Ellen Brandenberger, Senior Director of Product Innovation at Stack Overflow, asks whether it’s possible to implement AI ethically.

As artificial intelligence (AI) continues to reshape industries – driving business innovation, altering the labour market, and enhancing productivity – organisations are rushing to implement AI technologies across workflows. However, while doing so, they should avoid overlooking the need for reliability. It’s crucial to avoid the temptation of adopting AI quickly without ensuring its output is rooted in trusted and accurate data.

For 16 years, Stack Overflow has empowered developers as the go-to platform to ask questions and share knowledge with fellow technologists. Today, we are harnessing that history to address the urgent need to develop ethical AI

In setting a new standard for trusted and accurate data to be foundational in how we collectively build and deliver AI solutions to users, we want to create a future where people can use AI ethically and successfully. With many generative AI systems susceptible to hallucinations and misinformation, ensuring socially responsible AI is more critical than ever.

The Role of Community and Data Quality

The foundation of responsible AI lies in the quality of the data used to train it. High-quality data is the starting point for any ethical AI initiative. Fortunately, Stack Exchange Communities has built an enormous archive of reliable information from our developer community. 

With over a decade and a half of community-driven knowledge, including more than 58 million questions and answers, our platform provides a wealth of trusted, human-validated data that AI developers can used to train large language models (LLMs).

However, it’s not only the amount of data available but how it is used. Socially responsible use of community data must be mutually beneficial, with AI partners giving back to the communities they rely on. Our partners who contribute to community development gain access to more content, while those who don’t risk losing the trust of their users going forward. 

A Partnership Built on Responsibility

Our AI partner policy is rooted in a commitment to transparency, trust, and proper attribution. Any AI product or model that utilises Stack Overflow’s public data must attribute its insights back to the original posts that contributed to the model’s output. By crediting the subject matter experts and community members who have taken an active role in curating this information, we deliver a higher level of accountability.

Our annual Developer Survey of over 65,000 developers found that 65% of respondents are concerned about missing or incorrect attribution from data sources. Maintaining a higher level of transparency is critical to building a foundation of trust. Additionally, the licensed use of human-curated data can help companies reduce legal risk. Responsible use of AI and attribution isn’t just a question of ethics but a matter of increased legal and compliance risk for organisations. 

Ensuring Accurate and Up-to-Date Content

It’s important that AI models draw from the most current and accurate information available to keep them relevant and safe to use. 

While 76% of our Developer Survey respondents reveal they are currently using or planning to use AI tools, only 43% trust the accuracy of their outputs. On Stack Overflow’s public platform, a human moderator reviews both AI-assisted and human-submitted questions before publication. This step of human review provides an additional and necessary layer of trust. 

This human-in-the-loop approach not only maintains the accuracy and relevance of the information but also ensures that patterns are identified and additional context is applied when necessary. Furthermore, encouraging AI systems to interact directly with our community enables continuous model refinement and revalidation of our data.

The Importance of the Two-Way Feedback Loop

Transparency and continuous improvement are central to responsible AI development. A robust two-way communication loop between users and AI is critical for advancing the technology. In fact, 66% of developers express concerns about trusting AI’s outputs, making this feedback loop essential for maintaining confidence in the output of AI systems. 

Feedback from users informs improvements to models, which in turn helps to improve quality and reliability.

That’s why it’s vital to acknowledge and credit the community platforms that power AI systems. Without maintaining these feedback loops, we lose the opportunity for growth and innovation in our knowledge communities. 

Strength in Community Collaboration

At the core of successful and ethical AI use is community collaboration. Our mission is to bring together developers’ ingenuity, AI’s capabilities, and the tech community’s collective knowledge to solve problems, save time, and foster innovation in building the technology and products of the future. 

We believe the synergy between human expertise and technology will drive the future of socially responsible AI. At Stack Overflow, we are proud to lead this effort, collaborating with our API partners to push the boundaries of AI while staying committed to socially responsible practices.

  • Data & AI

Related Stories

We believe in a personal approach

By working closely with our customers at every step of the way we ensure that we capture the dedication, enthusiasm and passion which has driven change within their organisations and inspire others with motivational real-life stories.