Google has unveiled Bard, a new conversational AI chat service that aims to rival the success of ChatGPT and capitalize on the growing demand for intelligent chatbots.
Bard sets itself apart by retrieving information directly from the web, making it a versatile tool capable of coding assistance, math problem-solving, and aiding in writing tasks.
The announcement of Google Bard came on February 6, when Sundar Pichai, CEO of Google and Alphabet, revealed this innovative addition to the company’s AI offerings.
While Bard was an entirely new concept, it drew upon Google’s Language Model for Dialogue Applications (LaMDA), introduced two years prior, to power its launch.
What is the functioning mechanism of Google Bard?
Bard relies on Google’s most advanced large language model (LLM), PaLM 2, an upgraded version of PaLM, which was released in April 2022. PaLM 2 enhances Bard’s efficiency, performance, and problem-solving capabilities, improving its functionality and user experience.
During its initial release, Bard utilized a lightweight model version of LaMDA. This decision ensured lower computing power requirements and scalability to accommodate a larger user base.
LaMDA, built on Google’s Transformer neural network architecture, which was open-sourced in 2017, shares similarities with GPT-3, the language model powering ChatGPT.
Google’s choice to develop and employ its own LLMs, LaMDA and PaLM 2, sets Bard apart from other popular AI chatbots like ChatGPT and Bing Chat, which utilize LLMs from the GPT series.
Which individuals are granted access to Google Bard?
At the recent Google I/O event, it was announced that Bard would no longer have a waitlist and would be accessible to the general public. The previous waitlist, which opened on March 21, 2023, granted limited access to users in the United States and the United Kingdom on a rolling basis.
Bard initially supports Japanese and Korean; however, Google has plans to expand its language support to include an additional 40 languages soon, catering to a more diverse user base.
The controversy surrounding Google Bard
Bard’s launch had controversy. During a demo, Bard provided inaccurate information about the James Webb Space Telescope (JWST), stating that it had taken the first pictures of a planet outside our solar system. Users quickly pointed out the factual inaccuracy, raising concerns about the chatbot’s reliability and accuracy.
A Google spokesperson addressed the issue, emphasizing the importance of a rigorous testing process. They also mentioned initiating the Trusted Tester program to address and rectify such inaccuracies. Bard’s current performance has faced criticism, with users noting its struggle to answer basic questions, longer response times, and the absence of automatic source inclusion.
In comparison to more established competitors like ChatGPT and Bing Chat, Bard has been described by Google CEO Sundar Pichai as a ‘souped-up Civic.’
Prior to Bard’s release, Google’s LaMDA faced scrutiny when a former engineer raised concerns about its potential’ sentience.’ Google dismissed the claim, placing the engineer on administrative leave and eventually parting ways with them.
Google’s transition from LaMDA to PaLM 2 is expected to address several of Bard’s current challenges and improve its performance.
The AI chatbot market continues to expand. Google’s entry with Bard aims to offer a competitive and feature-rich option for users seeking conversational AI assistance.
Alongside their development of AI image generator Imagen and AI music generator MusicLM, Google acknowledges the potential risks associated with these models and strives to address concerns related to creative content, biases in training data, and cultural appropriation.