaxy Z Fold 5. Unfortunately, no source was mentioned. In the future, we can very well imagine a direct integration of Bard into Sheets, and this is probably what Google is aiming for.
A more reasoned Bard, better at mathematics and computer programming
Another change implemented by Google: Bard “detects computation requests and executes code in the background, making Bard more capable in math tasks, coding questions, and string manipulation“, can we read on the page of updates of the chatbot. Google specifies that its tool cannot solve questions of this type every time.
More generally, Bard has seen his reasoning and logic skills improved, Google says in a June 7 blog post. If Bard relies on a large language model (LLM), for tasks that require logic, we can not only resort to that. Simply because an LLM only predicts the continuation of a sentence and cannot reason logically.
This is why Bard runs another program in parallel while generating the responses. According to Google, this would improve the accuracy of answers by about 30% on “count-based word and math problems“.
The Watt Else newsletter is THE unmissable event dedicated to the mobility of the future. Register here!