Google launches chatbot Bard in Arabic

RIYADH and DUBAI: Google launched its latest generative artificial intelligence experiment, Bard in Arabic, on Thursday, having initially introduced it in English in May this year, to allow Arabic-speaking people to utilize their creative capabilities and increase productivity.

Google is intentionally calling Bard an “AI experiment” — not a chatbot — allowing the company to explore a “new paradigm in computing,” said Najeeb Jarrar, regional director of marketing at Google MENA.

“We’re learning together how large language models can be helpful and how to minimize poor experiences,” he told Arab News.

The Arabic language consists of several dialects, making it a challenge for AI models. Bard, however, is based on Google’s most recent language model, PaLM2, which can understand information in multiple languages.

It is designed to recognize questions in over 16 Arabic dialects, including Egyptian Spoken Arabic and Saudi Arabian Spoken Arabic, and can reply to questions in Modern Standard Arabic, Jarrar said.

It also understands input even if it contains mixed languages such as inserting sentences in Arabic with other languages, along with a user interface that supports right-to-left writing. 

“I have been using Bard since its release in the Middle East, in the English language. My use for it was to summarize some videos and reports,” said Osamah Essam Eddin, a technical content creator.

He explained how he used both Bard and ChatGPT and compared the two. “I use Bard more for search or (to) lookup updates about a piece of information. It is excellent for anything related to searching such as searching for a specific brand, specific feature, and such,” he said.

Currently, Bard is only available for personal use. When asked about how businesses can use Bard, Jarrar said: “As we launch Bard in new languages including Arabic, our focus will primarily focus on users’ experience and how they can benefit more from Bard.”

There is also no news regarding advertising and revenue models for Bard.

It is primarily designed to boost productivity through features like exporting Python code to Replit; sharing Bard chats with friends; and image search.

Google has already integrated products like Lens, Gmail, Docs and Collab into Bard with plans for “further integration,” Jarrar said.

“We are used to thinking of computing (as) narrowing the world’s existing information, and now it’s about applying the information and expanding it into new ways of creation and creativity,” said Jack Krawczyk, senior product director at Google and one of the leads at Bard, during a roundtable earlier this week.

Addressing privacy and misinformation concerns associated with AI, particularly generative AI chatbots, he said that Google is taking a “bold and responsible approach,” which means engaging with privacy regulators before launching.

Image search, for example, is currently only available in English because Google wants to “understand how this new form of creativity operates in a single language” so that it can build systems that essentially “maximize helpfulness and minimize harm” in other languages, Krawczyk said.

“A lot of people talk about the race that’s happening right now in AI and we believe there’s only one race — the race to get it right. And in that race to get it right, we’re taking this responsible approach,” he added.

Arabic was among over 40 languages Bard was launched in and rolled out across Europe on Thursday.