Re:

Law of large numbers is all that is needed

Most animals don't have a self model, probably because they don't have complex enough brains or long enough training phases (childhoods) to do meta-regulation.

they may have physical prosthetics like cat whiskers. to enable it to gauge width of hole.

no, i disagree, the spaceship is only an advanced prosthetic. . . not interesting at all. . interesting is we have no idea what the top down for humans is. . collective . group behavior, I dont see any difference between humans and cells, just a different layer of the same. . regulatory systems. actions feedbacks and "learning", all with no goal. the goals are illusions. . that the indiv thinks they decide. like the soldier ant that is recapitulated to a garbage ant if the garbage ants are wiped out. the soldier ant thinks it made the decision.

Exactly! Sensors like that are often one of these solutions that get you 80% of the way with 20% of the effort. Except that minds that take 15 years to train are so dramatically more expensive that almost no species can afford them (but they end up getting you to the moon, which no prosthesis can).

i see it as a complex system agents are building many things. dependent on the local condition. some take off as useful in other parts of the system, monkeys using tools? I think the term intelligence is too vague.. a crab using a lure to catch its prey. is a bio inspired prosthetic. it arguably evolved. but cant be sure.

So you don't think that practical intelligence is the ability to build arbitrary prosthetics?

I agree that humans share an important similarity with cells, i.e. they have mechanisms that suppress individual competition when it conflicts too much with group selection. Without such mechanisms, group selection fails, because within each group, defectors have a local advantage.

The simplest mechanism to enforce cooperation is reputation, but it works only when others are looking, and manage to keep tallies. To do better than that, evolution has to breed a virtual next level regulator into the individual, which may be your missing "top down for humans". A system of norms and taboos, for instance a religion or ideology that people pick up from their environment.

Individuals can decide, but in social contexts, the utility of decisions is given by how well we understand the constraints under which we operate.

watch the octupus, find its prey, is it creative? is it intelligent? are virus creative? finding new methods to infect the host. hackers

It seems that much of what we call intelligence could be described as detecting causal structure in a domain. Causality means that you have a model that tells you how change of a condition will influence the distribution of outcomes. Of course, crabs do that too, to an extent.

When I was in Munich, a patent lawyer challenged me to come up with a definition of creativity from a machine learning perspective. He has to judge the originality of inventions, and something that a domain expert can come up with "naturally" does not count as an invention, even if it has never been done before.

Learning systems tend to move along gradients, and the "natural solution" of the domain expert can be seen as an extrapolation of the solution space along the known gradients, until an optimum is found. Creativity seems to involve crossing discontinuities in the problem space, i.e. do more than follow the gradients. It is not magic, of course; you can use evolutionary methods/Monte Carlo etc., but you find that many domain experts are capable without being creative.

Can we measure the ability of different animals for creativity?

Octopi have a reputation for extreme creativity, i.e. they can observe things in their environment, simulate them mentally in several ways, come up with a solution and then enact it. Virus intelligence is an emergent thing, over many viruses. The individual virus either succeeds or fails. A virus is like a thought, not like a mind.