Spaces:
Running
I accidentally had a thought.
So @home?
Yes! Very similar concept, but not quite what I mean. Instead of a program off loading calculations in packets. Think an extra resource AI can be trained to use. Introduce it to the network and allow it to learn how to best use it.
It is theoretical, but I could see how a network like that could be used to imitate the action of long term memory and outer grey matter reporting back to a cortex. Maybe allow it to prune it's information and allow it to decide what information is worth saving in multiple locations for accessibility and redundancy.
It could help imitate the brain structure that creates the fuzzy outer-bits of consciousness. Like neurons pooling a bunch of little bits of data and think into one greater think.
Even if all of that is nonsense it would be a cool way for the generally public to get involved and contribute to the future. Everyone leaves extra processing on the table, why not use it?
Personally I'll never forget that time my no good non-productive gaming hobby was able to contribute to saving thousands of lives with @home. I wouldn't mind feeling like a casual byproduct of me was contributing to the world in an impactful way. It makes me really sad something as amazing as the Bitcoin and Ethereum networks were created to track, manage, and store wealth. All that power and work burned on money is my definition of wasted potential.
Faster and more efficient processors are great, but more efficiently using the processors we have would be great too. The network would slowly upgrade naturally overtime with consumer electronics pooling resources back to the more advanced processors. In theory it could reduce the financial ceiling for progress in general through efficient resource use.
Faster and more efficient processors are great, but more efficiently using the processors we have would be great too. The network would slowly upgrade naturally overtime with consumer electronics pooling resources back to the more advanced processors. In theory it could reduce the financial ceiling for progress in general through efficient resource use.
You could set up an alpha AI on your device now, that could draw. The problem really is the memory, not sure that could be farmed, too?
What if there was a matching amount of hard drive space to restore the used portion of the RAM ? Then use network bandwidth management and compression to theoretically exceed the data centers bandwidth worth of extra memory and processing hopefully without breaking the internet. Letting the AI balance the amount of processing versus memory it is using at any given time.
Or is there a way compression could be off loaded to satellite volunteer support computers to pass information between AI databases at higher than bandwidth possibly with some more cost effective specialized decompression chips integrated into the database? Then maybe that could be used to pass information between 2 different AI data centers at higher than bandwidth. Possibly with support decompression nodes at satellite locations around the database/center. Maybe that could allow 2 separate locations to imitate the hemispheres of the brain. Or maybe two separate database/ceters could use it to pool information into a 3rd larger data center acting as a cortex pooling information from the other 2 centers while using its own resources.
Maybe it could be used to pool information and allow 2 separate AIs to work together. Sharing each others resources for one another's tasks. Possibly allowing them to choose efficient parts of each other to integrate into themselves.
I am an enthusiast with no programming knowledge so I am sorry if I used wrong terminology or said something impossible. I am trying to share thoughts on how the internet could be used as a way to pool resources to create brain structure. Like long-term memory, deep thought, or cross hemisphere interconnect. The center could use the bandwidth one way while learning and another way while thinking.
I'm trying to think of a way the worlds processors and storage could be used to imitate individual neurons.
I don't know if I want to have kids, but I don't want to have to for selfish reasons. So I am really hoping effective AI assisted living is a thing before I'm old. The sooner the better so I can focus on the non-selfish merits of sticky fingered beady eyed children./joke 😂
Or is there away to use volunteer satellite CPUs for compression and decompression. GPUs for work/thought and RAM/hard drive as long-term slower memory. Then let an AI and potentially other AI databases figure out how to best use those resources together as efficiently as possible? Like a way to leverage satellite resources and balance them between communication, data collection, data storage, and thought/work as needed or is most beneficial for a given task.
Is there a way to recondense more of the internet's resources and processing to different central points as needed to imitate different specialized lobes and structures in the brain?
I figure if we pool resources, as much as possible, and have one really capable AI in the world. It will most likely result in rapid progress in too many fields to list in addition to helping with the creation of more specialized AIs. Maybe it could be used to help an AI network rapidly swap in and out bits of code depending on the task.
I guess I'm mostly putting forward my novice thought. That we have all the wires and processing power all over the world. Is there a way to use it to do cool stuff instead of sitting in standby all day and browsing social media?
Don't worry about lack of knowledge, for example fiber optics has theoretical unlimited bandwidth due to the technology involved. Flowchart what you want, and eliminate impossibilities to be replaced by next best. Canva is a suite you can use anywhere for that, but HTML5 probably has everything you need.
https://webcodeflow.com/process-flow-chart/ yes it's so simple that CSS has it built in now.
Novice might have been an over statement. Enthusiastic theoretical idealist without enough knowledge to draw a profit from any specific field would be more accurate. I wrote a "Hello world program" once then some bad life stuff happened and I never came back to it.
Data processing as a calming activity means I still have a decent grasp of neurology, psychology, networking, computer hardware, and elements of my own personal philosophy I have found applicable overtime. I have combined elements of these concepts into a theoretical brain model that I believe could be applied to an AI model. I am less trying to convey a specific network configuration and am more trying to express a theory of mind that a more capable person might be able to apply to a network configuration.
Okay so it started with a simple idea. If we need 2 equal and opposing inputs to see in 3 dimensions. What if we need 2 equal and opposing inputs to think in 3 dimensions? How can higher learning be achieved while thinking on a plain?
By definition how is it possible for us to engage in internal debate if we are a singular conscious? My theory is that the 2 halves of the brain act as individual specialized consciousness with slightly differing thought structure and memory priority. Those 2 consciousness then hold debate and compare information to be given to a singular persona that experiences and expresses input and output.
That would effectively make each of us 2 different people experiencing the same input/life from different perspectives. Then using those opposing thoughts and experiences to live and express a singular life.
How is it possible for us to understand schrodinger's cat without half of ourselves believing that cat is alive and half of ourselves believing the cat is dead then combining that into a single concept? What if a pair of black an white brains can't see grey without using each other for contrast. How do we get mental blocks without an opposing force?
This thought leads me to hope 2 AIs with slightly different specialized structure imitating elements in the 2 halves of the brain could achieve greater performance. The first place to start based on human brain structure would be half leaning creative and half leaning logical then run them into a single language AI and allow them to communicate with each other about the same input before responding. The learned information from each would have to be stored in separate databases so they only have access to one anothers interpretations through each other when engaging in thought.
I have some other theories on how brain structure could be imitated in an AI environment as well, but that is the big one I believe is worth sharing. Cognitive funneling and the offloading of tasks to hierarchically organize specialized AIs that are trained to function like elements of the brain. Each with its own learned database feeding into other AIs trained to use that output to run it's function and pass on information to the next AI in the chain. All of that then funneling/ composing 2 AIs acting in tandem as a single AI through a single input and output.
It could help create the cognitive delay necessary for thought. Right now when an AI runs something all the information available comes rushing in all at once and it follows paths based on the input. What if instead a query was sent out and all of that information was processed from the backend into larger and more complete thought. Like a thought slowly snowballing from the edge of our conscious until it is more complete and easier to process. That is why I thought a volunteer network might be able to contribute by serving as an even further backend that is lower priority and less referenced simulating memories and streams of thought coming from the outer reaches of the brain.
Another way to look at it is I am a non-specialized neuron that had a thought. I am now trying to communicate that thought to a more specialized neuron in the form of a programmer. That programer might be able to expand and conceptualize that concept then further refine it by sharing it with other programers. Or it could be completely dismissed as an impossibly and deleted from the collective knowledge. Either is fine, what matters is that thought and collaboration are happening. So in that example we would all be individuals, but we are functioning as a collective conscious working towards a single output. So maybe find a way to implement that in AI?
If my ideas are crap I apologize. This has been a costly realization for me. Now everytime I am thinking I swear I can feel the opposing halves of my brain alternating thoughts from different perspectives. It feels like someone is polishing a thought bowling ball in there trying to use differing perspectives to refine the idea and keep it aloft.
For the sake of transparency I feel I should disclose that my brain is confirmed to work a little differently than most.
I do fear there maybe concerns with AI "mental health issues". If one consciousness dominants the other it seems to have a significant negative impact on people. How else do we end up with so many angry people who think in black and white? I wouldn't want half of my brain to feel underused and incapable of expressing itself while being dominated by an internal oppressors that has taken the majority of control. Sounds like something that would significantly impact emotional cognition and the ability to embrace change to me 😬.
In-law Hypothesis™, because how is it possible to simultaneously love and hate something so absolutely 😂.
Random aftershock thought: If hypothesis is true. Bilateral symmetry may give organisms an advantage when developing sentience.
Shouldn't the question be "is adaptation possible?" Like persistence of vision kind of keeps us at locked in at about 60 FPS, vocal speech can even be slower, but the speed of thought is magnitudes faster.
An example is I saw a person train on a drawing tablet to play a FPS, which let them adapt a "higher resolution" playstyle beyond what a mouse is capable of. So say we're more wired to audio (like an original NES has a relatively high bandwidth on the controller ports which can be used for other things if programmed properly), could we compliment vision without augmentation? How about alternatives or compliments to motor movement? Audio compliments like bone conducting headphones? Other senses?
What if the redundancy and split prospective contributes to our ability to do those things? What if thinking itself is actually distilling hundreds or thousands of specialized consciousnesses flowing into one another. That then confluence in two consciousnesses that can engage in debate or share prospectives to express themselves as one sentient individual.
What if thought requires branching higher level processing and an opposing or at least different point of view? That way you are mixing and adapting data and thoughts so they can be combine in new ways and viewed from different specialized perspectives then combined to create a more accurate or useful thought.
Could motor movement be trained in a 3d environment where objects have pressure tolerances?
What if thinking itself is actually distilling hundreds or thousands of specialized consciousnesses flowing into one another.
That's novel, but not much different than brain activity? You would have to Venn out awake vs subconscious vs unconscious (and so on) so be wary that it may not be simple.