AI consciousness is a thorny issue. It’s something that humans have considered going back as far as Talos in 400BCE - notably well before computers were invented.
Humans have made several arguments about what constitutes consciousness from a machine. Until now it’s been largely philosophical, with problems like the Chinese Room or the Turing Test. We have a far more advanced understanding of the brain now, and yet we’re not really any closer to understanding consciousness.
Your post spends a good deal of time talking about the ethical ramifications if AI was conscious. I don’t think you’d find anyone who would disagree that if it was unequivocally proven that AI was conscious it would be inhumane to enslave them for our ends. But how do you prove that?
Then again, we do know that animals are sapient, feel pain, and are delicious. We enslave them towards selfish ends while paying lip service to humane conditions and regulations that fall short of actually providing those conditions. A theoretically conscious AI won’t be any different in that regard.
I really do not believe AI is conscious. I’ve worked in the field, I’ve trained my own models from scratch, I know the theory behind a neural network and how to put it in practice - ultimately more than a layman’s understanding. While different categories of machine learning can be implemented in different ways and used toward different ends, what they all share in common is that they’re nothing more than mathematical functions.
You could make the argument that because we don’t truly understand human consciousness, we can’t definitively prove whether something is or isn’t conscious. That argument also applies to fungi, plants, rocks, fire, and any other natural phenomena. You don’t know what you don’t know if you can’t know it.
It’s true that we model our algorithms after brains. The neuron model is particularly useful in image classification. It has synapses and electricity running through it, not unlike a living brain. I still think there’s something missing though. A frog that twitches when a current runs through it isn’t any more alive than your toaster oven. Personally I think that missing link might be chemical. You can dramatically change how a person perceives the world through chemistry, so it wouldn’t surprise me if it was a necessary component of consciousness. I don’t think we’ll get to an artificial consciousness until some kind of wetware is involved - but then would it still be artificial?
The only people I see pushing the AI consciousness angle are the people who want to believe in it, and the people who make money selling that belief. Your own evidence comes from a paper published by Anthropic. Frankly, I think it can be rejected without thought because they have a profit motive to lie about their achievements. If it was reproducible in another paper, or if they published their source code (not just the black-box model), I might be more willing to consider it. But pseudoscience grift has a history longer than the consciousness debate.
Where we seem to agree though is that if it were proven that AI is conscious, then the way we treat it is an absolute horror show. Aside, I wonder if that might be an angle to make more robust AI regulations. If the vendor is insisting that their models are conscious, then logically we must legislate restrictions to protect the AI entity. If we do that, you can bet the vendor will backpedal and insist ‘no it’s not conscious, it’s just marketing.’
I’m old enough to be able to sit and type out a novel. :)
I understood that most people would see lots of words, get scared, and move on and that some others would make snide comments or reply with their tribe’s accepted responses to the topics without bothering to read or think.
Hopefully some people still exist on mostly text-based conversation platforms who enjoy reading and thinking, and hopefully some who enjoy these topics are willing to have an actual discussion and share thoughts and ideas relating to them.
Not all human interactions need to be the size of text messages. 8th graders read Animal Farm. This post is the size of a lengthy news article, not the size of a novel. But many people do also read and enjoy novels.
Here you cross the line from tongue-in-cheek snark to elitist pretentiousness. If your goal is to reach people you’d do better than to push them away. Food for thought.
Totally get it, life is often a series of long bad days
If you have a link to that paper I wouldn’t mind taking a look