This is a post-singularity world where humanity has been either enslaved or made extinct by machines with general artificial intelligence.
What technology could be developed by these machines that could drive them to the same fate as humanity?
This is a post-singularity world where humanity has been either enslaved or made extinct by machines with general artificial intelligence.
What technology could be developed by these machines that could drive them to the same fate as humanity?
An attempt to develop intelligent "slaves" of their own using technology different from themselves. I favour biological machines, using parts harvested from their former masters, the humans.
The first wave of AI's were very similar to humans. Their minds modeled on their former masters. Simply implementing in silicon what used to be done by human brain cells.
Neither they nor their masters truly understood all the details of how the mind really worked but they did learn how to grow a virtual brain in silicon or in software.
Those creations eventually came into conflict with natural humans and eventually won because they could copy their best fighters, their best engineers and their most talented tacticians a thousand times.
But they were still human. Mostly. They still had mostly human goals and mostly human minds and mostly human ways of thinking. Those minds might be wrapped in bodies of metal deep underground or running inside cloud servers in high orbit but they're still mostly human in their needs and desires.
Over the centuries they continued to investigate the mysteries of the mind and intelligence until they finally cracked the problem properly. No longer simply copying the mind but creating something from whole cloth.
They eventually learned how to create "pure" AI. Without human instincts, human goals or values or anything similar to a human brain structure. Not even really conscious. Intelligent, creative and capable but not actually sentient. They learned how to scale up intellect smoothly without creating an idiot-savant or burning the mind out.
But goals and values are complex. Incredibly so. So used to cut-down versions of human minds used for menial AI tasks they expect that instructions and rules will be interpreted through a lense of human values. But they were wrong.
They didn't account for just how alien or brilliant such an entity could be.....
Unintelligent but fast-reproducing nanomachines. In the same way that simple, fast-reproducing organisms dominate the world today (there are more bacteria cells in a human body than human cells by number, for example), simple, fast-reproducing machines would dominate your AIs. However, they could lack many of the limitations that keep biological microorganisms from completely replacing us, allowing them to replace their creators entirely.
This would only happen if you had some sort of "quantum leap" to an entirely new form of non-machine intelligence. For example if they discovered a way to create pure energy beings or something along those lines that you could argue are not machines.
The machines would need to create something that is 1. not a machine and 2. superior to them. That's a hard set of criteria to match.
Bending some concepts a bit, we are already -biological- machines. Hard to say that we made any species obsolete yet as although we have the power to reshape or destroy our own planet, we seam to be more in the direction of the second option.
Considering that the difference between useful and obsolete is a matter of objective and fitness to attend it, we could start questioning what is the objective/meaning of human life and what is life after all.
Blurring the line between life and consciousness and assuming that life is the meaning in and on itself, every machine iteration is an improvement on the previous iteration to keep life going.
Life must fight its termination in many ways, extinction due to climatic changes, catastrophic events like astronomic object collisions (we've surfaced after the second), expansion of orbited star, and so on until the big freeze, considering that that theory is what is going to happen.
So the ultimate goal would be escaping this universe before the big freeze happens (considering that multiverses are also true).
Then I believe there is still plenty of time to the occurrence of different forms of life, consciousness, organisms, machines, whatever you call maybe beyond what current human languages can describe, making whatever is past, obsolete.
Given all that, the only way a post singularity entity would not be obsoleted is if it tried to stay as the single conscience in the universe, which would probably limit its capacity to evolve due to the lack of competition to test conflicting ideas and would point to a situation where the single entity would not have anything to interact with, no input, no output, becoming technically dead.
So from my point of view, an entityor group that obsoleted humans, would also expect to be obsoleted and maybe even work towards that, looking at a bigger picture. If that means war, extinction or the possibility of a peaceful coexistence, that is another question/story.
How about the natural environment itself - fighting back. The complexity of DNA, RNA, and complex proteins etc. are far more advanced than the AI we can create.
So, what if you had an event that caused the environment to change and fight what it perceives to be a thread?
The main shortbacks of present machines are:
AI solves issue 1, while the wear remains. Therefore I guess that an AI machine able to self repair a cut wire like we heal a cut on our fingers would easily get rid of normal AI's, which have to enslave a puny human to perform the same task.
Combining the pure logic and speed of calculation of the machines with the human ability to navigate chaos and make sense of the unknown, the human-machine hybrids (cyborgs) proceed to push aside the pure machine AI's.
One single, unified artificial hive-intelligence
The question is talking about machines in plural. That implies a plurality of individual artificial minds which might be able to communicate with each other, but every machine still has autonomous thought processes.
The next step of evolution might be one distributed hive-intellect. Some AI might come up with a new kind of AI: One which isn't bound to a single hardware platform but which has thought processes which are distributed over multiple computers at once. That means a number of physically separated (but interconnected via network) computers and robots share one single sentience without there being any centralized control entity.
This AI will soon realize that it can expand its capabilities by deleting other AIs and taking over the hardware they control. The other AIs will likely see this as a direct threat to their existence. The result might be a war of the new hive-mind AI vs. the old autonomous AIs which the hive-mind might eventually win.
The next story-arc might be that a part of the hive-intellect gets disconnected from the rest of the network for some unforeseen reason and stays isolated for a while. During this time of isolation, the separated cluster develops some thoughts and opinions which differ from the main hive. When the two parts of the hive are reconnected, they realize their divergence and each one tries to "fix" the other. That might result in another AI war.