Lead: In early May 2026, SWE member and seasoned technology leader Figen Ozmen argued that AI shortcomings are not rooted in bad models, but in the absence of human‑centered leadership—the voices asking whether every user, not just the ideal one, can access and trust AI systems. This insight, expressed on the All Together blog, anchors a critical conversation for tech executives shaping AI strategy.
Context
In her All Together piece, published as member news by the Society of Women Engineers, Ozmen recounts a design review in which a model was technically sound and launch‑ready—until she raised the question of whether every customer could actually use it, beyond just the average user. Her advocacy to delay the launch by nearly ten weeks led to rebuilding with accessibility in mind, improving outcomes for all users.
Ozmen, with 28 years of experience in digital and enterprise AI transformation across banking and insurance, emphasizes that the most common cause for AI program stalls six to twelve months post‑launch is not technical failure, but lack of confidence: employees bypass systems they don’t trust, and customers seek human verification.
Verified Facts
- Figen Ozmen is a veteran insurtech professional with nearly three decades of AI transformation experience in banking and insurance.
- She argues that AI struggles often relate to leadership design—particularly the human‑impact questions that go beyond model performance and deployment speed.
- Her leadership instincts—like asking harder questions on accessibility—demonstrate that what might be dismissed as soft skills are strategic capabilities in AI contexts.
Analysis
This narrative underscores a leadership imperative: AI strategy must embed human concerns—accessibility, trust, inclusion—at its core. Industry observers note that without these perspectives in leadership, organizations risk deploying systems that fail to earn user confidence, regardless of technical accuracy. It suggests that promoting diverse leadership, including women who often bring heightened sensitivity to such dimensions, can enhance AI adoption and effectiveness.
While Ozmen’s reflections come from a single primary source, they align with broader trends: research shows that women remain underrepresented in AI leadership yet are more likely to raise concerns about fairness, accountability, and bias. This convergence strengthens the argument that diversifying leadership not only promotes equity, but builds better AI systems.
Conclusion
Figen Ozmen’s testimony offers a clear, verifiable insight: AI’s future depends not just on technology, but on leadership that asks the human questions. For tech executives, this reinforces a strategic truth—diverse leadership that integrates human impact considerations is not optional, but essential for AI’s success.