Editor’s note: This article is based on insights from a podcast series. The views expressed in the podcast reflect the speakers’ perspectives and do not necessarily represent those of this publication. Readers are encouraged to explore the full podcast for additional context.
Artificial intelligence is evolving at extraordinary speed, but alongside innovation comes the critical responsibility of building systems that are safe, fair, and trustworthy. In a recent episode of the “CAIO Connect” podcast, host Sanjay Puri sat down with Sarah Bird, Chief Product Officer of Responsible AI at Microsoft, to discuss how organizations can build AI that people trust.
The conversation explored Bird’s unique career journey, the growing importance of responsible AI frameworks, and why evaluation and governance will shape the next generation of AI systems.
Bird’s career journey is an unusual and fascinating one. At just 19, she began working on processor design for the Xbox 360, diving deep into the complexities of hardware engineering. That early experience gave her a powerful foundation in systems thinking, understanding how large, complex systems operate and anticipating how they might fail.
Over time, that same mindset naturally extended into the world of artificial intelligence. As AI systems became more sophisticated, Bird recognized that the challenges surrounding them were not purely technical, they also involved human impact, ethics, and accountability. Her role at Microsoft now focuses on ensuring that AI technologies are developed responsibly, with safeguards that address fairness, bias, safety, and transparency.
READ: AI governance, copyright, global collaboration: Key insights from Regulating AI by Sanjay Puri (June 22, 2026)
During the conversation, Bird reflected on founding Microsoft’s FATE (Fairness, Accountability, Transparency, and Ethics) research group in 2017. At the time, responsible AI was far from a mainstream topic in the technology industry.
Convincing organizations to invest resources into ethical AI practices was not always easy. However, Microsoft’s leadership recognized early on that responsible AI would become essential for the long-term adoption and credibility of artificial intelligence systems, “I think Microsoft’s leadership, one of the reasons I love doing the work at this company is they’re very long term thinking.”
Today, what began as a research initiative has grown into a comprehensive framework that integrates responsible AI practices across product development, governance, and engineering teams. Bird’s work demonstrates that responsible AI cannot be treated as an afterthought, it must be embedded into the entire lifecycle of AI development.
Another key topic in Bird’s discussion was the role of open innovation in the AI ecosystem. “I think that it’s really important to have this open innovation ecosystem where anyone can come and contribute, anyone can sort of inspect what’s going on,” she said. Bird has long been involved in open-source initiatives and believes that open collaboration plays a crucial role in advancing research and innovation.
However, she also emphasized that openness must be balanced with caution, “the right thing to do is really make sure that we can offer this technology in a more controlled way.” As frontier AI models become more powerful, releasing them publicly without careful consideration can introduce risks, including misuse, misinformation, or unintended consequences.
Organizations therefore need to weigh the benefits of open access against potential societal impacts. Responsible decision-making about how and when to share powerful technologies will be an important challenge for the AI community moving forward.
One of the most compelling insights Bird shared is that AI evaluation may become the most important technical challenge in the coming decade, “I’ve been saying this for a long time, that evaluation is one of the most important tools and honestly still one of the hardest problems.”
READ: Mehdi Jomaa on AI governance, global south voices, and the future of leadership (June 22, 2026)
Traditional software development relies heavily on deterministic output systems that produce predictable results. Generative AI, however, produces probabilistic responses that are much harder to measure and verify.
As a result, organizations are now focusing more heavily on building evaluation systems that test model performance, accuracy, and reliability. Bird predicts that future software workflows may increasingly revolve around designing evaluation frameworks rather than writing code from scratch.
As AI systems become more integrated into everyday life, trust will become the defining factor that determines their long-term success. Bird emphasized that responsible AI requires collaboration across disciplines, including engineering, policy, ethics, and social sciences.
Her team at Microsoft includes experts from diverse backgrounds engineers, researchers, linguists, and policy specialists working together to address the complex challenges posed by modern AI technologies.
The conversation between Bird and Puri highlighted a powerful truth: the future of AI will not be defined solely by technological breakthroughs but by the systems we build to ensure those technologies are safe, fair, and worthy of public trust.