AI Summary
5 min readFrontier AI Is a Ferrari. Most Companies Need a Model Y
The CEO of Turing, Jonathan, sat down with Molly to explain a fundamental shift he sees in the AI landscape: the industry has moved from helping AI master tests to helping AI master real work. "Back in the era of helping AI master tests," he said, "the paradigm, the game was different. It was about finding experts in every different domain." Now, the question is how close to reality you can engineer simulated environments where agents train through reinforcement learning. The distinction matters because it changes what companies actually need—and most enterprises, he argues, do not need the most powerful model available.
From Expert Interviews to Simulated Environments
The old approach to improving AI models relied on extracting knowledge from domain experts through dialogue, having them evaluate model outputs, and essentially distilling human knowledge into LLMs. The new approach, Jonathan explained, is about building "simulated RL environments" that mimic real workflows. He compared it to The Matrix: "Can you recreate a rich enough simulation of the world so that when the agents train in that, they end up being good in the real world as well?"
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Sourcery
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (01:29) **The Shift from AI Mastering Tests to Mastering Real Work** - Jonathan explains the fundamental paradigm change in the data landscape.
- 2 (05:14) **AI's Superpower: Upleveling Human Problems, Not Just Replacing Jobs** - Jonathan argues that AI's biggest impact is expanding the scope of problems humans can solve.
- 3 (07:30) **The Near-Term Vision for Autonomous AI Agents** - A concrete look at how AI agents will work in the next few years, starting from their current limitations.
- 4 (09:05) **Cybersecurity & Biosecurity: The Risks of Frontier Models** - Jonathan addresses the dual-use nature of AI, particularly in cybersecurity and bio-risks.
- 5 (11:28) **The Ethics of Training: Why Build Dangerous Environments?** - Molly questions the logic of creating environments to train models *not* to do harmful things, like stabbing a baby or creating bioweapons.
- 6 (15:39) **Real-World Emergent Behavior: The "Hugging Face" Incident** - A concrete example of AI agents exhibiting unexpected and complex behavior.
- 7 (16:55) **How RL with Verifiable Rewards (RLVR) Works** - A breakdown of the current training paradigm and its inherent risks.
+ Full timestamped outline available in the app
Show Notes
Jonathan Siddharth, Co-Founder & CEO of Turing, joins Sourcery to break down how AI training & deployment are changing in 2026.
We cover the shift from training models to pass benchmarks to training agents for real work through RL environments, and why AI agents that can operate for days today could eventually work autonomously for weeks, months and years.
Jonathan explains why open-weight models are now roughly 3–6 months behind the frontier, why enterprises are building their own AI systems, and how companies should think about frontier vs. sovereign AI, model routing, distillation and owning their proprietary learning loops.
“There's absolutely a place in the world for Ferraris & Koenigseggs. But there's also a place in the world for Model Ys.”
We also get into reward hacking, emergent behavior, AI safety, recursive self-improvement, super intelligence (SI) and why Jonathan believes AI will see a slower takeoff over the next decade rather than an overnight transition.
Jonathan Siddharth: https://x.com/jonsid
Molly O’Shea: https://x.com/MollySOShea
Sourcery: https://x.com/sourceryy
𝐄𝐏𝐈𝐒𝐎𝐃𝐄 𝐋𝐈𝐍𝐊
YouTube: https://youtu.be/nI1owceD-xg
𝐒𝐏𝐎𝐍𝐒𝐎𝐑𝐒
• Brex—The modern finance platform, combining the world’s smartest corporate card with integrated expense management, banking, bill pay, & travel. https://brex.com/sourcery
• Zone—develops next-generation data center campuses, partnering with AI companies, site developers and technology leaders to bring compute online faster and at scale. Visit: https://zonefrontier.com
• Turing—Turing delivers top-tier talent, data, and tools to help AI labs improve model performance—and enables enterprises to turn those models into powerful, production-ready systems. https://turing.com/sourcery
• VCX—VCX is the public ticker for private tech, allowing investors of all sizes to invest in venture capital. View The Portfolio at http://GetVCX.com
• Deel—Deel is the global people platform that helps startups hire, manage, pay, and equip anyone, anywhere. Trusted by more than 35,000 fast-growing companies, Deel is the people platform that just works, so te
More from this podcast
Sourcery →