Is Artificial General Intelligence a Mathematical Pattern?
Examining the science and philosophy of human and artificial intelligence.
Posted March 28, 2018
What if the key to unlocking artificial general intelligence is a pattern that already exists, but has yet to be discovered? Artificial general intelligence, also referred to as “strong artificial intelligence” or “full artificial intelligence” is the ability of a machine to perform human-like cognition. What seems to be a straight-forward philosophical question is actually quite nuanced. Clues to the answer can be found in an interdisciplinary examination of computer science, math, philosophy, physics, synthetic biology, and neuroscience.
Artificial intelligence (AI) is a term that lacks a single unifying definition. The simplest explanation is that AI is machine intelligence versus biological human intelligence. AI is in the early stages of development, despite being a concept that is over 60 years old—the term originated in a Dartmouth publication circa 1955 . The resurrection of AI is largely due to recent trends such as the falling cost of computing, the rise of powerful cloud-based decentralized computing, the availability of big data for machine learning, and the increasing sophistication of computing algorithms. Today computer science technology enables machines to perform functions such as problem solving, learning, planning, reasoning, and recognition of speech, voice, images, and handwriting. Currently AI is more of a tool for point solutions—far from strong artificial intelligence.
If achieving artificial general intelligence is indeed a pattern that already exists, uncovering it involves mathematics, the science of patterns. Mathematicians look for patterns to form a conclusion, called a conjecture, and set out to support the proposition by creating a proof, or theorem. For example, mathematician Shinichi Mochizuki of Kyoto University published a proof called the Inter-universal Teichmüller theory (IUT theory) of the abc conjecture, one of the unsolved problems in math number theory. In both computer science and mathematics, algorithms are procedures for solving a problem. Computer science is inherently mathematical with corresponding methods in which to provide sets of instructions for machines. For example, today computers are able to “learn” from data sets, or teach themselves concepts. Machine learning is a subset of AI where the computer “learns” without explicit programming. The learning algorithms may be based on regression, instance, regularization, decision-tree, Bayesian, clustering, association-rule learning, Artificial Neural Networks, Deep Learning, dimensionality reduction, ensemble, and many other types of analysis .
Is math merely discovered, like an excavation by an archeologist on a dig, or invented, like an inspired poet? Mathematical Platonism is a metaphysical view that mathematical truths are discovered, not invented—mathematical objects are abstract and exist independently of our having the ability to think or describe it . Metaphysics is a branch of philosophy that is concerned with the fundamental nature of reality and being, which includes ontology (the study of the nature of existence), cosmology (the study of the origin and evolution of the universe), and epistemology (the study of knowledge and justified beliefs). If an object has an associated mathematical formula, then it is theoretically possible to express it in a computer algorithm. If math is a reality unto itself that awaits identification, would that imply that everything has a corresponding mathematical formula? Critics of Mathematical Platonism would argue that numbers are concepts that exist when the mind conceives of them.
Human consciousness can be described as a state of awareness, and being aware of one’s thoughts and surroundings. Can consciousness be programmed? Physics is a natural science that studies the nature and interaction of matter and energy, and mathematics is the tool of choice for physicists. Cosmologist, physicist, and Massachusetts Institute of Technology (MIT) professor Max Tegmark argues that consciousness is a mathematical pattern that can be understood as a state of matter with information processing capabilities . Using the analogy of the different states of matter (solid, liquid, and gas), Tegmark puts forth the concept that consciousness is also a result of an emergent phenomenon. He calls this state “perceptronium” . If consciousness is a pattern, in theory, a machine can be conscious if one ascribes to Tegmark’s hypothesis.
How formulaic is life itself? Can life be programmed? To answer that question, we need to look no further than at the recent breakthroughs in synthetic biology. The J. Craig Venter Institute created the world’s first synthetic life form with an entirely synthetic genome, a self-replicating bacteria called the Mycoplasma mycoides JCVI-syn1.0 in 2010 . The genetic code for this new species was digitized on a computer, then assembled biochemically . Life can be created with synthetic DNA inserted into genome-free bacteria. This was a single-cell organism. The next step in synthetic biology would be to synthetically create self-replicating multi-cellular organisms—a complex, and ambitious undertaking.
Will humans one day become an amalgam of artificial and biological intelligence? How realistic is a brain-computer interface (BCI)? Entrepreneurs and business moguls are entering the neuroscience market. Various approaches to uncovering how the human brain works include the use of optogenetics, fMRI, imaging, electrophysiology, high-resolution optics, genetics, spectroscopy, and biochemistry. The world’s first neuroscience accelerator, NeuroLaunch, was launched in 2014, serial entrepreneur and venture capitalist Bryan Johnson founded Kernel with $100 million US dollars of his own fortune in 2016, and billionaire Elon Musk entered the neuroscience market with the launch of Neuralink in 2017 . In January 2017, an innovative breakthrough in BCI was achieved by a research team led by Niels Birbaumer, a neuroscientist at the Wyss Center for Bio and Neuroengineering in Geneva, Switzerland. Researchers were able to communicate with patients with amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig’s disease, who had “locked-in syndrome” using functional near-infrared spectroscopy (fNIRS) that “measures brain hemodynamic responses associated with neuronal activity .” Four ALS patients were trained to regulate their frontocentral brain regions to transmit “yes” or “no” answers to questions. Responses were measured by the relative change in oxygenated hemoglobin (O2Hb), with reported results of an “above-chance-level correct response rate over 70% .” This was a first-of-its-kind study that paves the way for future brain-computer interfaces.
Technological singularity is the concept where machine intelligence exceeds the capability of human intelligence. If this can be achieved, what does this mean for the future of humanity? The answer to this question has profound implications for the future. Whether or not the universe is inherently mathematical, humans are advancing toward unlocking the mysteries of physics, consciousness, artificial intelligence, neuroscience, and life itself.
Copyright © 2018 Cami Rosso All rights reserved.
1. Rosso, Cami. “Why AI is Trending Now.” Medium. Feb 21, 2017.
2. Brownlee, Jason. “A Tour of Machine Learning Algorithms.” Machine Learning Algorithms. November 25, 2013.
3. Linnebo, Øystein. “Platonism in the Philosophy of Mathematics.” Stanford Encyclopedia of Philosophy. July 18, 2009. Revised Jan 18, 2018.
4. Tegmark, Max. “Consciousness as a State of Matter.” Chaos, Solitons & Fractals. Submitted on 6 Jan 2014 (v1), last revised 18 Mar 2015 (v3).
6. Smith, Michael. “Scientists Create First ‘Synthetic’ Cells.” ABC News. May 21, 2010.
8. Rosso, Cami. “Why Neuroscience is Trending in Business.” Psychology Today. Mar 20, 2018.
9. Chaudhary, Ujwal; Xia, Bin; Silvoni, Stefano; Cohen, Leonardo G.; Birbaumer, Niels. “Brain-Computer Interface-Based Communication in the Completely Locked-In State.” PLOS Biology. January 31, 2017.