The Potential of Artificial Intelligence
Placing Artificial Intelligence in the context of history makes clear its incredible potential
Human civilization can be traced as a progression towards increasing economies of scale across many domains. Most other trends can be reduced to this central thread. The most important and fundamental of these domains is thought: social, political, and economic systems have increasingly joined human thought together into a more cohesive unit. But there is a hard upper limit; we remain confined to our own minds, each thinking separately. We cannot merge our minds, or significantly expand our individual cognitive capacities. Artificial Intelligence provides the unprecedented opportunity to overcome the most significant barrier of economies of scale, and thus the advancement of human civilization, that we face. Understanding AI in this larger context drives home the truly massive potential the technology carries.
Economies of scale: The hallmark of civilization
Economies of scale refers to “the reduction of production costs that is a result of making and selling goods in large quantities,” according to the Cambridge Dictionary. The concept originates in economics, but it can be widely applied by generalizing the definition: a larger unit is more efficient than the sum of many smaller units of equal total capacity would be.
When examining some of the hallmarks of modern civilization, across a range of domains, economies of scale explains the improvements. It would be impossible to trace in detail such developments across history, and of course there are many exceptions and temporary reversals, but analyzing some of modern civilization’s hallmarks can give an idea of the pervasiveness of the concept.
- Healthcare
In healthcare, we have moved from independent doctors to large hospitals that centralize the provision of healthcare and large pharmaceutical companies and government research labs that centralize the development of healthcare. Many, if not most, of the advances in healthcare stem from, or at least were facilitated by, an increase in economies of scale. It makes intuitive sense; if healthcare is provided in a large institution like a hospital, it makes sense to have specialized equipment for rare diseases and processes can be standardized and streamlined. Unifying research in large pharmaceuticals allows the best minds to have access to more advanced technologies and laboratories.
2. Manufacturing
A similar trend is evident in manufacturing. Before the Industrial Revolution, the little manufacturing that did take place occurred in small units, limited to localized regions or even a single home. Think of blacksmiths, people sewing their own clothes, and local bakers. The Industrial Revolution, in essence, enabled economies of scale in manufacturing, which is why it was so revolutionary. Factories introduced massive economies of scale into manufacturing, making the process far more efficient and allowing for specialization, large dedicated machinery, optimized supply chains, etc. The steam engine facilitated this centralization by making transportation less costly and faster.
There are many other instances of increasing economies of scale, such as in business, trade, culture, and governance. Progress towards economies of scale does not always proceed linearly, but when looking at history it becomes clear that the overall trend points towards increasing economies of scale.
Thought, the fundamental domain
Ultimately, all human achievement flows from our minds, and so thought can be seen as the foundational domain of all other aspects of civilization. Economies of scale in thought is more fundamental than in other domains, and has ripple effects. It is also a more complex, comprehensive process.
Humanity has made significant progress in increasing economies of scale in thought. The patent system facilitates the contribution of individuals to humanity’s collective knowledge without fear of theft. Data storage and access methods, starting with the book and culminating today in databases and the internet, make it ever easier for individuals to add their contribution to an ever larger pool of knowledge, which in the modern age encompasses all of humanity, and to access that knowledge. Our educational system now draws from the widest possible pool of talent, where it used to be limited (first to the aristocracy, then to whites, then to men, etc), allowing a fuller realization of human potential to contribute to society. Thought, in terms of its output and input, has become more centralized and accessible, providing economies of scale.
My thought is not your thought
But we are all still individual Homo Sapiens, just as our cave-dwelling ancestors millennia ago. We think separately, and have to laboriously communicate the results of our thinking through language, which distorts and simplifies. There is no central executive that can regulate disagreements in thought across society and come to a final conclusion; the closest we have is the government and the market, neither of which have access to our actual thoughts. Although we have achieved economies of scale in thought through the creation of better systems of interaction, there is an upper limit to what can be achieved.
This is a massive source of inefficiency. Entire systems, such as the legal system, most of economics, and our political systems, are designed to try to create economies of scale in thought without being able to actually combine the units, namely every one of us. These systems are immensely complex, and many different versions have been tested throughout history, often with disastrous consequences (think Communism in the Soviet Union and China, or the Great Depression). War stems from the lack of a central executive. By ‘central executive’ I mean an entity that has control over and complete access to the routines of each sub-unit. This cannot be the case in a human system; an individual will always retain his independence of thought. Disagreements are thus inevitable.
Imagine if everyone could unite their cognitive capacity and their knowledge into a single unit of thought, encompassing the sum of all human processing power and memory storage. The resulting entity would almost instantly be able to reach conclusions on a wide variety of topics, and understand concepts that are incomprehensible, even unimaginable, to us.
A ‘core problem’
Drawing a parallel to manufacturing may help illuminate the concept. Before the Industrial Revolution, each village had its own blacksmith. Imagine if the machinery used was fixed in place and could not possibly be moved or created anew. You would have to devise a complex system to maximize economies of scale using the existing blacksmiths. You would seek to optimize the transportation network, figure out the best way for supplies to reach all blacksmiths as efficiently as possible, introduce specialization so that economies of scale can at least be achieved with each individual blacksmith, devise an educational system for blacksmiths, etc.
Now if the equipment of the blacksmiths is not immobile and non-replicable, as is the case in reality, all one has to do to achieve economies of scale is build a large factory. The metaphor illustrates the concept: the barrier to unifying thought is a ‘core problem’ in the sense that solving it would nullify the need for the systems designed to sidestep it.
Strong Artificial Intelligence — A parallel system of greater efficiency
Enter Strong Artificial Intelligence, or AI that operates at a level similar to or beyond human capability across most cognitive domains. AI is not a move towards economies of scale in the sense that we as humans are improving our own systems, but rather in the sense of the creation of a new, parallel system that can achieve what our own systems can’t.
Before continuing, it should be noted that it is assumed that ‘strong Artificial Intelligence’ is feasible. This is true because of the law of universality of computation. The reasons for this are addressed at the beginning of this article. We are focusing on the rationale behind trying to develop a strong AI, not its feasibility.
Why would a strong AI be the ultimate economies of scale in thought? For two reasons.
- AI will have vast cognitive abilities.
An AI will have access to vast computing power.
Humans can hold seven items in our working memory at a time; an AI could hold millions. There is no hard upper limit; hardware expansion (ie increasing RAM memory), software improvement, and new technology can all increase the size of an AI’s working memory. The same applies to long-term memory storage.
2. AI will have direct access to the sum total of knowledge.
Humans must laboriously access knowledge through external mediums, such as books or the internet, and process them. We can store only a limited amount of data in our long-term memory. We must thus pick and choose what we learn. As a human species, by definition, we know the sum of all human knowledge; but that knowledge is spread thinly across 7 billion individuals and countless books and other repositories. An AI would have access, at least initially, to the same data as the human species; but the crucial difference is it would have access to it as a single unit, allowing for vast economies of scale.
Discovering the theory of life, the universe, and everything
So what are the advantages of economies of scale in thought?
Just imagine if your own cognitive power had the same economies of scale. Instant access to all human knowledge, the ability to hold thousands of items in your working memory at once and draw connections between them, vastly improved computing power. And, although not addressed in this essay, most probably also a superior neural architecture better able to synthesize knowledge. Imagine the understandings you could reach about theoretical physics, the nature of the universe, the psychology of the human mind, the organization of our social systems, the way the Earth’s ecosphere is intertwined. And this all within the first few hours, or even minutes, and naming only what we are familiar with. What lies beyond this short timespan? We can hardly begin to conjecture; our human minds simply aren’t powerful enough.
The implications
So what are the implications? Will AI allow us to radically improve the human condition and achieve in mere days what would otherwise have taken centuries? Will it herald a new era in human history? Or does such an immensely intelligent entity pose an existential threat?
The truth is, no one really knows. Artificial Intelligence is so different from any other technology we have ever developed, and there is no precedent to look to. Never before has there been an alternative, more advanced intelligence aside from our own.
But this does not mean we should ignore such debate while we speed ahead with AI research.
Nor does it mean we should halt all research on AI.
We should proceed, but cautiously.
The potential is incredible; the risk equally so. Let us ensure we maximize the former.
Originally published on The Influencer.