# Artificial intelligence as a product of algorithms

Is this approach nonsense since reason produces algorithms as a method of solving problems. The algorithm is a sequence of events that perform a task till completion of the problem. Reason is intelligence and algorithms do not substitute reason or create it.

• Cellular algorithms give rise to structure of body and mind and the mind is constructed and gives us reason to question the nature of our creation but the reason does not come from our perception of these algorithms our perception allows us to perceive of these algorithms and enablesb us to understand our creation. – Bobs Jan 26 at 1:53
• "Is this approach nonsense". Is what approach nonsense? What are you trying to solve or understand? I cannot understand the question you are asking at all. If you want to talk about reasoning or intelligence then you immediately hit the problem of needing good definitions of them, in order to have a meaningful discussion. – Neil Slater Jan 26 at 8:57
• Iam trying to understand the construction of the mind from platonic solids. – Bobs Jan 26 at 9:02
• OK, that makes more sense. I am still not sure what you are asking about algorithms though, or what the main point of the question is. Questions about constructing or deconstructing intelligence directly are very hard to compose, and a lot of effort is required in order to phrase a question that can be answered. Maybe rather than putting forward your own ideas first, you can explain what you are trying to understand, and have your ideas/thoughts as a secondary thing. If you are new to AI, you need to be finding existing thinking around the big questions that interest you. – Neil Slater Jan 26 at 9:58
• Ian trying to understand how bodies are formed and how the body encapsulates the mind? – Bobs Jan 26 at 10:16

Artificiality

Artificial means designed by humans. It is easy to explain why.

Some would propose that artificial means not man made. However, that definition has shortcomings. It brings up the question of whether always defaulting word gender to one of two genders perpetuates gender imbalance.

The verb to make is not what we mean when we use the word artificial either. People make digestive bi-products, but we wouldn't categorize defecation as artificial.

For these reasons, design by humans is the egalitarian and unambiguous criteria for artificiality.

Intelligence

Intelligence is a mental ability that arises from thought, according to most people's view. Some place an academic slant on intelligence based on the general two dimensional view that arose in college board exams: Linguistic and logical. The reductionist claim is that there is a single factor and those that have ample quantities of it can beat everyone else in everything, which is not only supportive of elitism but not at all confirmed by common experience, genetics, or actual feature extraction form intelligence test results.

None of the above three perspectives is adequately definite for the purpose of answering this question.

For simplicity's sake, this answer will proceed with this definition, which certainly has features that would lead to discussion, but it is the one this answer will be using.

Intelligence is an adaptive capacity capable of sustaining one or more of a wide but finite range of goals and avoiding conditions qualifying under one or more of a wide but finite range of calamities, based in memory and new input.

This definition has a few interesting and useful features.

• Control that has no sense of the past is not intelligence. Even a PID controller, so old that it doesn't even have a self-calibrate button, is smarter than that.
• There is always some goal, such as collapsing the universe, that is beyond the capacity of the current level of intelligence, and the smarter the intelligence, the more advanced a goal imaginable, so this will always be true.
• There is always some calamity, such as eliminating all risk of death for everyone, that is also beyond the capacity of the current level of intelligence, and will also always be true.
• Obtaining a goal but not being able to sustain it is rarely useful and not necessarily indicative of intelligence.
• Avoiding a calamity for a microsecond followed by the failure to continue to avoid it is also rarely useful and not indicative.

Algorithms and Digital Circuitry

Algorithms are process definitions in a form that executes on a Turing complete machine based on concept definitions. For instance, there is a concept called multiplication, defined mathematically as auto-addition a specified number of times in the integer domain and extended into real, complex, and other domains through mathematical history. Consider a primitive CPU and the following definition for counting numbers well below the maximum integer value for the CPU.

$$\alpha \beta := \alpha (\beta - 1) + \beta$$

This recursive definition relies on previous definitions of addition and subtraction. The algorithm, assuming a CPU that permits reading, writing, addition, subtraction, increment, decrement, test, and branch, but has no multiplication in the arithmetic unit (which is like the very first von Neumann computers).

    clear R0
label j
decrement beta
test beta
if_positive j
write R0 to alpha_times_beta


The creation of hardware to produce instructions 4 through 8 in one instruction is another transformation. We have the following transformations.

$$\text{mathematical concept} \\ \Downarrow \\ \text{algorithm} \\ \Downarrow \\ \text{hardware acceleration}$$

This is the trend, and these are some of the concepts that have made it through both transformations and now reside in VLSI chips.

• integer multiplication
• floating point addition and subtraction
• floating point multiplication and division
• 2-D rendering
• 3-D rendering
• convolution
• vector arithmetic

The middle two are what made GPUs popular for graphics cards and the last two are what extended them for DSP (digital signal processing) over which GPU algorithms were created to perform artificial network computation.

AI as a Product of Algorithms

AI is a product of mathematics. Algorithms are the first line of implementation simply because one doesn't need to develop a VLSI chip to try an idea if an algorithm can simulate a scenario the mathematical concept describes. In languages such as C or Java or Python, the algorithm is explicitly described in procedural source code. Because of compilers and libraries, the tedium of describing machine operations in assembly code is no longer necessary, but the code is still procedural.

Functional and Declarative Programming

Functional programming moves in the declarative direction. It is a level of generalization beyond compiled languages.

Declarative programming (Prolog, DRools, ECL and some JSoN or XML based declarative approaches) let a container or framework determine the algorithms from declarations closer to those of their mathematical expression.

These two approaches produce algorithms, not hardware acceleration, but the development of VLSI layouts from higher level descriptions is also in progress and has been for decades. It is possible that algorithms and programming languages will be replaced by direct compiling of mathematics to 3-D printed VLSI circuits in this century.

Brain Hardware

Brains are already similar in some ways to hardware acceleration and may have emerged that way. Algorithms and anything resembling central processing may have arose in the brain afterward, as part of the already parallel neural functioning.

We have the following transformations proposed by the current intellectual establishment, which may or may not be correct and in correct order.

$$\text{genetic functioning} \\ \Downarrow \\ \text{electrochemical communication} \\ \Downarrow \\ \text{adaptation in real time via brains} \\ \Downarrow \\ \text{emergence of cognition} \\ \Downarrow \\ \text{emergence of logic} \\ \Downarrow \\ \text{emergence of mathematics} \\ \Downarrow \\ \text{emergence of digital design}$$

Analysis of Question Elements

Now we can consider, in context, additional ideas listed in the question.

Reason produces algorithms as a method of solving problems.

Absolutely true, but also add mathematics, which is built on top of logic, the most famous contributors of which are the ancient Greeks. The specific area of mathematics that led to digital design and algorithmic approaches were pioneered by Boole, Babbage, Watt, Shannon, Church, Turing, von Neumann, and others.

[An] algorithm is a sequence of events that perform a task 'till completion of the problem.

That is not accurate. Algorithms are serialized in memory as a sequence of lines of code (for compiled and interpreted languages) or instructions (for assembly code). However, flow charts are a better representation of algorithm. They are not two dimensional in their natural form, but rather five dimensional.

Three dimensions are required to represent an arbitrary graph topology (vertices and edges), which are represented in 2-D in flow charts by allowing line crossing, which indicates that the two lines are independent and not-directly-related edges of the graph. The other two dimensions are the type of instruction (start, end, test-and-branch, operate) and the details of that general instruction category.

Reason is intelligence.

Reason is part of intelligence, but not nearly all of it. Other content in this site enumerate other features of intelligence.

Algorithms [are] not [a] substitute [for] reason.

They are certainly not currently a drop-in substitution.

[Algorithms cannot] create it.

The dominant belief held by well educated postmodern people at this time is that cosmology, geology, evolution, and symbiogenesis are the combination of natural and genetic processes that created intelligence. There are other theories and belief systems for its creation, and new ones may be introduced later.

If the dominant theory among the educated is accepted, intelligence was the result of a finite, definable, and deterministic process. Under such conditions, there exists an algorithm that can simulate the creation of intelligence. This is not just an algorithm that is intelligent, but an algorithm from which intelligence can emerge. In fact, genetic algorithm investigation began on this basis.

Provability and Gödel's Uncertainty

The proposal that algorithms can create artificial intelligence can neither be proven or dismissed by proof at this time. This uncertainty exists even if we accept the above definitions of artificial, intelligence, and algorithm, which some will not. Humanity and its mathematical tools may in months, years, decades, centuries, or millennia prove the truth or falsehood of that proposal.

It is possible that neither it nor its converse can be proven. See Gödel's work.

If other problems are not solved by humanity regarding its own collective behavior, the proposal may re-emerge when other dominant life forms on earth develop intellectual capacity. In some ways, bacterial, algae, ants, termites, and the whole of life in the biosphere already exhibit characteristics that qualify as intelligence under the above definition.

• Great answer I have but one question is our creation and our world artificial to the gods? – Bobs Jan 26 at 13:11
• Just another question what could a physical execution of a model of intelligence in that medium offer an explanation of the intangible conscious and experience of a quality associated and meaning to human dream states or experiences do you think the mind simply occurs as a consequence of the model and does not involve any involvement with a diety apportioning the quality in order experience meaning in say lucid "astral" dreams? – Bobs Feb 1 at 21:54