GPT-4 just passed the bar exam in the 90th percentile without a single law school class. More unsettling? It also demonstrated theory of mind, correctly guessing what fictional characters were thinking based on limited context clues.

March 2026 brought a quiet bombshell from Microsoft Research: evidence that we might be witnessing the earliest signs of artificial general intelligence. Not the sci-fi version where robots take over, but something more subtle and arguably more significant. GPT-4 is performing complex tasks across domains it was never specifically trained for, suggesting we're moving beyond narrow AI into something approaching human-like reasoning.

The implications go way beyond chatbots. When an AI can ace legal reasoning, generate working code for visual art, and solve multi-step mathematical problems by breaking them down logically, we're looking at a fundamental shift in what machines can do. This isn't about getting better at one thing, it's about getting good at learning itself.

In This Episode:

> How GPT-4's cross-domain performance signals early AGI

> Why passing the bar exam without legal training matters more than you think

> What "theory of mind" in AI actually means for human-computer interaction

> The mathematical reasoning breakthrough that caught researchers off guard

James breaks down the Microsoft research that's got AI labs scrambling and explains why this March 2026 milestone might be the inflection point we'll look back on as the moment everything changed. No hype, just the technical reality of what these capabilities actually mean.

Timestamps:

00:00 The bar exam breakthrough

02:30 Cross-domain reasoning explained

05:15 Theory of mind in machines

07:45 Mathematical problem solving

10:00 What this means for AI development

Follow Unboxed for daily AI updates that cut through the noise. New episodes drop multiple times daily.

Learn more about your ad choices. Visit megaphone.fm/adchoices