Research from Stack Overflow indicates a strong overall preference from developers for open-source AI models and tools.
Open-source principles, fostering community contribution and transparency, have been cornerstones of internet development and platforms like Stack Overflow itself. Now, these principles are shaping preferences in AI.
Notably, even figures like OpenAI’s Sam Altman have acknowledged the potential drawbacks of an exclusively closed-source approach. Elon Musk’s xAI has also committed to open-sourcing the previous version of Grok following each new release.
Erin Yepis, Senior Analyst, Market Research and Insights at Stack Overflow, said: “Open-source AI is rapidly becoming a cornerstone for developers, especially those early in their careers. The high trust in open-source models for personal and learning projects underscores the community’s confidence in these tools.
“As the technology evolves, we expect its adoption to continue growing, driven by the next generation of developers.’’
A generational perspective
Unsurprisingly, experienced practitioners familiar with the benefits of open-source code – transparency and community collaboration – stand to gain significantly from the rise of open-source AI.
The survey found that a vast majority (82%) of respondents reported having “some or a great deal of experience with open-source tech.”
This finding aligns with Stack Overflow’s own Q&A trends, where “in the last 365 days, 40% of the top 1,000 tags used in questions and answers on Stack Overflow have been related to open-source software.”
Early-career developers – often the first to experiment with new technologies – show a strong inclination towards open-source AI for learning, even if they lack extensive prior open-source experience.
The survey revealed that “the largest proportion of those with no experience using open-source have the least work experience (12% of respondents with less than five years experience).”
This group also contained the highest number unsure if they’d even used open-source tools, suggesting its foundational nature can sometimes be overlooked. Overall, 9% of respondents were uncertain about their open-source experience.
When quizzed on preferred activities, maintaining or providing feedback on open-source projects topped the list (57% “like it,”) followed closely by engaging in online communities (50%,) and interacting with AI chatbots (49%).
Conversely, activities related to closed-source models faced more aversion: 37% disliked contributing to closed-source AI models, 30% disliked engaging with AI companies, and 27% disliked using proprietary tools for work or school.
Interestingly, younger developers (20-34) showed slightly higher positive engagement scores – particularly for AI chatbots – while more mature respondents (35-54) registered higher negative scores for using proprietary tech at work.
This finding suggests evolving work preferences, potentially influenced by interaction styles, with AI chatbots offering a different feedback mechanism compared to traditional community contribution. Online communities, however, remain crucial bridges for learning and collaboration across all age groups.
Trust leans towards openness
When asked to rate their preference for open-source versus proprietary AI across different tasks, open-source consistently scored higher.
The highest trust was placed in open-source AI for learning or personal/school projects (66% preference) and development work (61%). This contrasts with proprietary AI, which garnered trust from 52% and 47% for the same tasks, respectively.
Using proprietary AI for creative and strategic work received the lowest trust score (43%).
While developers with less than five years’ experience trust both types of AI for work at similar rates to their senior counterparts when looking within the open-source category (65% trust open-source for creative/strategic work vs 69% for those with 15-20 years’ experience,) the gap widens significantly for proprietary AI. Only 53% of the less experienced group trust proprietary AI for development work, plummeting to just 31% among the 15-20 year experience bracket.
“Trust in open-source AI exceeds trust in proprietary AI for the same activities,” the report summarises, though noting that more experienced developers show slightly narrower margins of trust difference between the two paradigms.
This trust is intrinsically linked to transparency; understanding the data models are trained on. Stack Overflow emphasises that “trust in data has been pivotal to the vision Stack Overflow is currently implementing on our community platform, where human-verified content is and has always been the driving force.”
Studies on platforms like GitHub further highlight the vast amount of data shared publicly, often by individuals (78% of data-containing repositories,) underlining the need for better discoverability to level the playing field for researchers and non-commercial projects.
Looking at specific Large Language Models (LLMs,) the survey compared awareness and preference.
Open-source models like DeepSeek’s R1/V3 and Meta’s Llama 70B ranked highly for awareness, alongside proprietary leaders GPT-4o and Claude 3.5/3.7 Sonnet. Here are the top five by awareness ranking:
- OpenAI GPT-4o
- DeepSeek R1
- Anthropic Claude 3.5/3.7 Sonnet
- DeepSeek V3
- Meta Llama 3.3 70B
Open-source AI: Bridging community, business, and ethics
Stack Overflow underscores that open-source AI is far from just a hobbyist pursuit; it represents a significant “legitimate business opportunity.”
Referencing a 2022 Stack Overflow podcast with Amanda Brock, CEO of Open UK, the report highlights sustainable business models around open-source, including paid support, maintaining an open core with proprietary features, managed services, dual-licensing, and donations/sponsorships.
However, challenges remain. Security is a notable concern, with 44% of survey respondents believing open-source AI poses a security risk. Yet, a slightly larger portion (48%) disagreed, potentially balancing the risks against the benefits of community oversight and collaborative improvement.
Crucially, overriding many concerns is a strong ethical conviction: an overwhelming 86% of respondents agreed that “open-source AI serves the public’s best interest.”
The Stack Overflow survey paints a clear picture: experienced developers’ long-held belief in open-source is merging with the next generation’s embrace of open-source AI, particularly for learning and its perceived trustworthiness. This convergence suggests a continuation in the tectonic shift in the AI landscape away from proprietary dominance.
“The findings in our latest survey highlight a convergence: a sustained belief in open source among experienced developers coupled with a growing embrace of open-source AI by the next generation,” the report concludes.
Realising this potential hinges on community strength and addressing practical challenges like discoverability. Facilitating knowledge sharing and improving the visibility of open-source projects and datasets will be vital.
Ultimately, bridging the experience gap and fostering trust may depend heavily on the vibrancy and accessibility of open-source communities themselves.
(Photo by Sébastien Goldberg)
See also: GitHub boosts Copilot with agents, new models, and MCP support
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.
Explore other upcoming enterprise technology events and webinars powered by TechForge here.
