CodeWhisperer enters the AI code completion market

At its re: Mars conference on June 23, the American tech behemoth Amazon declared the launch of CodeWhisperer.

Similar to GitHub’s Copilot, CodeWhisperer is an AI pair programming tool that can autocomplete entire functions with just a comment or a few keystrokes. This tool is the most recent addition to the rapidly expanding field of AI code completion solutions.

To train the system, which currently supports Java, JavaScript, and Python, the company used its codebase, billions of lines of open-source code made accessible to the public, publicly available documentation, and publicly usable code on public forums.

Jeff Barr, the chief evangelist of Amazon Web Services (AWS), is quoted in an Amazon Web Services (AWS) blog post as saying: Trained on billions of lines of code and powered by machine learning, CodeWhisperer has the same purpose. CodeWhisperer will increase your productivity whether you are a student, a seasoned professional, or a novice developer.

We are commencing in preview form with support for various IDEs and languages, according to the blog post. As a result, developers can use it right away in their favorite IDEs.

An IDE, or Integrated Development Environment, makes it easier for programmers to create computer programs.

The preview is compatible with code written in Python, Java, and JavaScript, utilizing VS Code, IntelliJ IDEA, PyCharm, WebStorm, as well as AWS Cloud9, according to the post. Support for the AWS Lambda Console is being developed and should be available very soon.

But Amazon pointed out that the algorithm constantly examines the code and comments and even takes into account the programmer’s individual coding preferences and variable names. It will then use this contextual data and the cursor position to produce its own distinctive code snippets.

According to the post, you can either utilize the recommendations exactly as they are or modify and improve them as necessary. We used billions of lines of code from open source repositories, internal Amazon repositories, API documentation, and forums to train CodeWhisperer (and we still do).

The developers who wish to test out Amazon’s new code completion tool can join the waitlist by filling out a request form.

According to the blog post, developers can install the AWS IDE Toolkit, set up the CodeWhisperer performance, and start using the tool after getting a preview access code.

If a developer enters “#See if a number is pr,” CodeWhisperer will propose that the word prime be completed, and if the solution is accepted, the AI-powered bot will then recommend a description and even an entire function, which developers can accept or change.

One of CodeWhisperer’s competitors, GitHub Copilot, recently became generally available. Kite Team Server, which supports over a dozen languages, and Tabnine, which allows developers to train the code completion AI on their own code, are two other contenders.

Kite, IBM’s Project CodeNet, and DeepMind’s Alphacode are three other AI code completion contenders.

Source link