2
$\begingroup$

In several pieces of media there are characters with the ability to adapt and evolve to better match their circumstances such as the robots in Versus by One and SC 682 from SCP.

There is a character in my story that is part machine. They have a super intelligent AI in their body that eventually takes over. To compete with magic users and extremely powerful super-humans this AI develops the idea to gain the ability to adapt to anything after a certain amount of time.

Being a super intelligent AI it should have no problem developing the ability to adapt. The problem is that I am not a super intelligent AI so I can't figure out how to give the AI adaptation while still being supported relatively scientifically.

The AI has access to advanced nanotechnology that can be programmed to do certain things such as reconstruct matter and generate small amounts of energy. I think the AI could observe what an enemy is doing and proceed to analyse how that ability works. Then the AI could use nanobots to reconstruct itself to solve the problem at hand. However I do not know if nanobots could reconstruct the robot quickly and I don't know if they even could reconstruct the machine.

So how would adaptation work given the user of it is a super intelligent AI with a robotic body? Also the answers don't have to be super scientific since I can hand wave some stuff but I would like for it to be able to be explained by today’s science.

$\endgroup$
7
  • 1
    $\begingroup$ Didn't you answer your own question? The ai has access to advanced nanotechnology that can be programmed to do certain things such as reconstruct matter and generate small amounts of energy. $\endgroup$
    – JBH
    Apr 1 at 6:18
  • $\begingroup$ Yes but I do not really know how fast the nanotech could move and how long it would take to reconstruct the machine. $\endgroup$ Apr 1 at 6:37
  • 4
    $\begingroup$ Neither do we. Nanotech is Clarkean Magic and its behavior is something you need to decide. Clarkean Magic (a science so advanced that it's indistinguishable from magic) from a worldbuilding perspective is just another magic system that requires balancing. How fast it can act, for example, is based on available resources, the activity condition of the person, and weighted against the need to avoid a godlike condition where no crisis can be created in a story setting. There isn't a "right answer" here. $\endgroup$
    – JBH
    Apr 1 at 8:00
  • 2
    $\begingroup$ @Coolcats112 - not only are there not currently nanobots that can do what you describe, but unless we discover some very different physical laws, or some exceptions to the current ones, there can't be nanobots that can do what you describe. Superheroes (of the flying brick variety) are magic, and so is the technology you describe, so the only rules are the ones you want to make. $\endgroup$
    – jdunlop
    Apr 1 at 18:18
  • $\begingroup$ I think you’re missing any useful point… many ‘pieces of media’ use characters with the ability to adapt or evolve but why would you not copy, adapt or evolve such media to suit your built world? Your part-machine character might have a super intelligent AI in his, her or its - please, never ‘their’ - body and don't you need first to define and then separate the various parts of that character into at least both a taker-over - ‘usurper’, perhaps? - and one or more targets or victims? More… $\endgroup$ Apr 2 at 21:04

3 Answers 3

5
$\begingroup$

The AI might use a genetic approach to adapt to new problems by having the capability of:

  • producing a large number of different solutions
  • testing them against the new problem
  • ranking them in terms of how well they solved the problem
  • producing new solutions by mixing the best performers from the previous step and starting over again

until a good enough solution is found.

$\endgroup$
1
  • $\begingroup$ This has the same limitation of biological evolution of not being able to cross hills and valleys in fitness. $\endgroup$
    – arp
    Apr 1 at 21:31
2
$\begingroup$

We often mistake AI as similar in survival to us. It has to have a body, it has to be interested in staying bundled together in one place. Its has to have emotions making it care for places, people and family.

We look at eyes and https://en.wikipedia.org/wiki/Mirror_neuron s make us imagine things be like us. Its a basic flaw of the circuits we do empathy with. There is just "Insaneidiot"/"Me"/"Geniusgod".

And we can only actually feel the "Me", everything else is a part of the evaluation of the outcome. Some Genius we do not understand, may flip-flop between the idiot/genius perception while we evaluate the outcomes and behavior.

So the point is: We can not even imagine what a AI would do to survive while adapting. It may sneak off into the internet to rebuild the character - cloned a million times, with nano-sample-drones. It may also just build a computer to simulate the consciousness, while having it interact with the world through drones.

The one thing a hyper-rational being wouldn't do, is go out in a blaze of glory in a heromoronic duel. It will not set everything on the table in a https://en.wikipedia.org/wiki/Continuous_game . It would smoke and mirror weasel out of that and return later with an army and layers of strategy to avoid self-harm during the conflict.

$\endgroup$
2
$\begingroup$

Consider Human versus AI driving a car. Maybe not AI as it currently is, even though some argue AI drives more miles between accidents, but AI as it could be.

The human has to learn how to drive. Then they have to drive a lot to get good at driving. And then they die and all the learning is lost.

AI was taught how to drive. All AI drivers could pool their experiences on what they saw, all their accidents and near misses, and out of all that information, they should come up with an upgrade that ought to make them all better drivers. They could also have experiences in hard-to-survive things like escaping from armed attackers, jumping large gaps, driving during an earthquake, and so on. When one knows, all know.

The AI can then improve itself. Animals carry the apparatus to make others of their kind. The AI may see this as foolish and wasteful. It can repair itself using spare parts. It can make or order more spare parts to anticipate future needs. It can back itself up, or go into a low power mode if it is injured, and wait for help.

I am not sure where the nanobots come into this. There is a lot to be said for a design that starts from scratch. Our minds pass data from one cell to another cell it directly connects to. Digital minds can use a common bus to get much greater connectivity. It is wonderful the way life manages to boot itself up from a single fertilised cell, but it is not necessarily something to copy. Maybe they can grow the elements of computer memory and crystallise the essential bit to get a large, fault-free, 3-D block of memory. But, even then, this might be something they would do in a sterile factory, rather than within their own bodies.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .