Google’s conversation AI tool Bard is expanding its capabilities to help software developers with programming tasks, including generating code, debugging, and code explanation.
Bard was launched earlier this year as a generative AI experiment to compete with ChatGPT and other language models. Although it did not perform as well as its AI peers in TechCrunch’s own testing on chatbot performance, this latest set of skills could help Bard keep up with ChatGPT and Claude, at least on paper.
Developers can ask Bard to “fix code that didn’t work” and it will help debug the code line by line. It can also translate code from one language to another and explain code snippets, which is a helpful feature for those new to programming. Furthermore, Bard can generate functions for Google Sheets, and users can export Python code to Google Colab.
Despite these new capabilities, Google Research product lead Paige Bailey cautioned in a blog post that Bard is still an early experiment and may provide inaccurate, misleading, or false information while presenting it confidently.
For instance, Bard may give developers working code that is incomplete or doesn’t produce the expected output. Bailey added that despite these challenges, Bard’s new capabilities can help developers in offering new ways to write code, create test cases, or update APIs. If Bard quotes from an existing open source project, it will cite the source.
In summary, Bard’s new capabilities offer software developers an exciting tool for generating, debugging, and translating code snippets, and its expansion into software development languages will undoubtedly be welcomed. However, its early experimental phase means that caution must be exercised when using its features.