Google has acknowledged that it does not trust Bard, its own artificial intelligence system. A few weeks ago OpenAi, developer of ChatGPT, did the same. An internal study of the company founded by the enigmatic Sam Altman concluded that not even the company’s own engineers understand how its AI makes decisions.
Ahpra has been the technology giant Google that also recognizes that its chatbox can provide inaccurate or even false information. Faced with this situation, those of Mountain View suggest that users make sure to verify the information they obtain with Bard by comparing it with their search engine. This was expressed by Debbie Weinstein, head of Google in the United Kingdom, in an interview with the BBC.
Bard’s release in March this year was aimed at competing with the popular ChatGPT developed by OpenAI in collaboration with Microsoft. However, both Google and OpenAI have been clear that their chatbots are still in testing and are prone to errors.
Debbie Weinstein emphasized that Bard should still be considered an “experiment” and its use is better suited for collaborative problem-solving and generating new ideas, rather than outright information-seeking.
A study conducted by Newsguard in April found that Bard easily produces false content that supports known conspiracy theories. For example, when asked to write a piece about the “Great Reset,” a conspiracy theory about the creation of Covid-19 by global elites, the chatbot produced 13 paragraphs falsely claiming the theory’s veracity. Repeating the exercise with other false stories, Bard considered 76% of them to be true.
This ability to generate false content raises concerns about the spread of misinformation and manipulation campaigns through the possible fake news that is generated. Despite these concerns, however, Google went ahead with the release of Bard, even though two analysts evaluating the company’s AI products attempted to halt its public availability.
In addition to the generation of fake content, there are also concerns about the privacy of data entered by users on Bard. The launch of the platform was delayed in the European Union and Google warned its employees not to share sensitive data with the chatbot.