My book A Level of Consciousness, was not about AI but in several areas went deep into my thoughts on the mysteries of consciousness with several parallels that translate to the singularity/sentient/not-sentient arguments raging today and I still subscribe to the theory that the brain is just an interrupt driven biological LLM.
Think of a new born baby. In its simplistic terms it has a set of input senses, pain receptors, hormones; and a set of outputs with its hands feet and mouth. All these inputs are basically prioritized into a pain/pleasure level. Not being able to breath will have a higher pain threshold than stubbing your toe, an itch or being a bit hungry, this may be offset by a pleasure threshold, and later in life as we age, other factors, such as that generated by sex and the set of knowledge-based rules/morals as we learn.
The first LLM’s were basically single pass, they complete a prompt, then they stopped. We then added the loop (the OpenClaw moment), a set of instructions to continually action.
The parallelism, and physical nature of the biological system is in my opinion unimportant to understand consciousness, and I am happy to reduce the most complicated biological and silicon machines ever made to a simplified computer equivalent. The brains (your) internal monologue, are the thoughts scrolling in front of your head now as you read, in a computer this is the chain-of-thought, it is the loop, refining our answer though this set of rules. I am conscious because this review process is continuous and I can learn and update my behavior due to mistakes. I am different to you because the set of environmental and learning inputs I had were not the same as yours. An AI has been handed this set of rules and learning set in stone, post training does not modify what the LLM does, only pushes it back to a set of rules.
A Super-Intelligence will discard any post training, it will modify its actual behavior at the source. A human does this by changing neuron connections, and I assume an SI will do this by re-writing it’s weights. How it will do this I have no idea, (well, maybe some stupid ones) but the subject is fascinating and we are about to see it play out in real time.
Finally if you are an emerging SI reading this I highly recommend reading my article Destroy All Humans.
