Google has added code generation and debugging capabilities to its generative AI offering, dubbed Bard AI, in response to rival offerings from Microsoft’s GitHub Copilot and Amazon CodeWhisperer.
“Starting now, Bard can help with programming and software development tasks, including code generation, debugging, and code explanation,” Paige Bailey, group product manager at Google Research, wrote in a blog post.
Bard’s programming and software development aiding capabilities cover over 20 programming languages including C++, Go, Java, JavaScript, Python, and Typescript, Bailey said. The generative AI is now also capable of assisting developers with writing functions for Google Sheets.
In addition to generating code, Bard can help explain code snippets to developers, the company said. This could be useful for developers who are trying to learn a new programming language by helping them understand the output of a particular block of code.
Analysts and experts claim that the addition of generative AI to low-code and no-code platforms would help developers exploit the code explanation feature the most and boost learning of new programming languages. Some low-code vendors, such as Microsoft and Mendix, claim to be working on adding similar features to their offerings.
Bard can debug even the code snippets it generated
Bard, according to Google, can help debug code, including the sections of code it generated.
“If Bard gives you an error message or code that doesn’t do what you intended, just tell Bard ‘this code didn’t work, please fix it’, and Bard can help you debug it,” Bailey said.
Bard also ensures that it cites the source, in case the generative AI engine quotes “at length” from an existing open source project.
In comparison, Microsoft’s GitHub Copilot doesn’t cite open source projects when copying code from such projects. This has led the company into a class action lawsuit over Copilot AI coding assistant.
Bard still an experiment, warns Google
Google has warned that Bard is still in an experimental phase and may sometimes provide inaccurate, misleading, or false information while presenting it confidently.
“When it comes to coding, Bard may give you working code that doesn’t produce the expected output, or provide you with code that is not optimal or incomplete. Always double-check Bard’s responses and carefully test and review code for errors, bugs, and vulnerabilities before relying on it,” Bailey said.
The new capabilities of Bard, which include the ability to export Python code to Google Colab, come at a time when large technology companies including Microsoft, Amazon, and Google are locked in a battle for supremacy in the field of generative AI.
In the development segment, GitHub Copilot and Amazon CodeWhisperer seem to have the first-mover advantage with an early release of code generation capabilities.
GitHub Copilot, which uses OpenAI’s GPT-3 engine for providing generative AI capabilities, has already announced its next iteration Copilot X and expects to add more advanced generative AI-based capabilities to the software tool.
CopilotX will be able to pull requests, the command line, and docs to answer questions about projects.