Tuesday, the 21st of March, Google opened up limited access to its Bard AI chatbot service, the company's official entry into the growing trend of AI chatbot deployments, such as ChatGPT and Microsoft's Bing chatbot. Interested individuals can join a waitlist to engage with Google’s human-like chatbot. Initially, Bard will be available only to US and UK users and some Pixel Superfans.
“We are expanding access to Bard in US + UK with more countries ahead, it’s an early experiment that lets you collaborate with generative AI. Hope Bard sparks more creativity and curiosity, and will get better with feedback,”
- outlines through a tweet Sundar Pichai, the company’s CEO
“You can use Bard to boost your productivity, accelerate your ideas and fuel your curiosity,” - outline Google researchers. “You might ask Bard to give you tips to reach your goal of reading more books this year, explain quantum physics in simple terms or spark your creativity by outlining a blog post.”
Bard is an optimized and lightweight version of LaMDA, a large language model based on the Transformer neural network architecture that Google first introduced in 2017. LaMDA has been trained on massive amounts of public Internet data, comprising trillions of words, and contains up to 137 billion parameters. In this aspect, compared to ChatGPT, whose database hasn't been updated since 2021, Bard's sources of information are substantially more current and relevant.
Bard is fundamentally a language prediction tool that is designed to produce human-like responses. When users ask Bard a question, it generates a response by predicting the most probable next word to use in its answer. In reality, Bard doesn't just generate a single response but provides users with an array of choices. The user is then prompted to select the best starting point according to their perception.
Considering the AI chatbot is trained on publicly available information, Google warns prospective Bard users that its output may exhibit biases and stereotypes. Google has acknowledged that its AI program may make mistakes, and the company's objective is to improve Bard through user feedback. During the initial demo last month, Bard made a mistake by generating incorrect information about the James Webb Space Telescope, which was then included in Google's promotional materials.
“Bard is an experiment,” - outlines the company. “Bard will not always get it right….”
