In a blog post by company CEO Sundar Pichai, Bard “draws on information from the web to provide fresh, high-quality responses”. In other words, this AI pulls internet searches to include in its answers to conversation prompts by users. One example included in the post shows an answer to a query asking it to explain the recent discoveries of the James Webb Space Telescope to a 9-year old. This demonstration shows that, at the very least, it works pretty much like the OpenAI ChatGPT, plus the extra benefit of internet searches to provide either better accuracy or more context. Google says that it is first releasing Bard with a lightweight model version of LaMDA, which requires less computing power. The company is also initially only allowing “trusted testers” to, well, test it. The AI will be made available to more people “in the coming weeks”, but no solid date or window was provided beyond that. When that happens, it would probably be fun to see the difference in the way the two different AI chatbots respond to the same query. Going back to the post, Pichai says that Google Searches will soon see AI-powered features that “distill complex information and multiple perspectives into easy-to-digest formats”. But it sounds like the internet search giant wants Bard to work well first before it gets integrated into its main product. And that’s definitely something to aim for, since there’s plenty of instances already of AI going wrong and getting shut down. The most recent example is the AI Seinfeld show called Nothing, Forever, that got banned from Twitch. (Source: Google)
