It demonstrates awareness, it’s not proving it’s aware. But this is nonetheless a demonstration of what it would look like because such output is the closest thing we have to measuring it.
We can’t measure that humans are self aware but we claim they are and our measure is simply observation of inputs and outputs. So whether or not an AI is self aware will be measured in the exact same way. Here we have one output that is demonstrable evidence in favor of awareness while hallucinations and lies are evidence against
There’s nothing gullible here. We don’t know either way.
Also citing lecun doesn’t lend any evidence in your favor. Geoffrey Hinton makes the opposite claim and Geoffrey is literally the father of modern AI. Both of these people are making claims from the level of measure that is to high level to draw any significant conclusion.
> Either way, you can’t just “teach” laws to statistical models and have them always follow. It’s one of the main limitations of statistical models…
This is off topic. I never made this claim.
> It demonstrated generating text, which people attribute to a complex internal process, when in reality, it’s just optimizing a man-made loss function.
All of modern deep learning is just a curve fitting algorithm. Every idiot knows this. What you don’t understand is that YOU are also the result of a curve fitting algorithm.
You yourself are a statistical model. But this is just an abstraction layer. Just like how an OS can be just a collection of machine instructions you can also characterize an OS as a kernel that manages processes.
We know several layers of abstractions that characterize the LLM. We know the neuron, we know the statistical perspective. We also know roughly about some of the layers of abstraction of the human brain. The LLM is a text generator, but so are you.
There are several layers of abstraction We don’t understand about the human brain and these are the roughly the same layers we don’t understand for the LLM. Right now our only way of understanding these things is through inputs and outputs.
These are the facts:
1. we have no idea how to logically reproduce that output with full understanding of how it was produced.
2. We have historically attributed such output to self awareness.
Shows that LLMs may be self aware. They may not be. We don’t fully know. But the output is unique and compelling and a dismissal that such output is just statistics is clearly irrational given the amount of unknowns and given the unique nature of the output.