Well, that is hard to answer. Being nobody has yet been able to define the word “intelligence”. Therefor, we cannot definitively group anything into “intelligent” and “unintelligent.”
Complex instructions
Sort:
Jan 16, 2025
0
#4
The wonderful fact is that certain types of process can generate complex functionality given basically two things:
- some sort of information that, in at least some cases, is replicated
- some random alterations to that information
This suffices for the process of evolution by natural selection, with the de facto goal of fitness (viable replication).
The reason it works is the simple fact that if some example of the information tends to get replicated more, it becomes more common relative to another variant that tends to get replicated less. While the process of random alteration has a mild tendency to reduce fitness in real examples of this process. The process of natural selection (the way fitter variants become more common) has a strong tendency to increase fitness. In a system with a reasonable balance of the two, the result is a continual trend towards higher fitness.
The process also generates variation, and in practice different variants may find different ways to achieve high fitness. There is a complex interrelationship here between the variation and the context of that variation. One fit variant becomes concentrated in one part of the tree of ancestry, and in one niche in the environment that is part of the definition of fitness, and another finds itself concentrated elsewhere.
All of this works well in many different forms on computers. While some sorts of AI exhibit part of the process described, other forms exhibit all of it - and these forms can solve hard problems, and exhibit complex functionality, without anyone designing that functionality.
In a sense, the only reason it is not obvious that life has to exist is that the mechanism which provides the replication for genetic information is very complex. It relies on the information itself being the cause of the generation of complex structures - eg proteins and what are made from them - in the previous generation. A fit genome only replicates if it is in a viable cell!#
This makes a difficult "bootstrapping" problem. It is pretty clear that life as we know it is not close to what led to life as we know it, which had to have relied on much simpler ways to replicate information, probably with far lower fidelity and far less precise control. But this is not a deal breaker: such systems can work, and can feasibly generate superior systems leading to life as we know it. It is clear that variation that improves fidelity and which provides improved control can provide large fitness gains, so if early life was viable at all, it is not so surprising it evolved very good ways of doing the essential functionality. Eg chemical evolution in an open environment evolving the tendency to encapsulate itself, and finding ways to make the information more robust (eg by translating it into a different form).
Jan 22, 2025
0
#5
Elroch wrote:
The wonderful fact is that certain types of process can generate complex functionality given basically two things:
- some sort of information that, in at least some cases, is replicated
- some random alterations to that information
This suffices for the process of evolution by natural selection, with the de facto goal of fitness (viable replication).
The reason it works is the simple fact that if some example of the information tends to get replicated more, it becomes more common relative to another variant that tends to get replicated less. While the process of random alteration has a mild tendency to reduce fitness in real examples of this process. The process of natural selection (the way fitter variants become more common) has a strong tendency to increase fitness. In a system with a reasonable balance of the two, the result is a continual trend towards higher fitness.
The process also generates variation, and in practice different variants may find different ways to achieve high fitness. There is a complex interrelationship here between the variation and the context of that variation. One fit variant becomes concentrated in one part of the tree of ancestry, and in one niche in the environment that is part of the definition of fitness, and another finds itself concentrated elsewhere.
All of this works well in many different forms on computers. While some sorts of AI exhibit part of the process described, other forms exhibit all of it - and these forms can solve hard problems, and exhibit complex functionality, without anyone designing that functionality.
In a sense, the only reason it is not obvious that life has to exist is that the mechanism which provides the replication for genetic information is very complex. It relies on the information itself being the cause of the generation of complex structures - eg proteins and what are made from them - in the previous generation. A fit genome only replicates if it is in a viable cell!#
This makes a difficult "bootstrapping" problem. It is pretty clear that life as we know it is not close to what led to life as we know it, which had to have relied on much simpler ways to replicate information, probably with far lower fidelity and far less precise control. But this is not a deal breaker: such systems can work, and can feasibly generate superior systems leading to life as we know it. It is clear that variation that improves fidelity and which provides improved control can provide large fitness gains, so if early life was viable at all, it is not so surprising it evolved very good ways of doing the essential functionality. Eg chemical evolution in an open environment evolving the tendency to encapsulate itself, and finding ways to make the information more robust (eg by translating it into a different form).
Do you think replication is a simple act not driven by information?
Do complex instructions come from minds, or could mindless processes write them? The difficulty of the instructions is a huge part of the question, but the medium in which they were written adds to the question. If they were carved in stone, painted on a cave wall, written with a pin or pencil, found in digital code, or set up in biological systems, what would that add to mindlessness being able to pull it off, or do all of them suggest a mind was at work?