Google has opened limited public access to its Bard AI chatbot service – the company’s official entry into the tech industry’s race to deploy AI chatbots such as ChatGPT and Microsoft’s Bing chatbot.
Anyone can currently to participate in what Google describes as “an early experiment that lets you collaborate with generative AI”, according to a published on 21 March. Such AI chatbots use their language capabilities to understand and answer a wide variety of questions and prompts, with Google suggesting that people can ask Bard for tips on reading more books, explaining quantum physics, planning birthday parties and writing party invitations.
But Google’s blog post also warns that Bard’s generative AI technology can reflect real-world biases and stereotypes in the chatbot responses – and it highlights how such AIs can “provide inaccurate, misleading or false information while presenting it confidently”.
Advertisement
Such AI gaffes were on display when an early Google Bard advertisement included a factual error about the James Webb Space Telescope. An AI-powered Bing chatbot demonstration by Microsoft also included .
“We should be concerned that the current hype around generative AI is prompting these companies to push their products into development really rapidly,” says at the AI Now Institute, a research centre in New York. “Now the companies are experimenting in the wild with them, versus really taking the time to mitigate potential risks prior to deployment.”
Google has shown some signs of approaching AI chatbots more cautiously than its rivals. The company has limited the “officially supported” uses of Bard by describing it as being unable to help humans write computer code just yet. That stands in contrast to the company OpenAI openly touting its AI chatbot’s code-writing capabilities.
A recent Google also showed Bard refusing to answer a medical question, given the significant consequences of answering such questions. Google researchers have already demonstrated an AI capable of accurately answering medical and health questions from medical licensing exams and search engine queries.
“I think it’s really notable that Google has had systems like Bard in development for a much longer time – they’ve been working on this system since 2015,” says Myers West. “And the fact that they did not push it to deployment earlier should be a signal about the potential risks associated with these technologies.”
Google’s Bard debut comes less than four months after the company OpenAI launched its AI chatbot known as ChatGPT on 30 November 2022. Google rival Microsoft capitalised on early public enthusiasm for ChatGPT by incorporating OpenAI’s underlying AI technology into a chatbot function for its Bing search engine in early February 2023.
Despite still having just a single-digit share of the search-engine market, Microsoft’s in early March by riding the wave of AI chatbot interest. That number included more than 1 million Bing preview users who signed up to check out the AI-powered chatbot function.
