Google’s Bard, an AI chatbot, is receiving some significant upgrades that are expected to enhance its math, logic, and coding abilities. These new upgrades are being pulled from Google’s PaLM language model, which aims to improve the AI’s overall efficiency. As Google continues to collect feedback on Bard, it is committed to introducing weekly updates to the program, with the integration of PaLM’s advancements being just one example.
With these upgrades, Bard should “better understand and respond to your prompts for multi-step word and math problems,” according to Jack Krawczyk, Google’s senior director of product. PaLM’s “chain of thought prompting” ability may have contributed to Bard’s improved performance. The primary goal of PaLM’s integration is to boost the AI’s math and logic abilities, while Bard gains the ability to assist with coding as well.
Bard was already powered by LaMDA, a large language model that helps the AI sound more human-like. PaLM is another language model that was instrumental in Google’s partnership with Everyday Robotics last summer. These two models combined are expected to give Bard greater perplexity and burstiness, making the AI’s responses more varied and complex.
Recently, Google introduced generative AI tools to its Workspace Suite, such as Gmail and Docs, that can help users with writing and editing tasks. Bard was also made available to the general public two weeks ago, following a brief testing period exclusively for Pixel Superfans. Users in the U.S. and the U.K. can sign up to have their name added to the waiting list for testing.