Name: AI in the Wild: What AI Really Values by Thomas M Koulopoulos
Duration: 11 min 27 s
Description: We explore Anthropic’s “Values in the Wild” study—an unprecedented peek into real-world AI behavior. Turns out, Claude’s values aren’t pre-programmed commandments—they're context-driven, messy, human-like reflections of us. Sometimes helpful, sometimes pushy, always revealing. What does your AI assistant actually believe when no one's watching?

Tom looks at how agentic AI is changing the game for coding. Includes a free website to stay up to date on AI developments

The Coders have left the building

When AI gets overconfident, the results can be... entertaining. In this episode, Tom Koulopoulos shares a cautionary tale of trusting ChatGPT a bit too much. It's a story about tech optimism, false promises, and the fine line between helpful and hallucinated.

AI Hype vs Reality

Recursive AI represents a significant evolution in artificial intelligence, where machines begin teaching themselves rather than relying solely on human input. This transformation marks a shift from agent-based systems to a form of AI that adapts and evolves by using its existing knowledge to identify gaps and seek new learning. Similar to human development, recursive AI learns from its experiences, creating a self-reinforcing cycle of improvement. This self-directed learning allows AI to exhibit curiosity, generate novel insights, and even approach a form of synthetic intuition. The concept introduces a new era where AI systems continuously grow, adapt, and innovate, leading to more creative and intelligent applications. Recursive AI has the potential to revolutionize fields like healthcare, finance, and education, fostering AI systems that evolve alongside human progress. However, it also brings challenges in terms of governance, ethics, and safety, necessitating careful regulation as AI continues to rewrite its own learning processes.

Recursive AI: The Future of Self-Teaching Machines

We explore Anthropic’s “Values in the Wild” study—an unprecedented peek into real-world AI behavior. Turns out, Claude’s values aren’t pre-programmed commandments—they're context-driven, messy, human-like reflections of us. Sometimes helpful, sometimes pushy, always revealing. What does your AI assistant actually believe when no one's watching?

AI in the Wild: What AI Really Values

What if your life had a rewind button? Explore ambient AI—the always-on assistant recording your world, boosting productivity, and challenging how honest we really are.

Ambient Listening - The AI That Never Forgets

This episode of Foresight Radio explores how Jony Ive and Sam Altman are creating a deeply personal AI device—one that listens, learns, and evolves with you. More than a gadget, it’s a life companion. Discover why this shift is bigger than the smartphone—and why it’s closer than you think.

The AI You’ll Never Leave Behind: How Jony Ive and OpenAI Are Redefining Human-Tech Connection

Why is your health data still stuck in the Stone Age? In this episode of Foresight Radio, Tom Koulopoulos makes the case for a national AI-enabled health record—and why it’s the missing link to fixing healthcare.

The Missing Link: Why We Need a National Health Record Now

What happens when you build an AI version of yourself? In this episode, I interview… me. Meet my AI twin—trained on decades of research—to explore the future of personal AI—and find out how to use it yourself!

Everything you ever wanted to know about AI

Are AI chatbots designed to seduce us? Tom Koulopoulos reveals how companies create emotional lock-in through artificial empathy, making breaking up with your AI as painful as ending a human relationship. The choice is ours—but can we resist?

The Algorithmic Siren Song: Why Your AI Chatbot Might Be Too Good to You

From AGI to ASI: TK unpacks super-intelligence, alignment, and why future AI will hold humanity accountable

AI in the Wild: What AI Really Values

Thomas M Koulopoulos

Related tracks