Google Teaches Generative AI Chatbot Bard to Write and Explain Software Code
Google has augmented its generative AI chatbot Bard to serve as a software coding assistant. Bard can now write, edit, and explain computer code, a feature seeing a rapid rise in demand ever since Microsoft subsidiary brand GitHub launched its Copilot AI pair programmer.
Bard’s programming tools work in nearly two dozen common coding languages like C++, Go, Java, Python, and TypeScript. The AI can translate among the different languages, and Python code can be sent to Google Colab. Google also highlighted Bard’s new functions can help fill in Google Sheets. Bard can also repair and redo code that doesn’t work, whether from an outside source or the code that it previously generated. Asking for assistance doesn’t require technical knowledge; just telling Bard that some code doesn’t work and asking it to fix the problem will prompt debugging.
The same natural language interface enables Bard to explain code to the user, showing what the code does to those who may not understand programming language. Google did caution that Bard is still in its early experimental stages and the code may not work or do what the user wants so it should be double-checked. The company sees a lot of value in the new tools as an expansion of its features.
“Despite these challenges, we believe Bard’s new capabilities can help you by offering new ways to write code, create test cases, or update APIs. If Bard quotes at length from an existing open source project, it will cite the source. In many cases, Bard can also help optimize your code by making it faster or more efficient. Just respond to Bard’s initial output with “Could you please make that code faster?” or ‘Find error handling clauses you might have missed,’” Google Research product lead Paige Bailey explained in a blog post. “Bard is already helping people with everyday tasks, from crafting presentations and writing lesson plans to inventing new recipes or planning a workout routine. With new coding capabilities, we’re excited to apply generative AI to accelerate software development, inspire innovation, and help people solve complex engineering challenges.”
Bailey pointed out that coding help was one of the most frequent feature requests for Bard, which fits with the interest shown in similar services. GitHub Copilot has been steadily widening its abilities since it launched, including with voice controls. It’s been encouraged by a survey of over 2,000 programmers showing that Copilot is making programmers happy and likely to code faster and feel “more fulfilled with their job.” to Copilot. Coding is also a central element for Amazon’s leap into generative AI with AWS. The CodeWhisperer project just became generally available and free to developers. CodeWhisperer serves as a digital pair programmer trained on billions of lines of code from its own database and open-source code and documentation, and can even mark and remove any functions that may be under a software license.