OpenAI and Salesforce have already integrated their standalone chatbots into larger, more comprehensive machine learning platforms that cover various aspects of their businesses. Recently, Google also announced that its Bard AI is undergoing a similar transformation. Bard AI has been enhanced to access real-time data from various Google applications, including Docs, Maps, Lens, Flights, Hotels, and YouTube. Additionally, it can tap into users’ stored personal data to provide more relevant and actionable responses.
Jack Krawczyk, the product lead for Bard, expressed his enthusiasm for being part of the team since Bard’s inception, noting that it has been six months since Bard was introduced to the world.
Despite the rapid spread of AI technology, many users still have reservations, either because they don’t see an immediate use-case for it or because they are concerned about the accuracy of the information it provides. Google’s Bard AI aims to address these concerns and build public trust by increasing transparency and providing more detailed explanations for its responses.
The latest version of Bard is designed to be more transparent and willing to acknowledge mistakes. It not only explains its confidence in an answer by referencing content from the web but also admits when it has made an error. This level of transparency is seen as a critical step in gaining user trust.
Users have provided feedback over the past six months, helping Google iterate and improve Bard. The chatbot can now handle more complex prompts and access real-time data from Google’s other apps, such as Maps, YouTube, Hotels, and Flights. Users can mix and match these API requests using natural language queries.
To enhance transparency, Google is explicitly linking to the sites it summarizes and introducing a Double Check feature. This feature allows users to independently audit Bard’s responses and search the web for supporting information. Responses with contradictory evidence are highlighted in orange, while well-supported statements are highlighted in green.
Overall, these enhancements aim to make AI more trustworthy and useful to users in a wide range of scenarios.
Google is introducing Bard Extensions, a feature that allows the AI to access users’ personal Google data, such as emails, photos, and calendar entries. This enables Bard to provide specific answers about users’ daily lives and perform tasks like searching for important dates in Gmail or summarizing email discussions. Users can also draft documents, like cover letters, based on the information available in their personal data.
To address privacy concerns, Google has assured users that their content from Gmail, Docs, and Drive is not accessed by human reviewers, used for advertising, or used to train the Bard model. Users have full control over opting in or out of the system and can grant or deny access to specific files. Initially, this feature is available to non-enterprise users in English, with plans for future expansion.
Google believes that access to personal data will help Bard provide more contextually relevant responses and enhance the user experience by offering deeper insights and assistance.