Digital Disruption with Geoff Nielson Why AI is Failing: Ex-Google Chief Cassie Kozyrkov Debunks "AI-first"
Nov 10, 2025
In this engaging conversation, Cassie Kozyrkov, former Chief Decision Scientist at Google and founder of Kozyr, dives into the pitfalls of the 'AI-first' mentality. She challenges the 95% failure rate in achieving AI ROI, emphasizing that success hinges on leadership and mindset rather than just technology. Cassie discusses the necessity of fostering a culture that embraces experimentation, the concept of 'AI infrastructure debt,' and the importance of nurturing human involvement in AI systems. Get ready for an insightful take on the real drivers of AI transformation!
AI Snips
Chapters
Transcript
Episode notes
Partner Instead Of Building Everything
- Partner rather than rebuild: organizations that partner are twice as likely to deploy AI successfully.
- Outsource specialized domains like AI security to experienced vendors instead of reinventing them in-house.
Humans Are Part Of AI Infrastructure
- AI infrastructure includes humans, expectations, training, guardrails, and escalation paths, not just cloud and GPUs.
- Treat people as part of infrastructure and design onboarding, measurement, and rollback plans before deployment.
Make AI An Individual Productivity Tool
- Use generative AI as a personal productivity tool: teach people to ask good questions, supply context, and judge output.
- Train employees to 'ask AI first' for non-confidential work and to treat AI as advice, not authority.