LLMs Won’t Create Superintelligence… Here’s Why

LLMs Won’t Create Superintelligence… Here’s Why

From NF-Twins by NF-Twins

March 19, 2026 · 1h 3m · Season 1 · Episode 145

About this episode

The episode discusses whether LLMs can lead to AGI and explores the limitations and future directions of AI.

In this episode (#145), the twins talk with Ganesh Krishnan founder of AIhello and Halzero.ai on whether LLMs actually are the path to AGI… or are we hitting a ceiling? We break down one of the biggest debates in AI right now: whether today’s models can ever reach true intelligence. Or whether a completely different approach is needed? We dive into why hallucinations happen, why scaling LLMs might not be enough, and why the future of AI could lie in systems that question their own data , learn through observation , and build a model of the world instead of just predicting text. This episode also explores the difference between hype and reality in AI, how intelligence might actually work, and what it takes to build real products in a space moving this fast. Why LLMs may hit a ceiling The real reason AI hallucinates Why current models blindly trust training data What “world models” are and why they matter Can AI ever question what it learns? The limits of self-driving AI systems Bootstrapping vs hype-driven AI startups Why distribution is now more important than product If you're building, investing, or just curious about the future of AI, this one will challenge how you think…

People in this episode

Host: NF-Twins

Guest: Ganesh Krishnan

Topics covered

  • AGI
  • LLMs
  • AI hallucinations
  • world models
  • self-driving AI
  • AI startups

Keywords

  • superintelligence
  • artificial intelligence
  • scaling LLMs
  • AI products
  • hype vs reality

Mentioned in this episode

Organizations: AIhello, Halzero.ai

More episodes of NF-Twins

Explore listener stats, chart rankings, contacts and more on the NF-Twins podcast page.