I agree with your thesis that there are two types of creativity and the current iteration of AI is limited to type 1. But I am not complacent about this limit.
Suppose that it becomes possible to make a genuinely self-learning device. Now, give it sensors to mimic all human senses, inclinations to prefer certain sensations over others, locomotion together with freedom to explore, and time. Could such a device, exposed to real and direct experiences of the world, eventually develop some degree of consciousness? Might it possibly exhibit characteristics of human consciousness like self-reflection, meta-cognition (awareness of one's own knowledge, and lack of knowledge), and abstract reasoning? If so, I think it could produce type 2 creative work, if it wanted to.
I think the key phrase there is the last. A device that wants to do something novel of its own volition would almost certainly be capable of type 2 creativity.
If one looks at a new-born baby human it exhibits little more intelligence in its first hours of life than an insect. It is the combination of inherited instincts and abilities, environmental stimuli, the ability to learn, and the particular structure of the human brain that enable it to become a fully conscious and creative individual. If these factors could be incorporated into the structure and experience of a device it might eventually develop the characteristics of human consciousness, the last of which would enable such a device to create entirely novel work. It would have type 2 creativity.
Whether we ought to even try to do this is an ethical question. I don't expect the practitioners of AI to pay more than lip service to it.
I believe this current limit is one that is inherit in the architecture. Meaning that we aren't going to see any further evolution with LLMs. They will get better at pattern matching, but they aren't going to develop further beyond that.
That is not to say that some other innovation might be able to do this and produce true reasoning. But I don't believe anyone knows how to make that happen presently despite many claiming otherwise.
I plan to write something more in-depth on the aspects you mention hopefully soon.
"Whether we ought to even try to do this is an ethical question." Yes, my answer to this is currently no for too many reasons to list here. But I've written quite a lot about it in many posts.
I agree about the limitations of LLMs. Never going to build a conscious creative being with them. But I suspect that developers have gained insights into what "learning" is which might open a path to what I discussed.
It would need to become an antenna for conciousness. The exploration of the quantum nature of reality is driven by it. If the fundamental nature of reality is quantum i.e probabalistic then the connection to the very fabric of reality requires a quantum interface. Not a binary one. The llm can only generate off of finite data sets. Connecting to the infinite would potentially solve this issue. Therefore the answer potentially lies in the interface with quantum computers.
Yes, true intelligence may be a physical phenomenon. We can't simulate it digitally. And we now have some plausible mechanism for quantum activity in the brain.
Why does recombination of the things that are already there, does not create novelty?
I wonder, if you could do this argument for genetics as well. The genetic alphabet is pretty limited. However, it is the basic code for novelties since millions of years. Perhaps, in thinking about artifacts created with AI, we also need something like a genotype/phaenotype distinction?
Simple recombination can appear as a type of novelty. However, it doesn't create new semantic information.
The process that results in gene modification is more complex and I would argue that process itself embeds semantic information in the result. Semantic information would also be the result of gene expression interacting with the environment.
Fun read! Thank you. There are parallels with left- and right-brain thinking worth exploring. "Understanding" has much to do with metaphor, figuration, humour, and irony. AI can't design a new paradigm inspired by observation and then go test it in the real world.
"I dare to say that it can never become type 2". Yes, I think this is correct. Certainly not on the existing type of AI architecture. And at the moment, I'm convinced nobody really knows how to move forward to type 2. Which is probably a good thing considering how many ways we have misapplied the current AI.
I agree with your thesis that there are two types of creativity and the current iteration of AI is limited to type 1. But I am not complacent about this limit.
Suppose that it becomes possible to make a genuinely self-learning device. Now, give it sensors to mimic all human senses, inclinations to prefer certain sensations over others, locomotion together with freedom to explore, and time. Could such a device, exposed to real and direct experiences of the world, eventually develop some degree of consciousness? Might it possibly exhibit characteristics of human consciousness like self-reflection, meta-cognition (awareness of one's own knowledge, and lack of knowledge), and abstract reasoning? If so, I think it could produce type 2 creative work, if it wanted to.
I think the key phrase there is the last. A device that wants to do something novel of its own volition would almost certainly be capable of type 2 creativity.
If one looks at a new-born baby human it exhibits little more intelligence in its first hours of life than an insect. It is the combination of inherited instincts and abilities, environmental stimuli, the ability to learn, and the particular structure of the human brain that enable it to become a fully conscious and creative individual. If these factors could be incorporated into the structure and experience of a device it might eventually develop the characteristics of human consciousness, the last of which would enable such a device to create entirely novel work. It would have type 2 creativity.
Whether we ought to even try to do this is an ethical question. I don't expect the practitioners of AI to pay more than lip service to it.
I believe this current limit is one that is inherit in the architecture. Meaning that we aren't going to see any further evolution with LLMs. They will get better at pattern matching, but they aren't going to develop further beyond that.
That is not to say that some other innovation might be able to do this and produce true reasoning. But I don't believe anyone knows how to make that happen presently despite many claiming otherwise.
I plan to write something more in-depth on the aspects you mention hopefully soon.
"Whether we ought to even try to do this is an ethical question." Yes, my answer to this is currently no for too many reasons to list here. But I've written quite a lot about it in many posts.
You might be interested to follow up with my essay on what might happen if we could build the wish-granting machine and it gives us exactly what we want. - https://www.mindprison.cc/p/the-technological-acceleration-paradox
I agree about the limitations of LLMs. Never going to build a conscious creative being with them. But I suspect that developers have gained insights into what "learning" is which might open a path to what I discussed.
Yes, I read the wish-granting essay. Very good.
Thank you!
It would need to become an antenna for conciousness. The exploration of the quantum nature of reality is driven by it. If the fundamental nature of reality is quantum i.e probabalistic then the connection to the very fabric of reality requires a quantum interface. Not a binary one. The llm can only generate off of finite data sets. Connecting to the infinite would potentially solve this issue. Therefore the answer potentially lies in the interface with quantum computers.
Yes, true intelligence may be a physical phenomenon. We can't simulate it digitally. And we now have some plausible mechanism for quantum activity in the brain.
FYI - https://www.youtube.com/watch?v=QXElfzVgg6M
Why does recombination of the things that are already there, does not create novelty?
I wonder, if you could do this argument for genetics as well. The genetic alphabet is pretty limited. However, it is the basic code for novelties since millions of years. Perhaps, in thinking about artifacts created with AI, we also need something like a genotype/phaenotype distinction?
Simple recombination can appear as a type of novelty. However, it doesn't create new semantic information.
The process that results in gene modification is more complex and I would argue that process itself embeds semantic information in the result. Semantic information would also be the result of gene expression interacting with the environment.
Fun read! Thank you. There are parallels with left- and right-brain thinking worth exploring. "Understanding" has much to do with metaphor, figuration, humour, and irony. AI can't design a new paradigm inspired by observation and then go test it in the real world.
In other words, the true Zombie AI-pocalypse.
Thank you very much!
"I dare to say that it can never become type 2". Yes, I think this is correct. Certainly not on the existing type of AI architecture. And at the moment, I'm convinced nobody really knows how to move forward to type 2. Which is probably a good thing considering how many ways we have misapplied the current AI.
"no architecture can ever match the brain."
Yes, this is a future topic I hope to write about soon.