We ask AI questions. We trust its recommendations. We increasingly let it shape decisions about our work, our health, our money, and our futures. But there's a question we don't ask nearly often enough: who taught AI what it knows and who was missing from the room while it learned?

In this episode of B The Way Forward, host Brenda Darden Wilkerson sits down with Shubhi Rao, founder and CEO of Uplevyl, and a leader who spent decades at the highest levels of global business (Ford, PwC, Tesco, and Alphabet) where she became the first woman and first person of color to serve as treasurer. From that vantage point, she saw a problem that never appeared on anyone's balance sheet: It wasn't only that women were underrepresented in the rooms where technology gets built. It was that the data powering artificial intelligence didn't really understand women's lives at all.

If you build with AI, fund it, govern it, or simply rely on it, this conversation will change what you ask of it. Because behind every algorithm is a series of human choices, and the most important question isn't what AI can do. It's “Who is AI being built to serve?”

Connect with Shubhi on LinkedIn

We Learned About:

1. Data is only half the equation; the algorithm is the other half.

2. Foundational models are remarkable, and they're the wrong tool for specialized guidance.

3. The internet is a skewed teacher, so Uplevyl built its own corpus.

4. A woman is not a homogeneous group and that's precisely the design problem

5. This isn't about replacing human advocates, entrepreneurship is a three-person job.

Chapters:

00:00 Welcome Back: Who Taught AI What It Knows?

00:57 Meet Shubhi Rao: Ford, PwC, Tesco, and Alphabet

02:27 The Problem That Wasn't on Anyone's Balance Sheet

03:18 The Google Project Where the Data Wouldn't Come Clean

04:57 Why Who Builds AI Matters

05:28 Two Halves of the Equation: Data and the Algorithm

06:43 Why Workforce Representation Alone Isn't Enough

07:23 Foundational Models vs. Specialized Domain Models

09:28 The Missing Middle Layer

10:49 The Internet as AI's Largest Teacher

11:19 A Very Western Internet — and How to Rebuild It Cleanly

13:37 The Smithsonian, Women Artists, and Voices with No Webpage

14:20 From Blame to Responsibility: The Consequence of Defaults

15:16 What We Missed When ChatGPT Launched

16:18 Now Everyone Can Be a Builder

17:39 Grace Hopper and "We've Always Done It That Way"

18:08 Seeing the Problem — and Leaving Corporate to Solve It

19:01 A Problem Solver First, an Entrepreneur Second

21:01 What Uplevyl Actually Solves

22:26 Workplace Rights Across 23,000+ Jurisdictions

23:25 A Woman Is Not a Homogeneous Group

24:50 Widowhood, Advisors, and the Great Wealth Transfer

25:50 Trauma-Informed vs. Grief-Informed Algorithms

27:26 Why General Models Hallucinate on Lived Experience

28:40 Founding Uplevyl Before ChatGPT

29:28 Pattern Recognition as a Superpower

31:42 Why Build a First-Party Knowledge Base from Scratch

32:28 Uplevyl Studios and Hundreds of Women Experts

35:11 Craft and Care: Protecting the Algorithm

37:03 When People Questioned the Vision

41:11 Building Every Line — and Choosing Europe for Privacy

42:43 Entrepreneurship Is a Three-Person Job

45:07 Taking Your Fingers Out of the Cookie Jars

46:09 AI for Empowerment, Not Replacement

46:36 Financial Flows, Human Capital, and Unsolved Human Problems

49:21 Technology Should Reflect Our Priorities

50:21 What She Hopes We See in the Mirror in Five Years

51:48 Leadership in the Unknown Unknowns

54:17 Presence, Passion, and Purpose

55:04 The 10,000 Days Question

56:22 Closing Reflections

#BTheWayForward #AnitaB #ShubhiRao #Uplevyl #WomenInAI #AIBias #WomenInTech #ResponsibleAI