The answer isn’t clear, but it will certainly be evident when we look back half a century from now. Technological breakthroughs, such as increasingly and surprisingly sophisticated artificial intelligence algorithms, make us question whether we’re living through a fundamental transition point in history.
It’s worth noting that a common aspect across all industrial revolutions is the sudden increase in the complexity of value creation ecosystems that underlie the economy, completely transforming the way we live, do business, and ultimately exchange value. By value creation ecosystems, I mean all the organizations, components, and interconnectedness required to transform matter and information into something of value, which is not always only of the monetary kind.
It is also self-evident from previous industrial revolutions that the only way to properly manage complex systems is through the responsible application of intelligence, whether artificial or biological. The fundamental question now is whether our biological, and collective, intelligence is enough to manage the increasing complexity that surrounds us. When we consider recent history, especially the misgivings of the past century, I believe the answer is no.
To qualify this hypothesis, we need first to define complexity and establish a way to measure it. Of course, it’s not that simple. According to Ray Kurzweil, complexity can be defined as “the smallest amount of meaningful, non-random but unpredictable information needed to characterize a system or process.” An inherent property of complex systems is that the rules describing the system at the local level can be extremely simple, but the emergent behaviour can be rich and often surprising.
To measure the complexity of any given system, the following attributes can be helpful:
- Size: The number of components;
- Connectivity: The number of connections between components;
- Hierarchy: The number of levels of organization;
- Diversity: The variety of components and their relationships;
- Circularity: The presence and number of feedback loops;
- Adaptation: The ability of a system to change and adapt in response to external factors;
- Emergent Properties: The emergence of new and unpredictable behaviour at the system level.
Considering past industrial revolutions, it would be a fair assumption to say we have been mastering the first three attributes whilstignoring the critical importance of the last four. Some of the pressing challenges we’ve been facing as a global society are a strong signal we failed to manage, for instance, the diversity of human behaviour and expectations, feedback loops in financial networks, large-scale adaptation to a new virus, and the unintended consequences of using fossil fuels like air pollution and global warming.
“There are no hard problems, only problems that are hard to a certain level of intelligence.” — Eliezer S. Yudkowsky, Staring into the Singularity, 1996.
Is it the case that the level of intelligence required to deal with all these aspects of complex systems may be beyond our human capabilities? A high-level analysis of business models may provide some insight into the question.
Traditional business models have been built based on a linear approach to value creation, like a production line, granting the organization full control over their entire value chain, sometimes with significant entry barriers, like regulation and capital requirements. Of course, that was a logical way to manage any business given the transaction costs associated with the use of internal resources were much lower than using third parties. In addition to that, customer choice was quite limited given economic “benefits” of mass production, mass marketing, and mass education.
However, the advent of the internet has driven transaction costs close to zero, and customers now have more choice than ever, with stores having infinite shelf space. These two change vectors — declining transaction costs and increasing customer choice — have demolished our old assumptions about value creation, given that customer lifestyles now cut across many industries (e.g. banking, transportation, consumer goods, media, etc.), creating an unprecedented demand for interoperability, something the upcoming Web 3.0 will only accelerate. Hence, it is a good moment to reconsider the biological constraints of our human capabilities to manage complexity and its unintended consequences.
We still believe that having a large volume of information can help us make better decisions, which is true in some cases. However, we are facing an increasing number of business scenarios where this is not necessarily true. We may be extremely well-informed as individuals, but at the same time, we may be collectively blind, given the biological and inefficient mechanisms we use to communicate, produce, and exchange knowledge. Think about it: we need to invest years and even decades to master specific knowledge domains, whilst more knowledge is created at an exponential rate. Like the Red Queen, we need to keep learning to stay in the same place.
So, what does all of this mean? In my view, it means we are stumbling on our own virtues as humans, and maybe, just maybe, our biological intelligence is reaching its limits in its ability to solve emerging complex problems, collectively.
Artificial Intelligence language models like GPT-3 will enable traditional businesses to be reinvented and disruptive business models to emerge, creating new markets and demanding a whole new professional skill set. It is an unknown territory we will need to explore, a new territory that may be fertile enough for us to sow the seeds of global prosperity.
The first reactions to such powerful technology are often misplaced. Will this thing replace my job? Is this the end of universities? What if this thing starts to replicate itself? And on and on. In my view, these initial reactions are missing the point, as if we had strong evidence to suggest that our biological intelligence, collectively, can address the extremely complex challenges we face today in business and as a global society. Just consider the United Nations Sustainable Development Goals to be addressed by 2030. It’s not only a matter of human effort or financial investment. These are extremely complex challenges that require an entire new approach to problem-solving.
We have evolved and educated our children to always find the right answer to a specific problem, and this ability gave us the power to dominate the planet, land on the moon, invent the unimaginable but also to cause unparalleled destruction. The advent of AI turns this on its head. Now we have all the answers we are looking for, but what is the question? Or more specifically, what is the prompt?
Now we have a powerful tool in our hands, and according to Spider-Man, that means our level of responsibility is bigger than ever. By the way, that’s why ChatGPT is stuck in 2021. AI is calling human ingenuity forward. What specific problem do we really want to solve instead of just virtue signalling? What piece of scientific, fact-based insight will unlock the solution? What is the safest and most sustainable way to implement the solution without shortcuts? Can an artificial brain contribute to the debate?
Well, it all depends on the data with which we feed the model. The benefits of AI-based solutions will be directly proportional to the quality of the data that underlies sophisticated language models like Davinci, the most powerful model known to date. When I say quality, I’m not only referring to the technical aspects of data but also intangible aspects like ethics, bias, morality, inclusion, fairness, etc. In other words, inherently human aspects.
The question now is whether we can use AI tools based on a framework of human values and ensure that AI models generate solutions consistent with our most cherished principles. These are challenging questions to answer. Therefore, the value businesses can create by leveraging AI is fundamentally dependent on both the quality of the data (which is more abundant than ever) and human discernment (which is arguably in short supply) and cannot be codified or automated (although AI based on Quantum Computing may prove me wrong).
I believe that intentionally combining these elements will unlock new avenues of economic growth that are not currently self-evident, similar to how being a web designer or a mobile app developer made no sense for most people in the late 1980s but emerged as a new field as entrepreneurs and large corporations ran new experiments and challenged the status quo.
I’m truly excited about what AI-based tools will enable and how their capabilities will transform business models, infusing digital and physical products and creating new ones. As we embrace this new technology and explore use cases that will benefit from its power, let’s consider the attributes of complex systems that AI can help us to manage. This approach will enable the relevant application of AI and maximize its business benefits and value for customers and society, while creating more bandwidth for humans to drive and accelerate creativity and innovation — the key ingredients of our future economic growth.