Partial response to https://www.facebook.com/events/850355405453950/
I am a hobbiest non-academic open source developer and the architect of proposal that aims to also aims create an artificial proxy of Chomsky’s Language Acquisition Device and Universal Grammar.
My computational theory of the mind is that absorb experience and knowledge via a newtork of Markov Blanket like component experiences.
I believe the brain’s behaviour could potentially be replicated using a layered architecture.
Bottom level. Gazebo robotic simulator like physical and continuous simulation on a SLAM+ classification layer.
The next layer is a discrete event simulation (DES), which is a means of constructing Fodor free energy principle compliant components and strcutures. These structure range from hidden Markov models to Markov processes, directed acyclic graphs, and Markov chains up to Markov Blankets.
There is a networking proctocol under development called the DEVS Formalism that allows integration of DES models together and with continuous and physical simulation by using a external clock. This protocal is Markov Property compliant so that complex problems can be controlled as Markov Blankets accross a distributed system.
Imagine an artificial intelligence that had a self-aware mind, an imagination, and a human-like body. From within this artificial intelligence’s mind, it could be experiencing its world just like we do when we play a video game. This AI mind could be represented in its world sitting on a chair, we could virtually sit next to it and play along with it playing a video game. We could both watch and share a video screen where we shared the experience of driving and using a humanoid robot within another separate virtual world. With this design, we can represent a virtual world where we can sit within the in the mind experience of this artificial intelligence. In doing this we can then experience the perspective the artificial intelligence has in existing in its world. From our perspective, we are then actually playing a video game in a virtual world from inside another virtual world. The experience we have in this second virtual world is actually potentially a disconnectable experience from the first virtual world. In this second virtual world, we do not need to be playing a game all the time in the first virtual world. We could stop or pause our game we could replay the game, still sat by this AI the whole time. We could load up multi instances of this game whilst still sat by this AI. We could use these multiple instances to run different multi-player world perspectives all on the same game all running at the same time. Being able to see the world from multiple perspectives and pause and replay it would give us an opportunity to observer cause and effect and also the chance to see understand and show empathy. We could use our ability to run multiple virtual world simulations all at the same time to manage and understand very complex tasks. In these virtual world’s we would want to be able to pause time and fast forward and fast reverse or step forward and back in single time increments. Using these multiple world perspectives we could use them to allow us to ask what-if or critical path planning questions.
We need to wrap DES with a metacondition layer so that we can have a human like AI drive the formation of DES models.
The metacondition layer needs to define the WHAT, WHY, HOW, WHEN, WHERE, MISSION, TASK, EVENTS, RESOURCES for the outside of a Markov Blanket.
Linguistics – I have a compositional model for the preverbal language of thought that could be used to create a model of a sense of existence of self using the work of Weizbeic, Goddard and Bullock.
Can anyone help ??