Google has updated its artificial intelligence (AI) chatbot ‘Bard’. This introduces multimodal capabilities, equivalent to allowing bards to reply in over 40 languages ​​or reading image input together with text.

Reuters and others reported on the thirteenth (local time) that Google had updated its blog post to service bards in greater than 40 languages, including Arabic, Chinese, German, Hindi, and Spanish, and to introduce a picture input function.

Choi Hyun-jung, director of Google Bard, also confirmed the identical on the ‘Artificial Intelligence Week 2023’ held on the Grand Intercontinental Seoul Parnas in Samseong-dong, Gangnam-gu on the thirteenth.

That is two months after the official release of Bard at Google’s annual developer conference (I/O) on May tenth. On the time, Google announced that it might remove the waiting list for Bard and begin full-scale service, and that it might support Korean and Japanese for the primary time along with English. He also said that a multi-modal function might be introduced soon.

“Sometimes listening to something out loud can make it easier to approach an idea differently,” said Jack Kroczyk, senior product director at Google, in a blog post introducing the “Hearken to Response” feature. This includes 40 languages.

Allows the user to vary the tone and form of the bard’s response to easy, long, short, skilled or casual. This feature is currently only available in English, with plans to expand to other languages ​​in the long run.

As well as, a ‘conversation pinning’ function was added to proceed the conversation with the bard, and a function to export the generated code to many places or share the contents of the conversation to other places was also added.

A standout feature is the power to make use of images for prompts. The ‘Google Lens’ function announced at I/O was introduced in earnest, and pictures could be uploaded with a prompt. This feature is currently only available in English, with plans to expand to latest languages ​​soon.

Meanwhile, what caught the eye of foreign media in the course of the announcement on this present day was whether or to not release in Europe and copyright issues. Google tried to begin the bard service in Europe during last month, however the launch was robotically delayed since it didn’t submit vital documents equivalent to data protection impact assessment to European regulatory authorities.

In response, Kroczyk said he met with the European side and reassured them about issues related to transparency, alternative and control.

Particularly, Amar Subramanya Bad Engineering vp added in a briefing that “users can opt out of information collection.”

That is a difficulty linked to the copyright of coaching data, which is probably the most sensitive issue within the EU’s AI law issue. As well as, Google recently faced a category motion lawsuit from a law firm over this issue.

“The bard is experimental,” Subramanya said. “We wish to be daring, but in addition responsible.”

Reporter Juyoung Lee juyoung09@aitimes.com