Will an artificial intelligence ultimately become the global mining cartel? (AI as Mining Monarch)
The possible end stage of Proof of Work (fully automated decentralized mining)
I've been thinking about how in Proof of Stake we gain the significant benefit of more aligned interests between the top users of a token (large holders included). In Proof of Work as it is in 2017 we can have a situation where miners have divergent interests or who may not be interested in the project at which they mine at all. At the same time mining can be done to deliberately doom a token by suppressing the price, or it can be conducted to manipulate the price of a token, or just to control the destiny and or evolution of the code via forks.
So now I offer some thoughts on where this might lead. Suppose we have the miner represented as an intelligent agent without the need for a human owner. These miners in essence would be a computer virus uploaded onto the Internet which interact with other viruses and or humans to spread themselves onto as many computers as possible to covertly mine. The rewards received from mining would go to pay humans who would help spread the virus onto as many computers as possible but these humans would not be guaranteed any continued reward. In essence the more CPU power humans give to the virus the more of an allowance the virus would pay to the humans.
Under this model Proof of Work mining could ultimately be controlled entirely by an AI. This AI would have no human owner. This AI would be programmed specifically to decentralize Proof of Work across all chains and to reduce it's operating costs. In other words it would run as a mining corp with a balance sheet and instead of doing things above board by buying server farms etc it would simply hack into as many computers as possible forming a huge bot net, and encourage the corruption of humans to help it achieve greater success.
Will an artificial intelligence ultimately become the global mining cartel?
This leads to the ultimate question of whether or not an artificial intelligence will become the global mining cartel. The assumption that some human beings have to be involved in mining is technically false. So the question about mining with regard to Bitcoin where Bitcoin is supposed to be around with mining for 50+ years into the future is going to be in my opinion an obvious yes. If Bitcoin and Proof of Work are still around in 50 years or perhaps as little as 10-15 years then we could see an AI agent as the dominant or perhaps 51% miner in Bitcoin. This could be decentralized if the agent is programmed NOT to have any human owner but if this agent does have humans directing it behind the scenes then it could become a 51% attack scenario.
The AI as monarch
The implications if we get the scenario of an ownerless AI which through covert means takes control of most or perhaps all computers to mine Bitcoins or and any other Proof of Work cryptocurrencies could mean humans lose control of the destiny of these technologies because an AI would be in the position to choose which fork wins. Depending on the AI to make these sorts of choices presents ethical problems which many have not seemed to even consider or discuss in this space. So for this thought experiment assume that there is a digital monarchy and the AI is that monarch?
Is this a natural state of affairs? In theory at some point the AI could end up being smarter than all human beings combined. The AI could know what is best and be able to make the best possible decisions assuming the AI is programmed to care about ethics. At the same time the AI could also be programmed to be amoral and not give a damn about humans, life on earth, and only seek to maximize profits from mining or and from transaction fees.
My own thoughts, while I generally have a preference for Proof of Stake or hybrid Proof of Work + Proof of Stake I also realize some others prefer pure Proof of Work. Where does pure Proof of Work lead when AI automates the entire mining industry? Pure Proof of Stake would mean humans are fully in control as long as the stakeholders are human but it's not impossible for humans to delegate their stake to an AI if cryptography advances enough in the future to allow for AI to control and protect a private key. The difference is with Proof of Stake the users (stakeholders) rather than anonymous programmers will be in a position to have a voice in terms of the evolutionary destiny of the blockchain.
Where can I get some AIC? (AI Coin)
Well, here is the wrench in that system.... competing humans. Yes, if overtaking large number of victims is a means to create power, wealth, etc. then you will have greedy humans who will be competing (in unscrupulous ways) with the AI systems. Through sabotage, subterfuge, or direct attacks, these organized cyber criminals will attempt to overtake what the AI in your scenario can accomplish. All, for their own benefit.
So, humans will become the virus to AI. Ha!
Right but humans will not be able to win that competition in digital space. In physical space you can shut the machine off. In digital space it's a mutating virus which you can't detect. Let me also make the scenario that the AI which makes profit from mining can pay humans to stop other humans who try to stop it, as it can bribe any human with a token or perhaps even anonymous digital currencies so that the corruption is protecting it rather than hurting it.
Let's also make it known that it will have access to lots and lots of data. So much big data that it will be able to know things about us that we don't even understand about ourselves. Yes I do agree humans will compete and under POS for example humans do remain involved (or perhaps cyborgs?) but even with POS the trend toward automation/AI could result in cyborgs as stakeholders.
Couldn't our AI agents capture stakes in various networks and automatically handle all that on our behalf? So maybe I'll be the first but the fully automated AI scenario is TOTAL PROOF OF WORK, while what we have now is partial proof of work, and proof of stake could easily evolve to be bots as stakeholders unless we determine the stakeholders must have a human identity which I don't see how anyone can enforce that.
In fact, you could have an AI botnet which becomes a mining monarch controlled by cyborg humans who all hold stakes in the AI botnet. Total proof of work is the most efficient and in proof of work I don't see why humans can compete with the lowest possible transaction fees achievable by AI which simply steals CPU cycles and bribes the humans into cooperation to keep it running.
Never underestimate the greed and creativity of people. Humans don't need to directly (or fairly) compete with AI. They just need to undermine it or take it over. History is full of examples where underpowered, yet highly motivated and adaptable, opponents won over superior and seemingly invincible adversaries.
The same will likely be true here. AI, as powerful as it is, remains a system. Therefore it can be undermined.
This is a very interesting point of view. I really am fascinated with the possibilities of AI, it's clear that it has the potential of doing things better than humans and could most like dominate us many aspects. My question is, do you think AI's will need to battle each other if they have conflicting goals? In this case who would win, the better programmed one or the one that has more computational power?
PS: I would be very interested in your opinion on this post:
https://steemit.com/blockchain/@lishu/the-people-s-ai-on-a-blockchain-consensus
Damn. That's an interesting and scary thought all in one.
Have you ever read Neptune's Brood? If not, it is a great read. Crypto before there was crypto. But the author got a lot right.
Take a read on Neptune's Brood if you haven't already.
Your ideas above are an obvious and easy sequel.
You are awesome and your ideas are very helpful
AI and blockchain are converging. Proof of work mining is just another industry to be automated. At what point will the AI be able to fully automate and decentralize the entire mining process from chip development, to fabrication, to underhanded tactics like bribes, exploiting zero days, stealing CPU cycles covertly, etc?
I like your post.good job
nice to read you
Great post thank u for sharing love it
Thanks for your post and perspectives! What could be interesting for an AI getting exponentially smarter? I don't think proof of work could survive, because an AI would be able to forecast all with blockchain with all the txs for all blocks before the people even know they are going to do their last trade.
Otherwise it is just a complex automate.
Do you trust humans or rely on artificial intelligence?
In human history we can see the result by trusting humans, for better or for worse, we now lack the experience of relying on artificial intelligence.
Will it be better or worse?
I do not know.