Unscripted with Jeff Pedowitz

Dr. Jennifer Rochlis spent two decades at NASA building Robonaut, a humanoid robot designed to work beside astronauts in the vacuum of space, then brought that human-machine trust expertise into autonomous defense systems. This week on Unscripted, she and Jeff break down why trust is not a feeling but an engineered, emergent property, and why NASA built in manual override for only eleven scenarios out of millions during spaceflight.

Augusto Gonzalez is an economist and researcher from Argentina who uses large language models to simulate entire human populations, running classic economic experiments on synthetic cultural agents. His work can match real anthropology (he replicated field results on the Hadza tribe of Tanzania) but it also revealed that these models flatten humanity's moral and cultural diversity toward a Western, wealthy, educated default, and that scaling does not fix the problem because the bias is baked into what is available on the web. In this episode he and Jeff Pedowitz pursue the deepest version of the AI question: whose humanity is inside these machines? Augusto explains why economists see incentive structures where engineers see technical detail, why naive prompting cannot represent a culture, why the danger is exporting subtle prejudices rather than values, why AI should not drive policy for populations it barely understands, and why he believes open source is the only path that lets every community represent itself.

Zachary Elewitz runs the enterprise AI lab at McKesson, a Fortune 10 healthcare company, where his team exists to de-risk the boldest AI ideas by building prototypes that prove value before the business commits resources. In this episode he offers the rare view from the buyer's seat: why the "we'll all be AI product managers" narrative is overblown, why he discounts what CEOs say publicly, and how an unflashy supply chain optimization saved eight figures a year with no disruption. He and Jeff dig into what executives actually want from AI (revenue and better ways of serving customers, not layoffs), the discipline of admissible risk and naming the failures you are willing to tolerate, how AI is changing customer behavior from "where should I buy" to "what should I buy," how he separates real vendors from buzzword artists, and why quantum computing and post-quantum cryptography are the thing too few people are watching.

Reid Blackman, former philosophy professor turned AI ethics advisor to Amazon, the FBI, and the Canadian government, joins Jeff Pedowitz to argue that the standard approach to AI governance is fundamentally broken. Top-down, policy-driven programs take a year or more to pass, arrive obsolete as the technology races from narrow AI to generative to agentic, and rarely change behavior. His alternative, from his new book The Ethical Nightmare Challenge, discards the values-first playbook in favor of a single pragmatic question: what are the nightmares? Name the specific bad outcomes for a given AI agent, determine the resources and training needed to avoid them, and let cross-functional teams do the problem-solving, pushing accountability to the front line rather than onto a single overwhelmed executive or an unscalable risk board. Along the way they explore why risk-obsessed enterprises are blind to AI risk, why leaders can no longer defer to the "techies," how to spot the buzzword artists in the ethics space, and why the leap from generative to agentic AI is a mountain most organizations have not begun to climb.

Zaheer Ali has built instruments for NASA, run shots at a laser fusion lab, and helped found a national center for nuclear security. Now he's an AI entrepreneur building the first true Space MBA. In this opener, he and Jeff get into where AI is real and where it's hype, why most businesses treat it like fairy dust, what actually makes AI work inside materials discovery and dealmaking, and how a practicing Stoic thinks about governing technology we can't fully control. A wide-ranging first conversation on AI, space, and staying clear-headed in the noise.