Google launches generative AI tool Bard in Arabic

Alphabet-owned Google launched its generative artificial intelligence platform Bard in Arabic on July 13 with distinct features that address Arabic speakers' needs, amid its rivalry with Microsoft-backed ChatGPT.

The conversational AI service can understand questions in 16 Arabic colloquial dialects including Egyptian, Saudi and Emirati, among others, but answers will be given in classical Arabic, Google executives said at a press briefing in Dubai.

“Bard will be available in the Arabic language across all corners of the Arab world as part of its global launch in 40 other languages,” Najeeb Jarrar, regional director of marketing at Google Mena, told reporters.

Bard in Arabic also features a user interface that supports the language's right-to-left script.

In addition, users can input their questions to Bard in both Arabic and another language simultaneously in a single sentence, a sociolinguistic concept known as code-switching.

They can also integrate words “borrowed” from another language and absorbed into Arabic for daily conversations.

In addition, users can ask Bard a question in English and receive an answer in English, which addresses speakers' various levels of proficiency in the Arabic language.

“A big team of Google's engineers and linguist experts worked together over the last months so that the product, Bard, will not just be a translation … but a product that matches our use in the Arabic language,” Marwa Khost, Google's communications manager for Mena, said during the presentation.

Besides giving their feedback on the quality and accuracy of the answers, users can also opt to hear Bard's responses out loud by clicking on the sound icon, a new feature that helps with the tool's accessibility.

The launch of Bard in Arabic comes amid a wider roll-out of the Google tool on Thursday in 43 languages, including Chinese, German, Hindi and Spanish, as the generative AI market competition heats up.

Bard, Google's answer to Microsoft-backed Bing and ChatGPT, is being introduced to 59 new geographies including Europe and Brazil. The latest update makes Bard available in 46 languages and in 239 countries and territories.

The growing adoption of AI technology can help boost global economic growth and raise labour productivity, Goldman Sachs said in a March report.

Widespread AI adoption can eventually boost annual global gross domestic product by 7 per cent in the 10 years after at least half of companies worldwide begin using AI technology.

The US investment bank estimates that AI adoption could raise global annual productivity growth by 1.4 percentage points over a decade, although it expects a more delayed impact on emerging market economies.

Globally, generative AI could cost the world the equivalent of 300 million full-time jobs to automation across major economies, the report indicated.

Bard in Arabic, which was released in the beta or “experiment” version, is currently free to use and there are no plans to monetise the tool, Mr Jarrar said.

Asked about the level of accuracy for Bard's responses in Arabic, he said that while he is personally “comfortable” with its accuracy rate, this remains an “experiment” and there's a large scope for improvement.

“Large language model and generative AI in general still require a person with expertise to look over the results before use,” he said. “Of course, there will be mistakes … it is a machine and it learns from the errors it makes.”

Google's parent company Alphabet lost $100 billion in market value in February after Bard made an error in a promotional video.

The huge improvements in AI over short periods of time will help increase accuracy, users can give feedback about the responses and the answers will continue to improve with research in new tech, Mr Jarrar said.

Asked whether the relatively limited Arabic content online will affect Bard's responses, Mr Jarrar said that Bard can source material from other languages and present it in the users' preferred language. He added that Arabic content has been growing and diversifying exponentially over the last few years.

When asked how frequently Bard will be updated to reflect new information, Mr Jarrar said there are two types of updates: the algorithms and the available information.

“Bard will not be updated in real-time but will be updated much more frequently so that things that have happened last week, will be available to some degree,” he said.

Risks of malicious use

As to the potential risks of using the generative AI tool for malicious purposes, Mr Jarrar said: “Google's approach for any new technology, and especially AI technology like Bard, is that we need to be bold but we need to be responsible.”

“We need to think how to do this service in the right way, not only for users but for society as a whole.”

Google's position as a “pioneer” of the web has meant that the company has developed tools over the years to detect and prevent abuse, he said.