Does AI Actually Speed Up Software Development? The Evidence

AI assistance is reshaping how projects are built. Whether it's a large-scale commercial application or a small, indie "bedroom" project. In this post, we'll explore the key differences, benefits, and challenges of adopting AI assistance in these 2 contexts and ask the question: Does AI speed up or slow down commercial projects?

Setting the Stage

AI has evolved from a futuristic concept to a practical tool in a developer's toolkit. Today, tasks like code generation, automated debugging, and testing assistance are not only possible but also increasingly common. Depending on the project size and structure, the impact of these tools can vary significantly.

Does AI Speed Up or Slow Down Commercial Projects?

Research provides compelling evidence that AI tools generally accelerate commercial development. A Singapore GovTech study found a 21-28% increase in coding speed across developers using GitHub Copilot, with an overall 12% productivity gain when accounting for all development activities ¹. Similarly, a large-scale GitHub study demonstrated a 6.5% increase in successful code contributions at the repository level ².

However, these gains aren't universal. The IBM watsonx study revealed that while 57.4% of developers felt more effective with AI assistance, 42.6% reported being less effective ³. This suggests that implementation and individual factors play crucial roles in determining AI's impact.

The Impact Varies by Developer Experience

Developer Level	Coding Time Reduction	Productivity Increase
Junior	33%	44%
Mid-level	24%	24%
Senior	15%	17%
Lead	13%	12%

Source: GovTech Singapore (2023) ¹

Overall more experienced developers see the less productivity increase.

AI Development Tools (Q1 2025)

AI-Enhanced IDEs

Resource	Description
Cursor	AI-powered code editor built on VS Code
WindSurf	AI-integrated development environment
Zed	High-performance collaborative code editor
Void	Modern AI-first code editor

Coding Assistants

Resource	Description
GitHub Copilot	AI pair programmer by GitHub
Amazon Q	AWS AI-powered developer tool
JetBrains AI	AI assistant for JetBrains IDEs
TabNine	AI code completion tool
Qodo	AI-powered code assistant
Supermaven	AI coding assistant
Aider	Command-line AI coding assistant

Code Generation

Resource	Description
v0	AI UI component generator
Bolt	AI code generation platform
Replit AI	AI-powered coding platform
OpenUI AI	AI UI generation tool
Lovable	AI app builder

Conclusion

AI assistance is transforming software development across different scales, but with nuanced impacts. Research consistently shows productivity gains in commercial settings, with studies reporting improvements ranging from 6.5% ² to 21-28% ¹. However, these benefits are unevenly distributed, with junior developers typically seeing the largest gains.

The effectiveness of AI assistance ultimately depends on thoughtful integration into existing development processes. As Toner-Rodgers demonstrated in scientific research settings, AI tools are most powerful when complementing human expertise rather than replacing it ⁵. For commercial teams, this means developing workflows that leverage AI for its strengths like code generation and understanding while maintaining human oversight in areas requiring judgment and domain knowledge.

As AI tools continue to evolve, organizations that develop balanced approaches to human-AI collaboration will likely see the greatest benefits in both productivity and quality outcomes.

On a closing note, despite the research presented in this article shows that commercial applications can be speeded up to ~20%, as a CTO and FullStack developer overseeing different commercial projects, I must emphasize that the benefits of AI assistance in development time will greatly depend on application complexity, like so, a simple CRUD application will probably hit the 20% mark and a complex regulatory web app solution will probably be on the lower 10's.

Footnotes

1. Ng, K. K., Fauzi, L., Leow, L., & Ng, J. (2024). Harnessing the Potential of Gen-AI Coding Assistants in Public Sector Software Development. ↩ ↩² ↩³ ↩⁴ ↩⁵

2. Song, F., Agarwal, A., & Wen, W. (2023). The Impact of Generative AI on Collaborative Open-Source Software Development: Evidence from GitHub Copilot. ↩ ↩² ↩³

3. Weisz, J. D., Kumar, S., Muller, M., Browne, K. E., Goldberg, A., Heintze, E., & Bajpai, S. (2023). Examining the Use and Impact of an AI Code Assistant on Developer Productivity and Experience in the Enterprise. ↩

4. Vaillant, T. S., de Almeida, F. D., Neto, P. A. M. S., Gao, C., Bosch, J., & de Almeida, E. S. (2024). Developers' Perceptions on the Impact of ChatGPT in Software Development: A Survey. ↩

5. Toner-Rodgers, A. (2024). Artificial Intelligence, Scientific Discovery, and Product Innovation. ↩