Foreign languages include Arabic, Chinese, German, and Spanish. Google is also launching new features such as the ability to add images in Prompts, listen to Bard’s responses out loud, and adjust Bard’s response to be longer, or shorter.
The new features are helpful if users want to hear the correct pronunciation of a word or listen to a poem or script. Simply enter a prompt and select the sound icon to hear Bard’s answers. Users can also change the tone and style of Bard’s responses to five different options – simple, long, short, professional or casual. This feature is live in English and will expand to new languages soon.
Amar Subramanya, vice president, engineering, Google, told reporters at a select media briefing that as far as limitations of Bard are concerned, “when the model is asked complex questions, from a factual perspective, it can, in some cases, not be accurate”.
“It can also reflect biases from the training data. It could be adversarial to the training model. These are some of the limitations of Bard. We do use answers from search, to keep it up to date,” he said.
He went on to say that Google’s goal with Bard is to generate its own content. “That said, there are instances when the model takes content from another source. In those cases, we clearly provide a citation in the user interface. The user is able to click on the citation and go to the source. In addition, for images, you can go to the thumbnail and go to the source,” Subramanya said.
Discover the stories of your interest
Responding to a question, without directly commenting on the lawsuit filed in the US against OpenAI‘s ChatGPT, he said, “Bard is trained on publicly available information on the web. That’s one of the sources of training for Bard. In addition, we also use data coming from the creator programme, and data coming from people using Bard. All these components are important in creating a model.”Subramanya also said at this point Google’s focus is not on monetisation of Bard.
Google is also adding new ways to pin and rename conversations with Bard in over 40 languages. Now, when users start a conversation, they’ll see options to pin, rename and pick up recent conversations on the sidebar. Google has also made it easier for users to share part or all of their Bard chat with their network. With shareable links, users can share their ideas and creations with others in over 40 languages.
Google is bringing the capabilities of Google Lens into Bard. Whether users want more information about an image or just need help coming up with a caption, they can now upload images with prompts and Bard will analyse the photo to help. This feature is now live in English (US), and will be expanded to more languages soon.
In May, Google moved Bard to PaLM 2, a more capable large language model, which has enabled many of its recent improvements — including advanced math and reasoning skills and coding capabilities. In the past few weeks, coding has already become one of the most popular things people do with Bard.
