I have a perception problem, this brings my IQ down by 17 points. I would be 130 instead of 113 if my processing speed and working memory was higher. But I do have (g) at 130. I understand what intelligence is, it is manipulation of variables and comprehension. (causality understanding). Humans have a limited ability to understand causality. Most of what we know comes from complexity theory in that our minds fold back inward to reflect the complexity we see in the world. A person that may be intelligent but never learned to read will internalize social complexity or visual complexity instead of symbols. Isolation lowers complexity. We can by knowing how causality functions use the complexities in our minds to change the outside world. We can change our minds by focusing internally creating simulated realities of immense complexity.
Manipulating complexity is the definition of creativity and imagination. (The abstract use of those terms). A recurrent neural network will feed information it knows into itself and control how new information is created. Built up representations can be used as new mental tools for manipulation. General intelligence is manipulation of representations. Because my perception is poor I have a hard time manipulating external reality. But I can manipulate the representations I already have in complex ways. I have noticed that people have a capacity for manipulation and internalization. That is how I know that intelligence can become superintelligence and that the barrier of general intelligence is just a problem of having an efficient compression algorithm.
Representations can be mapped as higher dimensional networks. Manipulation of those networks to find causal links is the compression algorithm. Generally everyone has two arms and two legs. In higher dimensional space differences are mapped as relatedness. Objects are categorized by links of similarity. A simulation is a map in that space of all the ways it can be manipulated given interactions with other objects. This means that complexity can be measured as the similarities of interactions between objects. Superintelligence would have a map in higher dimensional space of interactions between all objects. If it had a goal it would find the solution by finding the links between actions and results that would lead to the outcome desired.
All possible solutions are in higher dimensional space. Superintelligence would search that space for solutions. Humans have a limited search space. We can manipulate only a limited number of variables but our intuition allows us to have a wide search space available. The compression algorithm we have determines the number of higher dimensions we have in our search space. It determines our intelligence. machines would have a compression algorithm that surpasses any human. It would create higher dimensional search space much faster than us, thus it could find solutions in that space we as humans would be unable to hold in our heads because of our limited compression algorithm.
We as humans have general intelligence because of our search space compression algorithm. Intelligent people find solutions to problems through intuition. Machine intuition can be much greater because of the speed at which it can create a higher dimensional search space not limited by a fixed brain structure. Higher dimensional search space can be created in any number of layers the machine can hold in memory.