Amazon Launches CodeWhisperer, an AI Pair Programming Tool Similar to GitHub Copilot

- Advertisement -


At the re:Mars Amazon conference today announced launch of CodeWhisperer, an AI pair programming tool similar to GitHub’s Copilot that can automatically execute entire functions based on just a comment or a few keystrokes. The company trained the system, which currently supports Java, JavaScript, and Python, on billions of lines of publicly available open source code and its own codebase, as well as publicly available documentation and code on public forums.

- Advertisement -

It’s now available in preview as part of AWS IDE Toolkitwhich means developers can immediately use it right in their preferred IDEs, including Visual Studio Code, IntelliJ IDEA, PyCharm, WebStorm, and Amazon’s own AWS 9 cloud. Support for AWS Lambda Console is also coming soon.

- Advertisement -

Ahead of today’s announcement, Vasya Filomin, Amazon’s vice president in charge of AI services, stressed that the company didn’t create this just to offer a copy of Copilot. He noted that with CodeGuruits AI code analyzer and performance profiler, and DevOps Guruof its troubleshooting tool, the company laid the groundwork for today’s launch a few years ago.

Image credits: Amazon

- Advertisement -

I count in technology is in a dot where we thought it was in Correctly time to do this,” Philomin said. “As well as it fits nicely With in Another parts what They have. it was a travel as well as we have simply Performed another parts in another once”.

Internally, Amazon is only testing the service with a small number of developers, mostly to keep the announcement under wraps.

Image credits: Amazon

The company notes that the system constantly reviews your code and comments, and even takes into account your own coding style and variable names. Using this context information – and where your cursor is – it will then generate its own custom code snippets.

It’s worth noting that CodeWhisperer does some things that are different from Copilot. First, while most of the code the system generates is new, every time it generates code that is close to an existing piece in its training data, it notes this and allocates a license to that original feature. It is up to the developer to decide whether to use it or not. This should alleviate some (though perhaps not all) of the copyright issues that may arise when using such a tool.

Image credits: Amazon

Another factor that Philomin emphasized is security. Based on Amazon’s own experience with managing large codebases and debugging things after something went wrong (using a formalized “error correction‘ process), and thanks to its experience with CodeGuru, Codewhisperer will scan the code for potential security issues.

“Security is always important at AWS, so we want to make sure the code we generate is secure,” Philomin said. “Now, obviously, we have generated the code and the developer can change it – and so Codewhisperer can say: start scanning the current source file. It will scan the security system and notify you of any security issues and vulnerabilities either in the generated or modified code.”

Image credits: Amazon

Filomin also emphasized that the team has been working hard to ensure that the code that Codewhisperer generates does not generate any biased code either. However, when this happens, the team has implemented filters to automatically remove this code.

The actual user/developer experience is quite simple. Developers can choose between different code suggestions (the demo I saw usually had at least two options). It can automatically fill in comments and only suggest features based on those comments. A simple example for this would be something #Check if a number is prime.

And while the system will work great for developers outside of the AWS ecosystem, Philomin noted that the team took great care to make it work well for developers who want to use it to access AWS services (think # Create an S3 bucket).

“When we say it has top-notch AWS API support, that’s just additional information. In terms of all our measurements and all our testing, we find it remarkably good and it will be state of the art,” he said. Since developers can now test the service, we’ll probably find out soon how well it works and how it differs from projects like Copilot. Anyway, having more choices in this space is great (and someone is definitely going to hack into a project that will automatically suggest suggestions from multiple pair programming AIs).


Credit: techcrunch.com /

- Advertisement -

Stay on top - Get the daily news in your inbox

DMCA / Correction Notice

Recent Articles

Related Stories

Stay on top - Get the daily news in your inbox