Google recently announced the development of its own AI chatbot, “Bard,” which was to be released shortly to compete with Bing AI search engine powered with OpenAI’s ChatGPT. Nevertheless, the launch did not occur, resulting in an internal reaction inside the organisation. Despite this setback, it appears like Google is still working on Bard and is providing early access to a restricted number of individuals.
The initial release of Bard will be limited and available only to Google’s devoted followers in the United Kingdom and the United States. The chatbot is based on Google’s own LaMDA (Language Model for Dialogue Applications). It uses a lightened and optimized version of the model. According to Google, the chatbot is an early experiment in engaging with generative AI, and Google is eager for user feedback to help enhance the technology.
Google also admits that huge language models like Bard and ChatGPT are imperfect and can make errors. As a result, feedback from a diverse spectrum of specialists and consumers is critical to improving the technology.
Google VP of Products Sissie Hsiao and Google Research VP Eli Collins have stated that it can help users be more productive, accelerate their ideas, and spark their curiosity. Users can interact with Bard by asking questions and refining their answers with follow-up queries.
One area in which Bard excels is in answering “NORA” queries, which have no single correct answer. However, like other generative AI chatbots, accuracy, sources, ethical barriers, and other issues will need to be addressed.
Google personnel have been testing Bard and offering comments to train the AI for the past month. This method includes fixing incorrect answers and enhancing the chatbot’s response accuracy. While large-scale language models have advanced significantly in recent years, they are still far from flawless and require human involvement to increase their accuracy.
Upon opening Bard, users will be presented with a blank chat box and a disclaimer that says Bard may display inaccurate or offensive information that does not represent the views of Google. Users can enter any question they wish into the text field, and Bard will read and display the answers. Users can rate the answer with a thumbs up or thumbs down, restart the conversation, or switch to Google’s search engine by clicking the “Google It” button.
Unlike Microsoft’s Bing chatbot, Bard has no footnotes with web sources, making it difficult to confirm the correctness of the answer. However, users can click “View other drafts” in the upper right to see more answers to the same query if they are unsatisfied with Bard’s response.
It’s worth noting that OpenAI’s newly released GPT-4 is believed to be more accurate and difficult to fool than prior models. Nonetheless, it will be interesting to see how it compares to Bard when it is out.
While it is unclear when Google will expose Bard to the general public, the fact that they provide early access to a restricted number of people shows that the project is moving forward. It’ll be interesting to see how Bard compares to other AI chatbots on the market and how it affects the future of conversational AI.