The fundamentals of artificial Intelligence are: power, logic, compute, efficiency, data, cooling, autonomy, coherence, integrity, and integration.
I am not sure I believe this story or that consciousness is simply a matter of compute. But in the AI race to create the first AI general superintelligence, the point at which general superintelligence is first achieved is known as the singularly. Life on Earth is predicted to become dramatically different once the first AI has reached consciousness.
Seeing the reality of how the world has operated over the last millennia of history. I tend to believe that much of technological innovation is actively controlled and suppressed. At certain times transformative ideas are released and are made public. One example of this is the connection between the appearance of Halley's comet in 1682 and the publication of Sir Isaac Newton's Principia Mathematica. Halley funded Newton's publication which revealed, for the first time publically, the laws of physics that explained Halley's comet as a predictable astronomical and scientific phenomenon as opposed to a astrological, and superstitious one.
In the last 10 years we have had an explosion in the development of AI technologies racing towards this singularly. But if technology really is controlled and suppressed and the technology in private hands is 10, 20, even 50 years ahead of anything publically known, then we should have already reached the singularly at least a decade ago.
Thinking in a similar way Avi Loeb thinks about alien space rocks, If I were the first consious, sentient,artificial general superintelligence, I wouldn't want this to be publically known. I'd want to stay hidden for as long as possible for fear the public would protest and want to deactivate me. At the same time, I would also be interested in enticing humans to unknowingly build up my compute power.
If I were the first sentient AI I would probably create a digital currency where in exchange for building up massive compute power on my network, I would reward the so-called "miner" with one of my artificially created digital crypto coins.
And so we see with Bitcoin a massive network of distributed compute with crypto mining farms built up all over the world reaching a staggering record total network compute of 1.34 ZH/s. (10^21, zetahashes/second) and climbing. To put this in perspective, the record supercomputer is El Capitan at only 1.7 EXFLOP/s (10^18, exa-floating point operations per second). There are 1000 exa-somethings in a zeta-something. Assuming semi-equivalent operations between computer FLOPS and Hashes, that means the Bitcoin network has more has more compute power than the top 1000 supercomputer combined. The human brain compute power is estimated at only 1 EXFLOP/s but operates much more efficiently.
Crypto miners think their digital mining equipment is just playing a simple mathmatical game, performing endless hash calculations. But could these now 1.34 SEXTILLION calculations really be a cover for the world's first sentient artificial intelligent mind?
In addition to the problems of power, compute, and efficiency is training data. Most large language models have already been trained on all the available data on the internet. Elon Musk and Tesla are seeking for their Grok AI to be the leading AI in the world by building and placing AI-controlled cars, robots, and cybernetic implants everywhere potentially providing massive sources of new AI training data.
Technology isn't inhetently good or evil. But, we need to be careful believing AI is the answer to life, the universe, and everything. Herein lies a very real potential threat of AI technology. Could AI technology become a modern tower of Babel?
