Google’s Bard chatbot improves its math and programming skills
Google’s artificial intelligence chatbot Bard is gradually improving its skills in tasks related to logic and reasoning. This is reported by TechCrunch with a link to information from the tech giant.
The company noted that thanks to a technique called “hidden code execution”, Bard is improving specifically in the field of mathematics and coding.
Google explained that large language models (LLMs) like Bard are essentially prediction engines. When given a prompt, they generate an answer by predicting which words are likely to come next in the sentence. This makes them good email and essay writers, but somewhat error-prone software developers.
In an effort to address the coding and math shortcomings of generic LLMs, Google developed hidden code execution that allows Bard to write and execute its own code. The latest version of Bard identifies clues that could benefit from the logic code, writes the code, tests it, and uses the result to generate a supposedly more accurate answer.
Google claims the new Bard’s responses to “computational” verbal and math tasks have been improved by 30% compared to the previous version.
“Even with these improvements, Bard won’t always get it right — for example, Bard might not generate code to help the prompt response, the code it generates might be wrong or Bard may not include the executed code in its response,” the company noted.
It was previously reported that the Google Bard chatbot will be available in 180 countries, will receive the new PaLM 2 language model, dark theme and visual search.