Google’s artificially intelligent chatbots Bard and Baidu’s Ernie Bot are combined with the search engines of both companies. With this, the tech giants are competing with Microsoft, which is integrating OpenAI’s ChatGPT into its search engine.
US tech company Google and its Chinese counterpart Baidu have both revealed that they will be adding a competitor to OpenAI’s ChatGPT to their search engines. Shortly before, it was announced that Microsoft plans to integrate ChatGPT into its search engine Bing. With that, the world’s largest tech companies are now engaged in a public arms race.
“Google has always been known as the dominant search engine, if only for its reach and speed. This could be a way to finally give Bing a market advantage,” says AI researcher Mark Lee from the University of Birmingham in the United Kingdom.
READ ALSO
‘Language rules are more changeable than we think’
Language rules are not set in stone, says Marten van der Meulen. Yesterday’s language rules don’t have to be tomorrow’s.
Bart and Ernie
Google is now testing its AI, called Bard, with a select group of users. Sundar Pichai, the CEO of Google, said in a blog post that his answers will soon appear in Google search results. The chatbot will “turn complex information and multiple perspectives into easy-to-digest chunks.”
Bard is based on LaMDAa language AI that a Google employee claimed last year consciousness developed. That was a claim that met with much skepticism among scientists.
Baidu, also known as the “Chinese Google,” has announced a similar project. Its chatbot Ernie Bot is reportedly now being tested internally. It should be ready next month. Reuters reports that Baidu wants to launch the AI as a standalone chatbot first, and then gradually integrate it into its search engine.
Errors
Large language models like ChatGPT can write poems, scripts, and essays, and perform a wide variety of tasks based on written directions. They can also write concise, direct and tailored answers to queries. This could replace search engines, which still refer users to websites that may contain answers to their questions.
However, a problem with current AI models is that they often provide information that sounds accurate, but is not correct. “ChatGPT is really good at making things up and making up references,” says Lee.
Recently, in a Twitter ad, Bard made a mistake about the James Webb Space Telescope (JWST). The ad shows Bard’s response to the question “What new discoveries from the JWST can I tell my nine-year-old child?” As a third suggestion, Bard says, “JWST took the very first pictures of a planet outside our own solar system.”
Astrophysicist, among others Grant Tremblay of the Harvard-Smithsonian Center for Astrophysics pointed out that this is not true. “I’m sure Bard will be impressive, but for the record, JWST didn’t make ‘the very first image of a planet outside our solar system.’ The first image was made by Chauvin et al. in 2004′, he wrote on Twitter.
‘Ironically, you will get the correct answer if you search the original Google for ‘what is the first image of an exoplanet’. It’s funny that Google, in rolling out this massive multibillion-dollar investment, didn’t fact-check its own website,” says Tremblay.
Not ready
Philosopher Carissa Veliz from the University of Oxford says the flaw, and the way it slipped through the system, shows how dangerous it is to rely on AI models. ‘It perfectly shows the main weakness of statistical systems. These systems are designed to give plausible answers depending on statistical analysis. They are not designed to give truthful answers,” she says.
“We are certainly not ready for what is to come. Companies have a financial interest in being the first to develop or deploy certain systems, and they just rush through it,” says Véliz. “We don’t give society time to talk about it and think about it. They don’t even think it through themselves, as this ad clearly shows.’
In response, a Google spokesperson said: “This highlights the importance of a rigorous testing process. That’s something we’re starting now with our Trusted Tester program. We will combine external feedback with our own internal testing to ensure that Bard’s responses meet a high standard of quality, safety and validity.”