Bard, the AI chatbot from Google has begun to roll out, but only to a select group of users who must be at least 18 years old. Unlike its viral rival ChatGPT, it has a “Google it” button that accesses search and can access current information from the internet. Additionally, it cites Wikipedia and other fact-checking websites.
Google, however, forewarned that Bard would have limitations and that it might disseminate false information and exhibit bias. This, they say is due to the fact that it “learns” from information that comes from the real world, where those biases are still present. As a result, it is possible for stereotypes and false information to appear in its responses.
According to OpenAI, the company that created ChatGPT, it had more than one million users within a week of its November 2022 launch. Microsoft has made significant investments in it and last month added the product to its Bing search engine. It has also made plans to integrate the technology into its office suite of programs, which includes Word, Excel, and Powerpoint.
With its version, Bard, which initially debuts in the US and UK, Google has been a slower and more cautious runner in the generative AI race. To use it, users must first sign up. Like ChatGPT, it will be able to imitate other writers’ writing styles, but it will be unable to express opinions or adopt a persona.
Bard is a descendant of Google’s earlier Lamda language model, which was never fully made available to the general public. When one of the engineers who worked on it asserted that its responses were so convincing that he believed it to be able to perceive and feel things, it did, however, gain a lot of attention. He was fired after Google denied the accusations.
Bard has access to up-to-date information, whereas ChatGPT’s knowledge base only goes as far as 2021 and cannot, for example, respond to queries about the recent elections in Nigeria, or about Tanzania detecting Marburg disease today.
Bard has filters to stop it from sharing harmful, illegal, sexually explicit, or personally identifiable information, and it is programmed not to respond to offensive prompts, but “like any method these guardrails will occasionally fail,” said Zoubin Ghahramani, vice president of Google Research.
Despite the excitement surrounding this type of technology, there are also horror stories about some of the more disturbing things that ChatGPT has been forced to do. Additionally, there are worries that these potent tools, which are still in their infancy, may ultimately pose a serious threat to a wide variety of jobs.
There is also the theory that chatbots may eventually completely replace the lucrative internet search industry, which is especially pertinent to Google. Why struggle through pages of links from search results when you can just get one well-written response? Google can’t afford to withdraw from the competition.
According to Reuters, Google claims that it will closely watch Bard to ensure that it abides by its own “AI principles,” which prohibit the development or reinforcement of bias.