In public discussions, Demis Hassabis has repeatedly stressed that AGI is not just an engineering problem. He calls for a “new Kant or Wittgenstein” to guide where society and intelligence should go next.

What, exactly, is this missing piece?

To understand, we can briefly revisit philosophy:

• Ontology asks: what exists? What is the nature of reality?
• Epistemology asks: how do we know what we know? What is the nature of information?
• But history has largely ignored a critical middle layer—Topology, the study of structures, relationships, and the integrity of forms across domains of reality and information.

This hidden layer is what ensures that intelligence, whether human or artificial, remains coherent as it expands, interacts with new formats, and navigates complexity.
Without it, reasoning falters even when memory and computation are flawless.

I have begun formalizing this layer under a new framework—Metaphysics 2.0—which treats ontology, topology and epistemology as three interconnected primitives of cognition. This triangular structure stabilizes reasoning, guides planning and aligns memory with attention and format.

But the full story, including experiments and applications, will have to wait.
For now, consider the question:

If the mind of AGI is a cathedral under construction, what is the hidden scaffolding that makes it sacred?