A world reliant on increasingly advanced AI coding tools suggests that software creation becomes inexpensive, potentially sidelining traditional software companies. According to an analyst report, “vibe coding will enable startups to duplicate complex SaaS platforms’ features.”
This sparks concern and predictions of the demise of software companies. Open-source projects using agents to address resource limitations should be among the first to gain from affordable code, but reality is more complex. The influence of AI coding tools on open-source software has been varied.
Industry experts note that AI coding tools have introduced as many challenges as solutions. Their accessibility has led to an influx of poor-quality code that risks overwhelming projects. While creating new features is simpler, maintaining them remains difficult and could fragment software ecosystems further.
Thus, the narrative is more intricate than mere software abundance. The forecasted, imminent decline of software engineers in this new AI era might be premature.
Quality vs. Quantity
Open codebase projects are experiencing a drop in submission quality, likely due to AI tools lowering entry barriers. “For people unfamiliar with the VLC codebase, the quality of merge requests is dreadful,” said Jean-Baptiste Kempf, CEO of the VideoLan Organization overseeing VLC.
Kempf remains hopeful about AI coding tools but believes they’re best suited for experienced developers.
Similar issues exist for Blender, a 3D modeling tool maintained as open source since 2002. Blender Foundation CEO Francesco Siddi stated LLM-assisted contributions often “wasted reviewers’ time and affected their motivation.” Blender is working on an AI tool policy but currently does not mandate or recommend them for contributors or core developers.
The situation has become so problematic that open-source developers are developing new tools to manage it.
Earlier this month, Mitchell Hashimoto introduced a system limiting GitHub contributions to “vouched” users, effectively ending the open-door policy for open-source software. As Hashimoto stated, “AI removed the natural entry barrier that allowed OSS projects to trust by default.”
Bug bounty programs also encounter similar issues, with open-source programs like cURL halting their bug bounty initiatives after being swamped with what creator Daniel Stenberg termed “AI slop.”
“In the past, substantial time was invested in the security report,” Stenberg explained at a recent conference. “There was inherent friction, but now there’s no effort required, and the floodgates are open.”
Frustration arises as many open-source projects still benefit from AI coding tools. Kempf notes they simplify building new modules for VLC when managed by experienced developers.
“You can provide the model with the entire VLC codebase and say, ‘I’m porting this to a new OS,'” Kempf mentioned. “It’s useful for senior developers to write new code, but challenging for those unaware of what they’re doing.”
Competing Priorities
A significant concern for open-source projects is differing priorities. Companies like Meta value new code and products, while open-source efforts focus on stability.
“The issue varies from large companies to open-source projects,” Kempf noted. “They’re rewarded for writing code, not maintaining it.”
AI coding tools emerge amid a highly fragmented software landscape.
Open source investor Konstantin Vinogradov observes that AI tools confront an enduring trend in open-source engineering.
“On one hand, we have an exponentially growing codebase with an exponentially increasing number of dependencies. On the other, the number of active maintainers may be slowly growing but isn’t keeping pace,” Vinogradov stated. “With AI, both sides of this equation have accelerated.”
This introduces a novel perspective on AI’s impact on software engineering with concerning implications industry-wide.
If engineering equates to producing functional software, AI coding simplifies the process. However, if it involves managing software complexity, AI coding tools might complicate it. Active planning and effort will be required to maintain control over the expanding complexity.
For Vinogradov, this situation is familiar for open-source projects: ample tasks with insufficient skilled engineers.
“AI doesn’t increase the number of active, skilled maintainers,” he noted. “It empowers the capable ones, yet all fundamental challenges persist.”