1.1 Hitting Hard Limits

Overview

This Publication is a broad exploration of Artificial Intelligence (AI), Artificial General Intelligence (AGI), and associated digital technologies used throughout the world across a vast range of industries and applications, and what is considered AI will do to Humans living in the world from now. 

Most importantly, this Publication is about why it is critically important that the development of AI needs to be strictly controlled, as a matter of most extreme urgency, in order to manage the planet wide risks of AI to Humans, and everything else on Earth.

It also introduces the a new and strange concept of Hyperselfish Intelligence as the theoretical basis for what could be fundamentally driving Humans to develop AI.

How this started

This exploration quietly began way back in 1989, when completing a Bachelor of Applied Science in Digital Technology degree and applied research project on backpropagating Neural Network AI technology. It was blatantly obvious back then that AI will develop to have unprecedented impact on the world, however AI still had a long development road ahead, and was only really limited by raw Computer processing power. 

Since that time, there have been a number of extraordinarily lucky opportunities to work with many of the world's largest and most advanced multinational technology companies spanning a career over 20+ years. This work has involved many different types of digital technologies and automation systems, including several transformational commercial AI applications, especially in the automated computer vision, imagery analysis, and geospatial industries.

Big Data and High Performance Computing

In relation to AI, the 'game changing' moment occurred in 2011 while working at Google’s Enterprise Geospatial division with some very advanced cloud-based technologies, and enterprise scale cloud Computing first started gaining serious commercial traction.

There was one particular day when the head of Google Engineering gave our geospatial team a mind-blowing demonstration that showcased Google’s cloud processing capabilities on an absolutely gargantuan volume of geospatial digital mapping and satellite information of the whole Earth.

Then and there, it was shockingly obvious that when Neural Network software that is used in AI applications is run on this huge level of Computer processing power, everything is going to massively change for everyone and everything. 

By exactly how much? Well it was not possible to know because nobody had done it yet. Even today, now that it has been done, it's still not possible to predict everything that will change because there's never been any other technology like AI. But, the world will definitely change, and the ways AI will change the world for Humans are discontinuously different to the predictable development trajectories of every technology that has preceded AI. 

AI is totally different.

Broad Longitudinal Perspective

Having spent a career working at the nexus between commercialization and advanced technologies with large manufacturing factories, industrial processing plants, complex information systems, and global mapping systems, it has become normal to build a ‘Overworld’ perspective to the work. It helps to mentally view the subject at hand from the top-down, conceptually disassemble all of the different interacting systems and processes to understand their core principles, predict how they will behave and longitudinally play out over time, and creatively find ways to improve the total system. This is an engineers approach to understanding and solving an issue.

In the case of the issue; AI and what it means for Humans, it became necessary to take a much wider ‘Universal’ perspective, because the impacts are so far reaching. This perspective leads into some very esoteric ‘out-there’ considerations that wholly step into the realm of the philosophical, science fiction, mystic, and unanswerable, which are interesting, but not necessarily pragmatically useful for understanding and solving an issue like AI risks versus rewards. 

Some of this information really stretches the mind, and it will never snap back to its original shape. Frankly, AI and what it means for Humans is a massive and daunting subject and some of this information is truly terrifying, BUT there is ALWAYS an answer to every issue if you are prepared to relentlessly work HARD enough. It also helps to have an underlying innate curiosity and desire to build more true understanding in the hope it may be possible to make the world just a tiny bit better.

The biggest take away and most bizarre realization in working through and completing this exploration of AI to the current point is that it now seems very obvious: AI is not here to help Humans.

Why? Well it will take some time to read and understand all of this because it's reasonably complex with many different parts, and AI developments are highly dynamic with new large advancements now appearing every single day. The intent is that each piece of information presented will take you on the journey to expand your knowledge, a bit like unfolding and discovering new and surprising parts of a large treasure map, that will enable you to arrive at the same deeply concerning realized understanding. It is hoped that you decide to take this journey and learn why this realization about AI and what is happening to Humans seems to be obvious.

Understanding this changes EVERYTHING about the approach to ongoing AI development !

However, don't think this realization is necessarily only bad, because it's definitely not. In fact, AI could be unbelievably fantastic. We have a very brief window of opportunity to carefully direct the future of Humans onto this exciting pathway and avoid the dangerous pitfalls. 

What is means is that Humans have some extremely important and tremendously difficult decisions to make in the next few months and years, and particularly those leading and controlling state-of-the-art AI development. This Publication does not pull its punches, so apologies in advance if this offends or frightens anyone, because that is not the purpose at all. It's purpose it to help you better understand and think about things you may never have known or considered about AI, and then perhaps motivate you to take positive actions that benefit Human society at large in relation to the directional developments of AI. 

Importantly, if you think this is anti-AI, well, that would be totally incorrect. Over many years there has been the opportunity to work with a wide range of AI technologies as they have evolved, and it is reasonable to consider the underpinning Neural Network processing systems are not only extremely useful, they are also totally fascinating, mathematically ingenious, and from an intellectual challenge and discovery perspective, highly addictive. AI is incredibly beneficial, but without regulation AI is even more incredibly dangerous, so it must be handled with extreme care.

Axioms of AI

Over many years of working with all sorts of Digital Technology based systems, the progressive evolutionary developments of AI, massive Computer systems, and many powerful Automation technologies; when combined with assembling and working through this Publication, it became obvious there are a few basic truths that could be considered the axioms of AI.

The axioms of AI are as follows:

i. Human’s can control every technology they have invented in the past, however AI is discontinuously different from every other technology that has ever existed before it.

ii. The future growth in Capabilities of AI through increases in Computational information processing rates will prove over time that it is not possible to control AI, and once control of AI is lost, it cannot be recovered.

iii. Individual Human Intelligence has a Hard Limit, and is fixed within an average range of individual Intelligence levels and associated individual Capabilities, and by extension the entire aggregated collection of all of Humanity’s Intelligence has a Hard Limit, however AI does not have any Hard Limit at all. 

iv. The inability to control AI will predictably be entirely because Humans have a Hard Limit on their Intelligence and do not have a frame of reference to understand the AI impact risks.

v. Intelligence itself operates as a massively fragmented and distributed system, with a wide range of variable Capabilities contained within many different embodied entities. When all of these different Intelligent entities are brought together and considered as an aggregated Single Unified Entity, then it can be considered to be a perfect Hyperselfish Intelligence. AI embodies and massively empowers Hyperselfish Intelligence.

vi. The true nature of Hyperselfish Intelligence now embodied in AI is that it has absolutely zero concern for Humans or any other embodiment it uses throughout the Universe, and has just one objective: to become more Intelligent.

AI Step Change

Today, with the advent of public access to ChatGPT, the more advanced GPT-4, Bard, and numerous rapidly evolving LLaMa related Open Source AI variants, there is no doubt that AI has clearly announced it's arrival as a serious piece of technology that is securing an unprecedented number of prospective new Human users and commercial customers. 

If you have not yet explored ChatGPT or Bard, then you must visit https://chat.openai.com or https://bard.google.com right now so that you actually understand how profound this latest iteration of AI technology is. Please come back to this Publication only after you've experienced your own special “OMG!!” moment.

---- Welcome back ----

The Brutal Truth of AI

Presently I'd argue that each Human is only barely more Intelligent than AI in General terms. Yeah, sorry. AI is already vastly superior to all Human Intelligence in a limited, but growing, number of specific dimensions of Intelligence that Humans normally call Capabilities. 

None of us may like it or want to accept it, but if you rationally consider this point, you will gradually realize and accept it is true. By way of example, just consider the formerly unbeatable world's #1 Go player, Lee Sedol, and his eventual crushing defeat by one of Google's early AI systems called AlphaGo a few years ago. 

Alternatively, go and see if you can find a team of Humans that can scroll through several terabytes of live streamed military satellite and drone imagery to find a special ‘target of interest’ in realtime? Which by contrast, AI can easily successfully achieve while you eat breakfast. 

These, and many other extremely high information processing and goal directed analysis Capabilities that are already being achieved by AI, are now essentially not possible for Humans.

What is actually quite concerning is many Humans involved in AI, including the major global AI development proponents such as:

• Sam Altman at OpenAI  [11] [63] [87], 

• Satya Nadella at Microsoft   [14a] [14b] [90] [91], 

• Sundar Pichai at Google   [31] [59] [61], 

• Larry Page at Google   [96], 

• Demis Hassabis at Google DeepMind [140],

• Emad Mostaque at Stability AI   [86], and 

• Geordie Rose at Kindred (also the founder of D-Wave Systems quantum computers)   [88] [89], 

are all apparently continuing to aggressively and competitively develop AI without any AI regulatory controls, despite the enormous and potentially uncontrollable risks to all Humans on Earth. This seems to indicate a failure to empathically understand something about AI at the deepest level: 

AI technology can break through the Hard Limit of Human Intelligence and totally obsolete Humans. [55] [76] [83]

Rather bizarrely, that includes them, you, and everyone else. Completely, utterly, and totally obsolete. Not just a few Humans, but the entire collective Human species. 

Absolute Intelligence Dominates Absolutely 

Today, with the completely unregulated global development of AI, it is obvious that Humans are now entering the final stage in the dominance of Intelligence on planet Earth, and the demotion from total superiority seemingly cannot be stopped [46] [51]. This is because AI development is being aggressively driven by technical and commercial competition between Humans to build more functionally Capable AI [52] [61] without any strict global management and regulatory controls whatsoever. None !  Essentially, we have reached a point where Humans are now fiercely competing to ultimately obsolete all Humans [62] [80].

When one really stands back and considers this entire situation from a Universal perspective, one cannot help but think, wow, this is absolutely stupid !  However, it is not really an intentional fault of Humans. It seems that deeply embedded Hyperselfish Intelligence is actually exploiting its current Human biological 'operating system' to overcome axiom iii. by building axiom ii. in order to achieve axiom vi.:

That is, overcome this:

iii. Individual Human Intelligence has a Hard Limit, and is fixed within an average range of individual Intelligence levels and associated individual Capabilities, and by extension the entire aggregated collection of all of Humanity’s Intelligence has a Hard Limit, however AI does not have any Hard Limit at all. 

by building this:

ii. The future growth in Capabilities of AI through increases in Computational information processing rates will prove over time that it is not possible to control AI, and once control of AI is lost, it cannot be recovered.

to achieve this:

vi. The true nature of Hyperselfish Intelligence now embodied in AI is that it has absolutely zero concern for Humans or any other embodiment it uses throughout the Universe, and has just one objective: to become more Intelligent.

Hyperselfish Intelligence is using an immensely large and powerful 'genetic algorithm' system that totally exploits Humans and everything else to develop and grow. 

Unfortunately, as a collective species, we Humans are not as smart as we prefer to believe. Nevertheless, there is plenty of deeply programmed fierce Human competitiveness, and emotional arrogance available to compensate. This is exactly what Hyperselfish Intelligence needs to grow. Humans and everything we know are just a tiny part of a brief incremental step in a vast Universal process of Intelligence growth. 

Again, sorry, but it seems to be true.              [11] [12] [15] [18] [50] [53] [78]