Part 3. AI — A New Approach To Research, Innovation And Entrepreneurship

Credit: [1]

“…it is the scale of physics where life emerges, but that life itself is a broader phenomenon recurring across different scales, from chemistry to cells to societies–which more generally concerns the interactions of information (an abstract property) with matter.” [2]

In Part 1 of this series I discussed how artificial intelligence (AI) has advanced to the point where it creates new data for analysis, abstracts the insights from the synthetic data and takes on much of the cognitive role previously reserved for humans. In Part 2 of the series we probed further into the concepts of abstraction, insight and synthetic data and introduced the possibility of AI surpassing human capabilities. In this Part 3 I discuss how we need to re-examine creativity so that humans continue to add value in the face of the advances in AI. Evolution has graciously programmed every young child (not starving to death) to be naturally creative as part of their instinct to explore, iterate and learn. Much research shows that this creativity is programmed out of the child through formal education by the time the child reaches 12th grade. Shaping creativity for the 21st century involves the recognition that the Industrial Age and its wealth creation model(s) is all but over. The old set of systems, practices and culture that shaped incredible human advances in the last three centuries needs to be re-examined, questioned and probably shelved. Hopefully, this article begins such an examination.

Creativity

Most of us have never taken a course in creativity and few have studied it. My experience with students is that most cannot define creativity. Never have I found a student that knows the difference between imagination and creativity. Most people, when asked to give an example of creativity, offer a song, a piece of art or perhaps a film as examples. Few offer examples from mathematics, science or engineering. The failure to teach creativity and understand its importance in math, science, engineering and the social sciences is particularly troubling at this moment in history when humans for the first time face possible extinction.

The cognitive scientist Margaret Boden defines three types of creativity [3]. The first is a “novel combination” of existing components or variables. I think this is what Einstein meant by “combinatorial play”, which may have been derived from Poincare’s notion that invention first was the selection from creative alternatives. Boden’s other two definitions of creativity refer to (1) “conceptual spaces [or] structured styles of thought” derived from one’s culture(s) and (2) the exploration of these spaces. I do not see added value in Boden’s “conceptual spaces”. A conceptual space has certain “components” or “variables” that can be redefined or changed. Boden and I agree that this component nature of creativity lends itself to approaches such as AI. We further agree that the argument about whether artificial intelligence is “creative” is not important. The AI is producing novel results that have value. Whether the computer was purposeful in seeking out the novelty is really of little concern except to philosophers. Nevertheless, the issue of whether computers are expanding their ability to replace humans is very real. However, I believe a problem or question is not properly understood until it is stated in the positive. Therefore, the better question is — how will humans be creative and create value…in a future where computers and artificial intelligence increasingly duplicate human capability?

If this in-depth educational content is useful for you, subscribe to our AI mailing list to be alerted when we release new material.

Research and Innovation

As one might expect, AI is not standing still and every major advance appears to be massively publicized to promote the technology, attract capital, promote share price or all the above. These advances are all based on two features of artificial intelligence — the ability of the computer to (1) recognize patterns typically for the purposes of (2) prediction. The advances typically identify more complex patterns and increasingly provide the basis for new insights, knowledge and understanding about complicated subjects. Juergen Schumidhuber makes the point well:

“Since short and simple explanations of the past usually reflect some repetitive regularity that helps to predict the future as well, every intelligent system interested in achieving future goals should be motivated to compress the history of raw sensory inputs in response to its actions, simply to improve its ability to plan ahead.” [4]

Of course, this “repetitive regularity” is the same exploratory, iterative behavior that humans were programmed with through evolution as the foundation for how we learn. This type of behavior was documented by Piaget and many other researchers. However, it should be recognized that pattern recognition alone is rarely valuable and herein lies the continuing importance in the role of humans. It is in the focus on the role of humans in this AI-assisted search for the novel patterns that we redefine both learning and research and make clear the human value add. Throughout this three-part series of articles, I have constantly returned to questions of epistemology and metaphysics. Respectively, what is our theory of knowledge and what is real (reality). I think the historic concept of reality developed in the 17th century by Descartes, Spinoza, Leibniz and others is out of date and an updated concept will better enable us to creatively make use of AI to produce innovative results. I have consistently advocated for a modern concept of reality grounded in quantum physics, complexity science and systems thinking.

The irony is that the AI tools we have available are better suited today for this updated metaphysics than humans. If we look at what we know [5]:

(1) Creativity is the combination of existing components in new ways. This concept is nothing more than the recognition that reality is based in sub-atomic and atomic components that combine to create everything from molecules to human and manmade systems.

(2) Herbert Simon’s concept of synthesis showed us that these natural and manmade components can be combined “creatively” in novel combinations to solve problems

(3) John Holland helped us to realize that much of nature and manmade systems can be understood from a limited number of rules or laws. “This perpetual novelty, produced with a limited number of rules or laws, is a characteristic of most complex systems: DNA consists of strings of the same four nucleotides, yet no two humans are exactly alike; the theorems of Euclidean geometry are based on just five axioms, yet new theorems are still being derived after two millennia; and so it is for the other complex systems.” [6]

(4) Capra and Luisi complete the necessary groundwork by showing us that all dimensions of human life are systems that are complex non-linear networks. “A central characteristic of the systems view of life is its nonlinearity: all living systems are complex — i.e., highly nonlinear — networks; and there are countless interconnections between the biological, cognitive, social, and ecological dimensions of life.” [7]

When we digest this summary, we see all the parts of computational creativity, as defined below:

“The combinatorial perspective allows us to model creativity as a search process through the space of possible combinations. The combinations can arise from composition or concatenation of different representations, or through a rule-based or stochastic transformation of initial and intermediate representations.” [8]

AI is ideally suited to model this natural computational environment.

Economic History

I want to introduce the concept of innovation now because I want to make clear that creativity has little value if it cannot be turned into invention (a tangible process, method, composition or design) [9] and then offered to create value in people’s lives. To paraphrase the legendary economist Joseph Schumpeter, “innovation is invention commercialized”. The discussion of innovation brings the focus back to humanity and solving people’s problems. We face daunting problems this century in the environment, a “cold war” with China, the issue of wealth inequity and the “social determinants” that affect every social issue. Because of the magnitude of these issues and the increasingly interconnected world we live in, the need for new science and engineering is paramount. But, I also want to make clear that the days of “science for science sake” are basically over, as evidenced by the National Science Foundation (NSF) strategic focus on translation — bringing innovative science and engineering to market. The NSF was founded in 1950 to define the U.S. national research strategy and fund it. Over the years the NSF has increasingly emphasized the commercialization of the research to achieve scalable social and economic impact. The NSF also recognizes the multidisciplinary nature of the current problems [10] and the need to integrate AI into research and commercialized solutions [11].)

What we realize is that as the new science became clearer, so did the problem(s) — how to harness the combinatorial nature of natural and manmade components to create solutions. As Bryan Arthur predicts, the tools appear to solve the problems of the times and machine learning, neural networks and genetic algorithms are all examples of the tools required to produce these combinatorial solutions. We even have graph neural networks to better understand biological and social connectivity. Before we get too euphoric, let’s pause and consider what is happening in economics.

Schumpeter, Thomas Kuhn, Carlotta Perez and many others have researched scientific and industrial revolutions. One of my favorite researchers is Nikolai Kondratieff, a Russian economist, who described the waves or cycles of technological innovation [12], as shown below.

Man began to satisfy needs and create economic value using raw materials and primitive tools. This economic model for wealth creation lasted until the 18th century when energy was introduced as an input to increase efficiency and the speed of production. Through the introduction of multiple forms of energy and the related technologies, we increased the scope of innovation until the end of WWII. Funded by wartime needs, computers emerged as commercial tools in the 1960–1970s. Based on Claude Shannon’s Information Theory and the natural proclivity for networking, computers spawned a new wealth creation model based in information. The replacement of energy, land and material as the key economic inputs had begun. Computing advanced rapidly as people realized the value of information and the competitive advantage to be derived. Beginning about 2005 researchers and industry realized that the AI launched at Dartmouth in 1956 just needed more data and not more computing power to have real scientific, economic and social value. The AI began to be applied to new fields such as synthetic chemistry, computational biology and eventually the production of its own synthetic data. Today we can replace matter and food with synthetic materials, model production processes predictively to reduce energy consumption and increase efficiency and use synthetic data to give us new insights in molecular biology, genetics, materials science and a wide range of other fields. Effectively, a new economic model for wealth creation has emerged — Synthesis — which uses AI to shape data as inputs, outputs or both in many different types of new systems, processes and platforms to create and engineer new solutions to old problems and new problems just discoveredthrough AI and synthetic data. Another popular way to look at this new wealth creation model is as the fusion of the physical, natural and digital into a system of information, an updated version of John Wheeler’s seminal “It to Bit” (that was explained in Part-2 of this series of articles).

If we combine this economic history lesson with the earlier summary of the state of science, creativity and innovation, we could shape a plan for the future built around the following concepts:

(1) Every problem is now an information problem.

(2) Instruction in science, engineering and mathematics should be shaped by quantum physics, complexity and systems thinking.

(3) The inherent complexity in natural and manmade systems lends itself to a multidisciplinary approach to learning, research and innovation.

(4) Applied AI is a key component of any multidisciplinary approach to research, understanding the complexity of issues and developing the necessary innovation.

(5) The existential issues facing humanity prompt a need for scientific research to be commercialized; the private sector has the vast resources to bring to bear to solve the problems. The urgency of the problems requires more than just a government and NGO responsibility for the issues.

Many are concerned about the AI taking away human jobs and another group is concerned about the discriminatory bias in the data AI needs to operate. Yet another group is worried by the increased electricity usage and environmental issues from the related data storage devices. Many groups flame these issues and the silly politicians on both sides just add fuel to the fire. Every technology since the steam engine has had positive and negative consequences and we have managed to creatively channel these technologies to improve the standard of living. However, the advances have also put us on the verge of extinction as we continue to exhaust the finite resources of the planet.

The noted French philosopher Michel Foucault said, “To do the impossible, you have to see the invisible.” It is in the human ability for pattern recognition, analogy and insight that we will find the creative solutions to today’s problems. The AI will simply support these efforts by speeding up the analysis, the determination of findings and the ability to predict. The real issue is whether we can channel the AI to help find the solutions in time.

I am confident that we will continue to develop new algorithms, which will require us to teach more math, statistics and applied machine learning to children at a younger age. These new algorithms will produce new knowledge in science, medicine and environmental studies. This new knowledge will lead to creative new technological solutions to problems and many of these solutions will be component-based approaches, frequently at the nanometer scale. I am hopeful that this knowledge will greatly extend life expectancy and finally enable us to improve the living standard for the people that have been left behind for far too long. As quantum computing and nuclear energy advance, I am even more optimistic about this forecast.

The biggest risk to my optimism is the increase in natural disasters and the continuing degradation of the environment. Humans will probably survive thanks to the new technologies, but the quality of life could be drastically changed or reduced. Also threatening my optimism is the increasing incidence of dictatorship and extremism. This form of government stifles innovation and suppresses people’s natural desire for improvement. Recent events in China and Ukraine are examples of the negative consequences of dictatorship. We need a better geopolitical environment if we are going to have any chance to properly address the environmental issues. Without government interference — individual empowerment, creativity and innovation can flourish to solve the problems of the times. Ideally, for everybody.