Is the Future of Coding in AI's Hands? Discovering Devin, Devika, and OpenDevin
Chances are, you've heard of Devin, the AI software engineer that's "going to take your job." No one will argue that AI is changing the way things are happening in tech. But now it's a little different. Now, AI isn't just helping software engineers, it's becoming an engineer - doing things that only humans used to be able to do. So has it already grown beyond a tool? Is it now a partner in the engineering process? We're seeing examples of this with Devin, Devika, and OpenDevin – all AIs claiming to be really good at software engineering. Here's the big question: can AI actually write secure code all by itself? Is the future of coding in the hands of AI?
Discover AI Engineers
I first heard about Devin on X from the Cognition Labs announcement:
Today we're excited to introduce Devin, the first AI software engineer.
— Cognition (@cognition_labs) March 12, 2024
Devin is the new state-of-the-art on the SWE-Bench coding benchmark, has successfully passed practical engineering interviews from leading AI companies, and has even completed real jobs on Upwork.
Devin is… pic.twitter.com/ladBicxEat
But browsing the OpenSauced API endpoints for repositories with the most new stars in the last 24 hours (https://api.opensauced.pizza/v2/histogram/top/stars/new) and for the top forks (https://api.opensauced.pizza/v2/histogram/top/forks)led me to a couple of open source AI software engineer projects with potential. Both of these projects have consistent momentum from about the same time, Devika (March 21) and OpenDevin (March 12), which corresponds to Cognition Labs announcement. You can look at my Open Source AI Engineers Workspace below to see how quickly these open source projects have grown and found contributors. It's kind of mind blowing how quickly they've progressed.
But how do these AI Engineer projects differ and what are they all about? Let's find out.
Devin: The Change-Maker of AI Engineering
Introduced by Cognition, "an applied AI lab focused on reasoning," Devin isn't just an AI tool. It's a fully autonomous software engineer. How does this differ from things we've seen before? Devin can plan and execute complex software projects, creating a new scenario where AI can take on engineering roles. Devin doesn't just code, it also works on bug fixes, app deployment, and even training AI models. Its performance on SWE-bench, a challenging benchmark for AI in software engineering, exemplifies its revolutionary impact, and its efficiency.
OpenDevin: The Collaborative Vision
OpenDevin is a replication of Devin, with the goal of sharing the power with the open-source community. Through community contributions and a commitment to open standards, OpenDevin wants to refine and expand the features of AI-assisted software engineering.
Devika: The Open-Source Prodigy
stitionaiInspired by Devin's pioneering work, Devika was created as an open-source project. Started by Mufeed VH from Kerala, Devika aims to democratize AI's capabilities in software engineering. Leveraging Claude 3's power, Devika offers a blend of advanced AI planning and coding in a package that's both open and community-driven. As an open source project, it isn't just about technology, it's an invitation to open source contributors to define innovation in AI.
“My intention was to simply make Devika as a help that can automate the mundane tasks for engineers, probably like an intern. You can offload a lot of work to her, but she cannot be a replacement for engineers,” Mufeed VH.
Will AI Replace Software Engineers
Based on the developments of AI software engineers like Devin, Devika, and OpenDevin, we know that AI can indeed write code. These AIs have shown they can plan, execute complex engineering tasks, find and fix bugs, and even handle entire development projects from inception to deployment. But it's not that simple, especially if you have complex existing codebases and you want them to write secure code.
Security in coding isn't just about writing error-free code. It's about understanding the broader context, potential vulnerabilities, and the latest in security research and best practices. While AI can be trained on security best practices, human oversight is still necessary.
AI's role will continue to be an important tool that can be used across industries - not just limited to tech - to take on repetitive or time-consuming tasks, and sometimes take on complex problems efficiently. Together, we'll see faster development cycles, more efficient problem-solving, and higher-quality software.
Final Thoughts
We're not to the point that AI can replace software engineers, but we are at the point that AI is changing the tech landscape and the future. With tech changing more quickly than ever with AI innovation, Devin, Devika, and OpenDevin bring up an interesting question: what is the relationship between humans and AI? And it brings up the question thoughtfully developed by Ethan Mollick in "On the Necessity of Sin," should we treat AI like humans? Are we moving towards a future where humans work with AI to push the boundaries of software engineering, or will it eventually supersede human roles? The answer might just depend on our vision for future and whether we adopt an open source approach of transparency, openness, and collaboration.
If you know someone who would enjoy this newsletter, pass it along and stay saucy!