315,000 Applications, 0.9% Hired: The AI Screening Revolution at Scale

What happens when a single company receives 315,126 job applications in one year and needs to fairly evaluate every single one of them? That's the reality Goldman Sachs faced for its 2024 internship cycle — and it's the question driving one of the most significant shifts in enterprise hiring today. In this episode, we dig into the real-world data from Unilever and Goldman Sachs — two of the most publicly documented enterprise AI screening deployments in history. Unilever cut time-to-hire from four months to four weeks, saved fifty thousand hours of candidate interview time, and saw diversity hires jump sixteen percent. These aren't projections. They're operational results from systems that have been running at scale for years. But the story isn't all smooth sailing. Early AI screening tools made some serious mistakes — facial expression analysis, "cultural fit" measurements that encoded existing biases, and candidate experiences so opaque that people had no idea what was being evaluated. We talk through what failed, why it failed, and what the enterprise playbook looks like today. If you're in HR, talent acquisition, or just curious about where AI is actually delivering results versus hype, this one's worth your time.