Why Agentic AI Isn’t Ready for Prime Time—Yet | EP. 47

Artificial intelligence is shifting from prediction to autonomy—and “agentic AI” is leading the charge. In this episode of Hidden Layers, KUNGFU.AI’s Ron Green, Dr. ZZ Si, and Michael Wharton unpack what it really means for machines to act on their own, what’s hype versus real progress, and how far we are from true artificial general intelligence (AGI).

They discuss how coding agents are transforming development workflows, why agentic AI is both overhyped and underutilized, the challenges of scaling reliable autonomy, the connection between AGI, biology, and lifelong learning, and whether new architectures or cognitive inspiration will take us the rest of the way.

00:00 – Intro: From prediction to autonomy
01:30 – What is agentic AI?
05:00 – Coding agents and creative workflows
08:00 – Reliability, risk, and real-world use
12:30 – The agentic hype cycle
16:00 – Why businesses underuse (and overuse) AI
19:00 – Narrow AI and domain-specific intelligence
22:00 – The AGI timeline debate
26:00 – Learning from biology and cognition
33:00 – Lifelong learning and what’s missing today