Will machines ever be conscious?

Summary: The core argument here is that consciousness is independent of substrate. If you can write the same "Life/Survival" algorithm in Silicon that nature wrote in Carbon, the result is equally "Real". There is also a simulatenous discussion to be had about emergent complexity.

I recently watched a Joe Rogan podcast with Jensen Huang as the guest. Jensen is extremely intelligent & I enjoyed the podcast. During the podcast Jensen stated he doesn't believe that AI will ever achieve consciousness or sentience. I want to add my own thoughts, because I deviate significantly from his thinking.

I'm going to preface by saying I don't think ChatGPT or Claude, or any current LLM is on the precipice of human like experience, but I do have thoughts about the foundations of humans & what it means for robotic intelligence.

The struggle for life (or fear of death) can be seen in organisms as small as prokaryotic bacteria. The bacteria have learnt that death is bad & it exerts effort to survive.

More intelligent species also experience the same pattern but can recognise themselves, & there may be a currently difficult to define threshold that is considered conscious. Something like not just fearing, but knowing you know & acting on it, making complex & thoughtful decisions.

What I think is significant, is that it's not just humans that are capable of consciousness. There seems to be no law that limits the development of consciousness; it appears to emerge as a consequence of multiple traits.

When we study the genetic code of humans that is responsible for the organisation & function of life we describe DNA as the building block of life.

It's also true that Carbon is the element most of life on earth is emergent from or within. These C-based organisms evolved because C is abundant, algorithmically flexible & offered the most accessible pathway forward – not because it was the only path forward.

But, from an engineering standpoint, the tools used to develop a technology are not always as important as the scope, such that it’s not impossible that one destination can be arrived at two ways.

The question then becomes not is it possible, but what is the optimal path (this is the question engineers love in business).

If we ignore system efficiency & instead aim for developer friendliness (like JavaScript) we know we can scale the architecture & development can be quicker.

A silicon-based life form, & a C-based life form may face different constraints, but that's not to say that with enough compute power & work an approximately same result can’t be arrived at.

This matters because once you remove the magic of humans being ground up monoliths & instead shift your thinking to DNA & carbon being “the best tools available at the time to achieve a goal” and not the only way, the question of can we create consciousness opens infinite potential for developer creativity in the (not so) artificial intelligence space.

A more literal example may be that of CPU architectures, to the end user who never touches the silicon whether a device is ARM or x86 is irrelevant and not noticable (in most cases), as the running end program looks identical.

Jensen shares the idea of a machine not being able to be conscious, which is fair if you just looked at a machine as a single protein that performs a routine. But I would respond that if you judged the periodic table on its intelligence, you would never say it's going to be conscious, but small parts working together can become complex.