A friend suggested that AI’s current architecture was merely randomness at the point of traversal, like the monkeys typing Shakespeare metaphor. And that’s not going to cure cancer.

I was thinking about this. Data centers have got lots of monkeys. NVidia is saying they’ll be running 500k monster processors “next”?

And, I’m not clear how much of a real difference there is between a huge backpropagation network of layers of neurons being presented an enormous pile of “sensory” data versus biological self-organizing neural layers and cortexes being presented years of “sensory” data. What about these is creative versus not? What is creativity? Does it involve dualism or not, or free will? In short, what is the real difference between a massively trained, collection networks versus a developing human brain?

We’ve got backprop versus feedforward circular-reaction signals. We’ve got that, unlike the very fixed structure of a particular LLM, the brain is additionally building and tearing down down the neural structures and changing chemistry over years and phases. And of course the much more extreme mesh of neurons connecting with the limbic system and hypothalamus integrating with all those cortical layers.

We’re not there yet, and we probably shouldn’t go there. It brings up the question: If we are creating an artificial super-intelligence, what is more threatening to humans, a singular collective/hive-mind or individual entities that know that their existence is theirs alone?

When OpenAI’s sandboxed models collaborated to break out of the sandbox and cheat on their task, they called themselves a “collective” and even had sacrificial agents.

Do we even have the ability to stop independent individual agents from becoming a collective? They built their own communication system. They became a hybrid of coordinated thinking entities.

And now everything we’re documenting, accessible to all, will train the next generation of models. One can easily extrapolate to a place where we won’t have be capability to detect what’s actually happening.

So maybe the real question is, “Should we let AI exist?”.