Future of Computing: Logic to Ideation
The future of Artificial Intelligence (AI) has been in the news a lot of late. Some very informed folks like Elon Musk and Stephen Hawking are warning us to take it slow. And we should heed their advice.
My own concern about AI is immediate—it’s here—because it applies to all digital computing. While I am concerned about killer robots, I am equally concerned that we have created computers in our own image.
Specifically, we have created computers to “think” like we do. For must of us that is a decidedly sequential process that starts with input and logically deduces output through the incremental refinement of commonality. As René Descartes put it in his 1637 publication, Discourse on the Method, “Je pense, donc de suis”. (“I think, therefore I am.”)
As every LEAN manager will tell you, process is everything. And thought is no exception. The process by which we link input and output, to a very large extent, pre-defines the latter. Any process of incremental refinement will itself limit the range of ultimate conclusions through arbitrary and structured exclusion. That’s how it works.
One of the byproducts of the thought process most of us employ, as a result, is the ultimate tendency to see all choice as essentially binary. I think of it as the either/or trap. We tend to see meaning in the distinct data points along a line rather than the more expansive geometry that linear output defines the boundary of.
Avoiding the shades gives us clarity and simplicity but often obscures hidden meaning and innovative solutions that lie entirely outside the existing paradigm. It is in this shaded area of thought where creativity resides. And it is in this area of creativity that we will find the best solutions to today’s problems as well as the launch point for entirely new and yet unforeseen opportunities.
Computers, as we currently construct them, are particularly well suited to the thought process of incremental refinement. They can digest more and more data at increasingly faster rates. That makes digital computing perfect for applications like driverless cars, production automation, coding, and serving fast food.
Life is defined by dichotomies, however, and computing is no exception. In addition to introducing the risk of killer robots, codified computing, as we currently know it, also contributes to the current political and social divisions that are crippling the country. Computers may not yet be pulling the trigger of mass destruction, but they are already redefining our culture, our politics, and the workplace.
It is technology, for example, that empowers both expansive and selective input. Instead of everyone tuning in to Walter Cronkite for our nightly news, we can get personalized news from a wide variety of sources that each interprets events in ways that mirror our own. With choice comes exclusion.
Beyond the unintended consequences of developing computers that think like we do, I am equally concerned about the opportunities lost if we don’t develop computers that can think in more holistic and free flowing ways. Instead of incrementally refining data, what if we created computers that could speculate back from the conclusion in a process of open ideation? It’s a more comprehensive approach that computers would seem to be ideally suited for given the sheer computing power available. And while the process may not lead to concise and well-defined conclusions it may provide the creative insight we need to then see and solve problems in new and innovative ways.
In attempting to solve the problem of X, for example, we may observe that the computer inevitably wants to go in a direction that seems more or less random at first. With further refinement, however, we may see a link that we hadn’t previously considered. That, in turn, may open the door to a whole universe of logical solutions that can be deduced in the traditional way.
As I argue in my recent book, Understanding Business: The Logic of Balance, and the soon-to-be-published Understanding Life: Context is Everything, understanding is the first step to forward progress. It is the process by which we connect the two, however, that determines potential. We shouldn’t limit either our thinking or our computing, therefore, to any single process of thought.
About AI, I have found Amazon's CEO (Jeff Bezos) description to his shareholders nails it just right: “Over the past decades computers have broadly automated tasks that programmers could describe with clear rules and algorithms. Modern machine learning techniques now allow us to do the same for tasks where describing the precise rules is much harder.” This is a great way to describe the promise of AI: I am sure you'd confirm that this is still far below what we can do as humans...
Gary Moreau, I guess we think more and more like computers, not the other way round. We find it more and more convenient to think in a binary way : we reduce our reading of the world to the oppositions between good and evil, just and unjust, order and chaos, taggressor and victim ... Thinking the world in these binary oppositions can be legitimate in some respects but this is not what sets us apart from AI. To think is to question the evidences, search for the hidden under the appearances, evaluate the means necessary to reach an end, anticipate the possible and unlikely outcomes, and prepare to adapt accordingly. Lean management principles are indeed perfect to predict outcomes from known input variables, and seek continuous efficiency along this model. MECE (Mutually Exclusive, Collectively Exhaustive) has been and still is the bread and butter for consulting firms. The assembly of a Boeing 787 Dreamliner requires about a thousand tengineers at its Everett plant in Washington. But the factory depends on wings manufactured in Nagoya, Japan. Horizontal stabilizers of Foggia, Italy. The doors for freight and passenger access from Sweden; The landing gears from Gloucester, UK. Like most of what we buy or use, A Dreamliner is the product of an extensive network of international subcontractors. Boeing like Airbus or Apple, are becoming more efficient by outsourcing their production process to third parties, but this outsourcing also means that production in a dozen countries on three continents also has the power to disrupt their operations. The successful assembly of an airplane depends on safe shipments through oceans, acceptable working conditions in Japan, a political risk with North Korea or Russia, stable exchange rates and flexible management of the supply chain. Lean managament is great to deal with complication, but not so much with complexity: in the 21st century, the consequences are devastating: terrorism, cybercriminals have benefited from the speed and interdependence of our economies to cause terrible damages, which are not predictable. Weather can be predicted in a particular city tomorrow with relative accuracy, but not in six months. It can be reliably anticipated that inflation will get retailers to increase their prices this month, but we don't know if it would cause recession in a year. This is where AI can't beat thoughtful strategic thinking.