For more than three months, Google execs have been following news about AI chatbots from Microsoft (MSFT) and AI research firm OpenAI. These projects have also captured the public imagination about the potential of artificial intelligence.
However, on March 21, Google also announced that it will launch Bard, an AI chatbot. Google has taken a relatively cautious approach and has so far only granted partial access to Bard. Google executives said in an interview that as of today, users in the United States and the United Kingdom could be added to the waiting list, and that access would be made available to more users in those countries later. Beyond that, Google plans to expand Bard to more countries and non-English languages.
Google's relatively cautious rollout of Bard is also their first public response to the recent chatbot craze driven by OpenAI and Microsoft, and is intended to prove that Google is capable of offering similar technology. But Google has been relatively cautious compared with its rivals, whose products have been criticized for introducing an unpredictable and sometimes untrustworthy technology.
Even so, the Bard is a significant addition to Google's most lucrative search business. Many in the tech industry believe that Google has more to lose and gain than any other big tech company from artificial intelligence, which could make a range of its products more useful but could also allow other companies to grab Google's huge Internet search business. Chatbots can respond instantly in complete sentences, and users no longer have to scroll through a list of results to find the answer they want, as they would with a search engine.
The Google Bard is currently launched as a standalone web page, with a main page consisting of a question and answer box. Image credit: Google
Google Bard is currently being launched as a standalone web page, completely separate from its search engine, in a tentative attempt to preserve one of the tech industry's most lucrative businesses (search) while launching new AI.
"Google is starting to get into this space because this is where the world is going." Said Adrian Aoun, a former director of special projects at Google. But Aoun, now CEO of healthcare startup Forward, also mentioned that Google's shift to the Bard chatbot could also help shift its business model away from relying on advertising revenue.
In late November, OpenAI unveiled ChatGPT, an online chatbot that can answer all kinds of questions from users, write academic papers, and even speak eloquently about almost any topic. Two months later, Microsoft, the company's major investor and partner, introduced a similar chatbot feature in its Bing Internet search engine, a move that also suggested the technology could transform a market that Google has dominated for more than two decades.
Google has been racing to roll out similar AI products since December. The company has issued a "code red" alert internally for ChatGPT's launch, making AI explicit as a core priority for the company. It also incentivized multiple internal teams, including those specialising in AI security, to collaborate and accelerate the launch of a range of new products.
Industry experts are eager to see how quickly Google can develop new AI technologies, especially given the breakneck speed with which OpenAI and Microsoft have released similar AI products.
"We are in an extraordinary period." "Said Chirag Dekate, a senior analyst at Gartner, a technology research firm. ChatGPT, he said, inspired many new startups, captured the public's imagination and led to greater competition between Google and Microsoft. "Now the market needs have shifted, so Google has to change with them." Decatur added.
Last week, OpenAI tried to improve its game with GPT-4, a newer technology. GPT-4 will also allow other businesses to incorporate the same AI technology that powered ChatGPT into a variety of products, from business software to e-commerce sites.
Bard may have frequently obfuscated facts, sometimes fabricating information without any warning, which many AI researchers have referred to as an "illusion." Image credit: Google
Google has been testing the underlying technology behind the Bard chatbot since 2015, but has so far held back from fully releasing the technology beyond a small group of early testers. After all, like the chatbots offered by OpenAI and Microsoft, it doesn't always generate trustworthy information and can show bias against women and people of color.
"We are well aware that these problems exist. We have to bring it to market responsibly." "At the same time, we can see the excitement in the industry and all the users who have experienced generative AI," said Eli Collins, Google's vice president of research and development.
Mr. Collins and Sissie Hsiao, Google's vice president of product, said in an interview that the company had not yet defined a revenue model for Bard, a chatbot.
Google announced last week that it would add generative AI capabilities to productivity apps such as Google Docs and Google Sheets, but users would have to pay to use them. The underlying technology will also be sold to companies and software developers looking to launch chatbot products or support new apps.
"It's still in the early stages of development." "We are also exploring how to incorporate these experiences into different products," Mr. Hsiao said.
According to The New York Times, Google's latest announcement is the beginning of a plan to roll out more than 20 artificial intelligence products and features, including a feature called Shopping Try-on, And a product that lets you customize a user's background while recording YouTube videos or using Google's Pixel smartphone video recording feature.
The chatbot Bard isn't integrated into Google's search engine, but a standalone web page with a question and answer box. At the bottom of the answers in the question and answer box, there is also a "Click on Google it" button. Users click the button and are taken to the traditional Google search results page on the topic.
Google executives defined the chatbot Bard as a creative tool that helps users compose emails and poems, and offers advice on how to get kids involved in new hobbies and activities. Google wants to look at how people use the technology and refine the chatbot's features based on usage and feedback, executives said. Unlike Google's search engine, though, Bard wasn't primarily designed to provide a reliable source of information.
"We see Bard as a complement to Google search." "We want to be both bold and responsible in using this technology to innovate," Hsiao said.
Like similar chatbots, Bard is generative artificial intelligence based on so-called Large Language models (LLMS). LLM is a technology that can learn skills by analyzing vast amounts of data across the Internet. This means that chatbots can often confuse facts, sometimes making up information without any warning, which many AI researchers also call "hallucinations." Google says it does its best to limit the practice, but admits it may not be entirely effective.
When Google executives demonstrated the capabilities of Bard, a chatbot, on March 20, it refused to answer medical questions because it couldn't access accurate and accurate information. The robot also misquoted sources when answering questions about the American Revolution.
Google added a disclaimer under the Bard chatbot's question and answer box warning users that "Bard may generate inaccurate or objectionable information that does not represent Google's views". In addition, Google provides three different answer options for each question. Through users' choices, Google can further understand the validity and relevance of the answers provided by Bard, which also helps Google to further improve the quality of Bard's answers.
Like Microsoft's Bing chatbot and similar AI chatbots from startups such as You.com and Perplexity, Bard also annotates its responses from time to time so that people can access their sources. And that content is already indexed by Google, so it has access to the latest information posted on the Internet.
This may allow the Bard chatbot to provide a more accurate answer in some cases, but not every time. Even with access to the latest online information, it can still misstate facts and generate incorrect information.
"Large language models are complex and difficult to fully understand and control." "Bard is no exception," Collins says of the underlying technology behind these chatbots.
