As part of today’s presentation, Google talked about the Bard AI chat bot presented recently, which is expected to become a direct competitor to the similar ChatGPT bot from Open AI. However, at the same time, Google showed a video in which its tool made a mistake when answering a question that everyone noticed.
Image Source: Google
In a short promotional video posted on Twitter, a chatbot billed as a “launching pad for satisfying curiosity” that can help uncover complex topics in simple words was asked: “What new discoveries of the James Webb Space Telescope (JWST) can I tell my 9 year old child?
Bard provided multiple answers listing the discoveries made with the James Webb telescope. And one of them turned out to be wrong: the AI bot reported that the James Webb was used to take the very first pictures of a planet outside the solar system, that is, an exoplanet. This is not true. The first images of the exoplanet were taken by the European Southern Observatory’s Very Large Telescope (VLT) in 2004, which was confirmed by NASA.
Concerns about inaccuracies generated by artificial intelligence systems that are not easy for humans to detect have been expressed before. OpenAI, which developed the ChatGPT AI bot, makes no secret that the technology has its limitations and acknowledges that sometimes AI bots can give plausible-sounding but incorrect or nonsensical answers to questions.
It is reported that OpenAI used all the English-language content of Wikipedia, as well as eight years of web pages extracted from the public Internet, and scanned books in English to train ChatGPT.
It can be assumed that Google used similar sources to train Bard, although the company has not yet disclosed how the AI bot was trained to generate responses and summaries of search results.
If you notice an error, select it with the mouse and press CTRL + ENTER.
