The AI industry is having an argument about what AGI actually is.
Jensen Huang, co-founder and CEO of NVIDIA says it's here, and defines it as a company worth $1 billion.
Google DeepMind disagrees, publishes a cognitive framework with benchmarks.
Both miss the point.
Huang's definition is market cap dressed up as science.
DeepMind's is closer. They treat intelligence as multidimensional, a set of interacting faculties like perception, memory, learning, reasoning, metacognition.
That's a real improvement over scaling laws. But there's still a gap.
The gap: a system can score well across every faculty on a cognitive profile and still fail to behave intelligently.
Why? Because intelligence is not the sum of faculties. It is what emerges when those faculties are organized under a unified dynamic.
DeepMind measures performance. It does not measure organization.
And organization is where real systems break.
A system that reasons but cannot maintain context. Learn but cannot transfer. Generates but cannot validate.
That is not partially intelligent. It is structurally limited. Averaged scores hide the point of failure. Integration is either there or it isn't.
Qubic's scientific team wrote this up in detail. Their position is grounded in cognitive science going back a century. Carroll. Cattell. Kovacs and Conway. The g factor isn't a sum. It's a hierarchy.
The summary: intelligence is what you do when you don't know what to do.
This is why Aigarth and Neuraxon don't look like other AI architectures.
Instead of maximizing scale or enumerating capabilities, they focus on how multiple interacting units produce coherent behavior across contexts that were not in the training data.
Integration first. Performance second.