Google has announced a significant expansion of its Bard AI, allowing it to access real-time data from various Google applications such as Docs, Maps, Lens, Flights, Hotels, and YouTube. This enhancement aims to provide more relevant and actionable responses, strengthening the chatbot’s capabilities and transparency.

Jack Krawczyk, the product lead for Bard, noted the rapid evolution of this technology over the past six months and acknowledged that some users remain cautious about AI, either due to a lack of immediate use cases or concerns about its accuracy.

The new iteration of Bard represents a significant leap forward. It not only offers confidence levels in its answers by referencing content from the web but is also willing to admit when it has made a mistake—a pivotal development in AI language models. Feedback from users has played a crucial role in refining Bard, making it more intuitive and imaginative.

Bard can now tackle more complex and extensive prompts, addressing user queries that involve multiple aspects. For instance, it can respond to inquiries like, “It’s my first semester in college, and I want to get involved, but also want to excel academically. Help me understand why balancing participation in school clubs and extracurricular activities while focusing on my studies is important.”

Google has followed the footsteps of OpenAI and Salesforce by allowing Bard to access real-time data from its various apps, allowing users to combine natural language requests. For instance, users can ask Bard to find flights to Puerto Rico and available hotels on Valentine’s Day in 2024 and then follow up with a request for nearby attractions near the chosen hotel.

To enhance transparency, Google explicitly links to the sources it summarizes and introduces a Double Check feature. This feature allows users to independently audit Bard’s responses and search the web for supporting information. Potentially unfounded responses are highlighted in orange, while well-supported statements are highlighted in green.

Additionally, users can opt-in to Bard Extensions, which grant the AI access to their personal Google data, including emails, photos, and calendar entries. This enables Bard to provide specific answers about users’ daily lives, saving them time and effort. Google has assured users that their content from Gmail, Docs, and Drive remains private, not used for ad targeting or training the Bard model.

Initially available to non-enterprise users in English, Google plans to expand these offerings in the future. This enhanced version of Bard is designed to foster more transparent and context-rich communication, harnessing the power of the open web to provide users with accurate and insightful responses.