4 min Devops

Google tests develop environment with AI help: another GitHub Copilot story?

Google tests develop environment with AI help: another GitHub Copilot story?

Google announces Project IDX, an online place for the development of full-stack web and multi-platform apps. A key element of the environment is the integrated AI assistant, which gets increasingly implemented on other development platforms as well.

Project IDX is a new test project from Google aimed at developers. It allows developers to work in an online development environment and create full-stack web and multiplatform apps.

Support for multiple coding languages is something Google is still working on. But in the meantime, interested developers can already start with JavaScript and Dart. The frameworks of Angular, Flutter, Next.js, React, Svelte and Vue, among others, are also supported.

The development environment of Project IDX is based on Visual Studio Code – Open Source. The environment runs entirely in the cloud and integrates with Firebase Hosting from Google. This allows developers to transfer their code from GitHub.

AI Assistant Codey

The thing that Google mainly wants to point out is the AI integration in the project. Codey is the AI assistant in Project IDX that suggests code, answers questions about code from any given project and performs code actions such as ‘add comments’.

Codey is an AI product from Google to help developers. Behind the assistant is a model based on PaLM 2. “We spend a lot of time writing code and recent developments in AI have created great opportunities to make that time more productive,” Google states in the announcement.

TechCrunch got a chance to test Project IDX before its launch. They report that the AI assistant is not yet on point. Codey cannot directly manipulate code and has trouble recognizing code selection.

‘Hello, meet my future ex-wife’

Among developers, Google test projects do not have a good reputation. After testing, no official product follows, or well-run products are sent to the afterlife without a thorough reason anyway. We notice this in the public reactions to the announcement on the website Hacker News. “The Google name has become an abomination at this point, they could at least try to roll it out as an independent start-up or something to give it a chance,” writes one user. Mainly the word “experimental” causes developers to stand on their hind legs: “It’s Google’s way of introducing someone as ‘Hello, meet my future ex-wife’.”

There are also concerns about the cloud aspect that characterizes Project IDX: “This looks like something Google would do to get code samples to train its AI.” Developers largely trust only one thing and that is Google’s badness: “Likewise, it is safe to assume that any data that can be collected will be sent to the mothership for whatever purpose they see fit.”

Developers have already seen it go awry in GitHub’s similar product. GitHub Copilot was accused of piracy by simply copying code from GitHub’s open-source licenses. The AI model also trains itself based on this public code. Meanwhile, the platform did find a way to deal with these problems. Through AI, developers now receive an alert when code from an open-source license is copied.

Good idea?

More and more companies are introducing AI assistants for developers. AWS has CodeWhisperer that uses contextual hints to generate code. And dates back to last year just like GitHub Copilot, but recent parties also show interest. For example, Stability AI also showed its latest assistant last week: StableCode. That already has a more positive outlook Project IDX due to broader support for programming languages: Python, Go, Java, JavaScript, C, markdown and C++ are supported.

Also read: GitHub Copilot X: productivity aid or a threat to developers?

StableCode has however a completely different target audience in mind. Its focus is on novice programmers while Project IDX focuses on full-stack development. That focus also allows it to differentiate itself from Copilot and CodeWhisperer.

Thus, the project may cater to the needs of a different group of developers, but we cannot ignore the fact that Project IDX does not offer any guarantee in the long-term. Codey was already presented at Google I/O earlier this year, and Project IDX represents the perfect place for Google to showcase (and possibly train) the AI.