Google Cloud * Gemini AI
Do you use AI at work? It’s a question that more and more of us are being asked all the time. With the approaching ubiquity of artificial intelligence-powered software assistance now appearing everywhere from our core use of the browser to the supermarket checkout kiosk, it’s starting to become hard to say no to this increasingly germane question.
Ask a journalist or author whether they use AI tools daily and the reaction is often something of a revulsion i.e. how could we possibly use AI tools to write, when what we do is write? But ask an arguably more highly qualified professional such as a physician’s assistant in a doctor’s surgery and you’re more likely (soon if not already) to get a yes. The application of AI in the medical field is ripe for exploitation as pattern recognition techniques start to help support diagnosis examinations and far more besides.
So then, how do software application developers feel about embracing AI tools to help them cut code at the command line and across the integrated development environments that they spend their lives inside? After all, if members of the developer community are the ones responsible for building these tools in the first place, surely they should be eating their own dogfood and applying intelligent code automation functions on the keyboard, right?
More Dogfood, Dude?
The latest research from Google Cloud suggests that as many as three-quarters of programmers use at least some AI tools to help execute and underpin their daily responsibilities. With worldwide population of developers forecasted by analyst house IDC to grow to 57.8 million by 2028 (it could be 58 million, but what’s 200,000 software engineers between friends?) the company thinks that AI tools should be available to all, whether they can pay for them or not. This is why we’re now seeing the arrival (in public preview) of Google Cloud Gemini Code Assist for individuals, which is essentially a free version of the Gemini Code Assist AI-coding assistant.
“Gemini Code Assist for individuals is globally available, powered by Gemini 2.0 and now optimized for coding. We fine-tuned the Gemini 2.0 model for developers by analyzing and validating thousands of real-world coding use cases. As a result, the quality of AI-generated recommendations in Gemini Code Assist is better than ever before and ready to address the myriad of daily challenges developers face, whether they’re hobbyists or a start-up developer,” said Ryan J. Salva, senior director of product management at Google Cloud.
“AI has become a central part of the way all large organizations produce software. We looked about and saw there has been a growing dichotomy of the haves and have nots… so large enterprises are willing to procure licenses for their teams to use AI, but students, hobbyists, freelancers and small businesses largely were falling behind. With this release, what we really want to do is continue a long tradition at Google of providing technology to people around the world for free. Our goal is to make it so it doesn’t matter who your employer is, what part of the world you’re in, what your income is… none of that should matter and now it doesn’t,” added Salva.
This free iteration of Gemini Code Assist comes with what its makers call a “generous” usage limit as measured by code completions per month. Where many free coding assistants offer somewhere around 2,000 code completions per month, Google Cloud is using its breadth and backbone to offer developers what the company ranks as “practically unlimited capacity” with up to 180,000 code completions per month with Gemini Code Assist.
It supports all programming languages in the public domain and also has a fundamentally important code review process element. Alongside the arrival of the Gemini Code Assist AI-coding assistant sits the public preview of Gemini Code Assist for GitHub, a route to AI-powered code reviews for both public and private repositories. Here we see developers get a helping hand that can detect stylistic issues and bugs and suggest code changes and fixes. They can offload basic reviews to an AI agent that can help make code repositories more maintainable and improve quality, allowing developers to focus on more complex tasks.
Inside An IDE, Ideally
As already suggested, working software developers spend a lot of their time coding in integrated development environments, usually known as IDEs.
“With the new, free version of Gemini Code Assist in Visual Studio Code and JetBrains IDEs, individual developers now have the same code completion, generation and chat capabilities that we’ve offered businesses for over a year, and that is already available for free in Firebase [Gemini in Firebase is an AI-powered collaborative assistant that reduces debugging time] and Android Studio [the official IDE for Android development]. Now, anyone can more conveniently learn, create code snippets, debug and modify their existing applications without needing to toggle between different windows for help or to copy and paste information from disconnected sources,” said Salva, in a Google technical blog detailing this technology’s arrival.
A chat feature enables programmers and developers to interact with the tool itself so that they can concentrate on application logic and user functionality, otherwise known as the “creative part” of software application development. This leaves the necessary, but repetitive steps (such as writing comments or structuring up automated tests that have been specified as a result of a formalized requirements process) to Gemini.
Spanish To Swedish, Serbian To Swahili
Developers can use natural language in a variety of languages (yes, don’t panic, Lithuanian, Swahili and more are supported) in Gemini Code Assist to generate, explain and improve code. For example, says Salva, a website developer might use a a prompt such as, “Build me a simple HTML form with fields for name, email and message, then then add a ‘submit’ button.” Equally, a user that wants to automate workplace tasks might ask Gemini to, “Write a script that sends a daily email with the latest weather forecasts,” and so on.
“Different developer teams may also have different best practices, coding conventions and preferred frameworks and libraries. To address this need, Gemini Code Assist for GitHub supports custom style guides for code reviews. Each team can describe which instructions Gemini should follow when reviewing code files in their repository. That way, Gemini tailors its code reviews to the needs of the repository,” noted Google Cloud’s Salva, noting that users only need a personal Gmail account to sign up.
If we the average user were looking for justification, affirmation and validation in terms of AI usage, wouldn’t the fact that programmers are (almost) wholly embracing these tools serve as a solid way to underline the opportunity?
Repetitive, Repeatable, Replicable
Let’s remember, the code assistant functions here are once again aligned to shoulder the repetitive, repeatable, replicable elements of programming (remember, computers are good at doing lots of the same task quickly in ways where humans start to make mistakes or fail as a result of rote boredom) so that developers can remain focused on the creative aspects of app functionality and the processes involved behind weaving algorithmic logic together.
Surely this is the time for developer dogfooding to start to come to the fore… plus anyway, that turkey-flavored one looks okay.