The human brain is a mathematically chaotic system. And human intelligence and wisdom etc. is an emergent property of that chaotic system.
It is much like the fact that hurricanes are not built into the fabric of oxygen, nitrogen and water molecules and atoms. If you have sparse chaotic nitrogen, oxygen and water molecules in deep space then hurricanes can never form.
But bring a very large number of those molecules into the gravitational field of a large planet, so that they form an atmosphere and oceans, heat that atmosphere unevenly via a nearby star and spin the planet then hurricanes emerge from the combined chaotic behavior of trillions of molecules.
Intelligence, and SPECIFICALLY conscious intelligence, is generated based on the same principles as a hurricane are in an atmosphere of nitrogen and oxygen molecules.
Intelligence in any particular domain is not built into the architecture of our brains.
Silicon based technology is 100% deterministic. No or very limited emergent properties are possible in a deterministic system. The vast majority of the behavior of a computer is designed by the electronic engineers who make the CPUs, ICs and screens etc., by the software engineers who design the operating systems, by the software developers who develop the apps and by the AI engineers who design the neural networks!
Simulating neurons and neural networks with software and data running on powerful; Nvidia chips does not change the fact that you are trying to build intelligence into the architecture of a deterministic system. And that is IMPOSSIBLE!
I am a former medical scientist with a respectable general understanding of brain physiology and nueron physiology. And I can tell you with absolute certainty that brain physiologists (at the cutting edge) still have NO CLUE how consciousness is generated in our brains, nor even what consciousness truly is. And conscious is an fundamental building block of human intelligence.
When humanity has only the DIMMEST conceptions of what conscious intelligence even is, let alone how it is generated, then it is ludicrous for businesses and the general public to expect that AI engineers to replicate human conscious intelligence in deterministic transistor networks on nvidia chips in their software simulated neural networks.
I am not saying that we should halt all AI research and development. Trying to replicated conscious human intelligence is one hell of a lot of FUN and a worthy CHALLENGE!!!!
But the expectations of the general public and businesses need to be FAR BETTER managed.
And corporate AI engineers need to STOP with their disgusting exaggerations of what they tell the above they are capable of achieving. I.E. They need to stop behaving like the snake oil salesmen of the old American west!
In my view we have reached a technological plateau with AI and its current iterations (chatGPT, google AI etc)
Sure they can be highly useful tools in particular narrow domains (e.g. protein folding). And sure they may become better refined over time in particular domains.
But also notoriously unreliable ones if your try to do more with them than their fundamental silicon architecture can cope with, e.g. self driving cars on other than dedicated roads or tightly controlled highways etc. Failure of self driving cars is generally rare at present (keeping in mind there are not many of them across the world compared to human drivers). But when they do fail then their failure is often CATASTROPHIC. Resulting in deaths and not just fender benders!
If humanity ever achieves C3P0 like AGI then it won't be based on our current silicon based technology. Or at least not based on silicon digital transistor networks.
