Google updates Bard Chatbot with ‘Gemini’ AI as it chases ChatGPT

For more than a year, Google has been developing technology that could rival ChatGPT, the sensational chatbot offered by San Francisco artificial intelligence startup OpenAI.

On Wednesday, the tech giant took another step in the ongoing race and released a new version of its own chatbot, Google Bard. The updated bot is available immediately to English speakers in more than 170 territories and countries, including the United States, and is powered by new AI technology called Gemini, which the company has been developing since the beginning of the year.

“This is the beginning of the Gemini era,” Google CEO Sundar Pichai said in an interview. “It’s the realization of the vision we had when we founded Google DeepMind,” the company’s AI lab. He said Google will implement three different versions of the technology across a wide range of products and services in the coming months.

Mr Pichai and Demis Hassabis, who oversee Google DeepMind, said Gemini is more powerful than Google’s previous chatbot technologies, and could generate more accurate responses and come closer to mimicking human reasoning in some situations.

“We are thrilled with Gemini’s performance,” said Dr. Hassabis.

When OpenAI amazed the world late last year with the AI ​​chatbot ChatGPT, Google was caught flat-footed. The tech giant had been developing similar technology for years, but like other tech giants – most notably Meta – was reluctant to release a technology that could generate biased, false or otherwise toxic information.

In March, Google released its own chatbot, Bard, to mediocre reviews. A month later, the company announced that it had merged its two AI labs – Google Brain and DeepMind – bringing together more than 2,000 researchers and engineers. And in May, the company announced at its flagship Google I/O conference that the new Google DeepMind lab had begun developing Gemini.

After founding the Brain lab in 2011, Google acquired DeepMind in 2014 and paid $650 million for the London AI start-up. DeepMind operated largely independently of the Brain Lab and the rest of Google for a decade, even attempting to exit the company in 2017. But as Google struggled to catch up with OpenAI, Mr. Pichai combined the two labs under Dr. Hassabis, a neuroscientist who co-founded DeepMind.

Google has released benchmark test results claiming that the most powerful version of Gemini outperforms OpenAI’s latest technology, GPT-4, in several key areas. It is better at generating computer code than previous Google technologies, Mr. Pichai said, and it can summarize news articles and other text documents more accurately.

Gemini is also designed to analyze images and sounds, but those skills won’t be rolled into the Bard chatbot until a later date.

Google built three versions of Gemini with three different skills. The largest, Ultra, is designed to tackle complex tasks and will debut next year. Pro, the mid-tier offering, will roll out to numerous Google services with the Bard chatbot starting Wednesday. Nano, the smallest version, will power a number of features on the Pixel 8 Pro smartphone from Wednesday, such as summarizing audio recordings and offering suggested text responses in WhatsApp.

Gemini is what scientists call a large language model, or LLM, a complex mathematical system that can learn skills by analyzing vast amounts of data, including digital books, Wikipedia articles and online bulletin boards. By identifying patterns in all that text, an LLM learns to generate text itself. That means it can write theses, generate computer code and even hold a conversation.

With Gemini, Google has also trained the technology on digital images and sounds. It’s what researchers call a “multimodal” system, meaning it can analyze and respond to both images and sounds. If you give him a math problem that includes lines, shapes, and other images, for example, he may answer in much the same way as a high school student.

However, that part of the technology won’t be available to consumers until sometime next year. Google also acknowledged that Gemini, like similar systems, is prone to errors. It can misrepresent facts or even ‘hallucinate’ – make things up.

Google Cloud, which offers AI and computing services to other companies, is eager to offer Gemini to customers as it competes for deals with OpenAI and Microsoft. After OpenAI briefly forced CEO Sam Altman to leave the company in limbo last month, Google Cloud created a migration plan in an attempt to steal its rival’s customers.

Customers can pay Google the same price as their current OpenAI rate and receive cloud credits or discounts.

Google said cloud customers would have access to Gemini Pro – its mid-tier offering – on December 13. Mr Pichai said some outsiders were now testing Gemini Ultra – the most powerful version of the technology.

Although Google has raced to recapture OpenAI’s AI lead over the past year, Mr Pichai said there was plenty of room in the market for all AI providers.

“It’s far from a zero-sum game,” Mr. Pichai said. “We have a sense of excitement about what we are launching. We also realize that we are still in the very early stages as we can see the continued progress we are making.”