Probably observation of best practices used by mankind would be enough, unless we go for splatter effect.
I would expect that similar features were built-in in some more simple drones. For example, a simple mining drone might be programmed to perform test drills on the asteroid in question, table the results, based on observations make a hypothesis that the asteroid is worth of mining, test the hypothesis by further mining the said asteroid and, if the rock lives up to the promise of initial drills, continue mining. That'd be scientific method implemented in silicon (or what ever is used in circuits) right there. Now next just change the objective function to something else and keep the solver.
The second level (memory and learning) would be keeping tables of asteroids mined (ships shot, etc.) so the amount of initial tests can be minimized. While applying this to new fields requires some kind of intelligence, within limits it can be simply applied. Now, if every drone hive has some kind of a master AI (drone mother or whatever), then it is possible that they have simple planning capabilities.
The whole can still have superhuman capabilities for two very simple reasons: Probably less need to rest/recharge to work, very little need to stop thinking, and dedicated "intelligent" systems. So while a human engineer might well be smarter than a runaway factory drone, the factory drone (when given design parameters by the "drone mother" or otherwise acquiring a new parameter set) can work 24/7 until it is done designing a model that fulfills the parameters.
This, assumably, would lead into a taxonomy of several levels of drone hives:
The lesser hives might well be runaway unintelligent drones with broken programming. If they can rewrite their own code and construct new machines, the principles of evolution working, those that still function after self-modification will go on to evolve the "next generation" of drones.
An advanced type of a lesser hive might have intelligent drones (preprogrammed "intelligent" for their task) mixed in, allowing more efficient evolution of the hive. The drones are still just products of broken programming, for example a stargate / station building drone with no termination order and corrupt blueprints.
The greater hives might lead be a runaway intelligent experiment similar Orphyx / Magnus or a lesser hive accidentally ascended into a greater degree of intelligence, possibly by overwriting some built-in constraints and/or mixing parts. While the above could be explained as automatons (continuing to do their thing until stopped), at this level the hive might be able to assess its options.
The final stage would be a greater hive gone rampant in the sci fi AI sense. If the lesser hives were essentially mindless (if brutally efficient at following their broken programming), the greater hives being able to plan ahead and/or autonomously change their goals (but still somewhat predictable to those who have studied them) the rampant hives are simply alien. Their "motivations" might depend on their origins or be completely twisted beyond recognition.
The majority of the hives are probably lesser hives or rather, somewhere between a lesser hive and a greater hive. Some greater hives might be on the road to rampancy, but there may be enough limitations and obstacles that they're still "only" managed to get approximately human- or "clever animal" level intelligence. It is notable that the more they modify themselves doesn't necessarily mean that they become more intelligent, just more alien - and unless the modifications are somehow assessed and controlled, the more likely they are to simply not work. Evolutionary pressure. Of course, a drone can escape from "humanity" more than once: Perhaps someone would see a drone hive as a resource and tamper with it, resulting in a change that might not happen alone. If we accept intelligent drones, then such of an intelligent drone might do this, resulting in a merger of hives.
In camp sci fi, an alien intelligence would of course take note of mankind trying to harm it and decide to destroy all humans, but this is quite human-centric, and on the scale of things not yet very bad. The programming might completely ignore humans (though if it has acquired combat drones, those might still shoot intruders based on old programming), though it might still be very dangerous in other ways even without "intending" to harm anything. We could think about Marathon's Durandal deciding that its goal is to escape the universe, and then preparing to wipe a constellation for raw materials and energy for an experiment on the subject. Humans in that constellation? What humans?
And yes, we have not observed rampant and superhuman level hives. There are three possible explanations. First is naturally the simple explanation that they do not exist, yes. The second is that they are not discovered. And the third is that they have been observed, but they have simply been alien enough that they have not been identified for what they are. For example, it might be a valid observation for a drone needing to travel thru a section of human-infested space that the larger the hive, the stronger response it generates. Therefore, the natural solution is to travel light - either distribute the hive into smaller pieces, send a small and fast building unit where it needs to go and then just beam over instructions to copy itself, or possibly even hide in plain sight. Copy a spaceship and use the network of the humans to travel? Pretend to be an interstellar piece of rock?
Author
Topic: Quick Questions involving Rogue Drones (Read 4751 times)