By the end of 2024, we anticipate having a fully-functional CORTECON (COntent-Retentive, TEmporally-CONnected) framework in place. This will be the core AGI (artificial general intelligence) engine.
This is all very straightforward. It’s a calm, steady development – we expect it will all unfold rather smoothly.
The essential AGI engine is a CORTECON. The main internal engine inside a CORTECON is a latent node layer, where the node activations are governed by (among other things) two parameter – an activation enthalpy parameter (epsilon0), and an interaction enthalpy parameter (epsilon1).
The unique thing about this latent layer is that node activations (governed by the two-parameter controls), are always and dynamically subject to free energy minimization. This is what keeps the CORTECON model more-or-less stable.
We’re using the term “more-or-less” because, by autumn of 2024, we should be experimenting with metastable states that will allow us to have neuronal avalanches. That is, we will be able to switch the system from relatively few active nodes to a large number of active nodes, and back again. We’ll also be able to control clustering, using the same parameters.
Rough-Cut R&D Plan
In early-through-mid 2025, we should be introducing control structures, feedback loops, and interactions between the latent node layer and an ontology layer, which should give us the ability to link the subsymbolic or “connectionist” (latent node) layer with structured ontologies, and use those ontology activations in feedback mode to help guide activations in the latent nodes. Together, all of this will comprise a single CORTECON.
The architecture just described is a bit more complex than that which we’ve been showing thus far; what we’ve been using to illustrate CORTECONs has focused on the latent node layer, together with a set of sensory and action capabilities that conceptually connect CORTECONs with Friston’s work in active inference.
