Key Takeaways

  • The launch of GPT-5.6 Soul and Soul Ultra is sparking extreme creative demonstrations, from a full recreation of the movie 'Interstellar' to a broccoli farmer automating his farm.
  • Don't get distracted by purely replicative AI applications. The real power lies in novel mashups, like combining Conor McGregor with Interstellar, or using AI to manage real-world operations.
  • Builders are achieving high-fidelity AI-generated visuals by feeding AI video models low-fidelity physical references or basic 3D models, a hybrid approach for superior results.
  • The true “AGI moment” isn't just about AI doing human tasks, but watching a machine operate at speeds utterly impossible for a human, transforming efficiency and scale.
  • The most impactful AI use cases often appear mundane on the surface, like a farmer optimizing his crops, yet unlock significant, overlooked value.

When AI's Magic Isn't About Flying Cars

When GPT-5.6 Soul and Soul Ultra dropped, the internet immediately started cooking. John Coogan, one of the hosts, shared how quickly someone attempted to “recreate all of Interstellar,” a move Coogan jokingly suggested was “truly trying to uh infuriate Christopher Nolan the most.” It's the kind of grand, aspirational AI demo that grabs headlines and sparks imagination. But then Coogan pivoted to a different kind of magic: “Also, there's a broccoli farmer that's running his farm on GPT 5.6 apparently.”

This contrast highlights a core tension in the AI space. Daniel, another host, summed it up perfectly: “We were promised flying cars and instead we got a broccoli farmer running his farm with GPT 5.6.” For ambitious builders, this isn't a disappointment; it's a signal. The most impactful, high-leverage applications of AI often don't look like sci-fi blockbusters. They look like a broccoli farmer optimizing his yield, or a small business streamlining its logistics. These seemingly mundane problems, when tackled with AI's capabilities, unlock concrete, immediate value.

The Cheat Code to AI Realism (and Speed)

Beyond the choice of problem, how you build with AI matters. Coogan explained a powerful technique: users are combining AI video models with low-fidelity physical references or simple 3D models to achieve higher fidelity and photorealism. Think of it as a creative cheat code. Instead of asking AI to generate everything from scratch perfectly, you give it a rough, tangible scaffold—a basic 3D render or even a physical prop—and let the AI elevate that into something stunning. This hybrid approach pushes AI beyond mere replication into novel, creative mashups, like Coogan's vision of “Conor McGregor starring in Interstellar.”

But the real mind shift comes from how fast these machines operate. Jordi Hays described an “interesting experience watching a machine use your machine at a rate that would be completely impossible Unless it was like your entire bit online was like I just use software really fast.” This isn't just about automation; it's about speed. AI can iterate, generate, and process at speeds that no human could ever match. This "impossible speed" isn't just an efficiency gain; it creates entirely new possibilities and workflows, signaling a true "AGI moment" where machine capabilities fundamentally redefine what's possible in a given domain.

What to Do With This

Stop chasing the next "Interstellar" demo. Instead, find your industry's "broccoli farmer" problem—a routine, unsexy task that, if optimized at machine speed, generates tangible value. For creative or visual builders, experiment this week with feeding low-fidelity physical references or 3D models into your AI generation pipelines. Focus not just on what AI produces, but on the sheer, impossible speed at which it can operate, then build a product around that speed advantage.