Bloomberg: usefulness of Bard chat robot questioned by Google insiders
CTOnews.com, Oct. 12 (Xinhua)-- Google and Discord jointly invited loyal users of their AI chat robot Bard to set up a chat room, Bloomberg reported last night. Among them, Bard product managers, designers and engineers discussed the effectiveness and usefulness of AI in the chat room, but some people began to question it.
Two participants in the Bard community on Discord shared details of the discussion in the chat room between July and October with Bloomberg. Dominik Rabiej, a senior product manager at Bard, said in the discussion that he "mistrusts" the responses generated by the big language model and advises people to use Bard only in areas such as "creativity" and "brainstorming". Rabiej also said that using Bard for programming is "also a good choice" because people inevitably have to verify that the code is valid.
In addition, Rabiej stressed the importance of the new "double check reply" button in Bard this month. It is said that the feature will use orange to highlight "possibly incorrect" content. He also reiterated that Bard didn't really understand the text it got, but responded with more text at the user's prompt. "remember, Bard, like any big model, is generative-it's not looking up or summarizing for you, it's generating text."
The product manager said bluntly in a chat in July that do not trust the output of the large language model unless (users themselves) can verify it independently. "We also want it to live up to your expectations, but it hasn't yet."
Another product manager, Warkentin, also wrote in the discussion, "humanized improvement is essential so that Bard can become a product that suits everyone. Otherwise, users will not have the ability to judge the function of the product, which I think would be a huge mistake."We don't need a product in an ivory tower, but a product that is available to everyone!"
CTOnews.com previously reported that Google added a "shared conversation link" feature when it expanded its deployment of the Bard AI tool in July. Users can export their conversations with Bard as web links and share them with others, allowing others to continue their conversations with Bard.
Gagan Ghotra, an analyst consultant on X platform, discovered the loophole, and Google mistakenly added links shared by some users to search results as an index, resulting in the leakage of sensitive information.
Gagan Ghotra warns users not to share any "private content" when talking to Bard, otherwise the relevant information may be leaked.
