Complicated, Complexity and the Possible Pitfalls of AI in Business

Do you think that complicated and complex mean the same thing?

It's probably a bit of an unfair question, considering the dictionary basically tells you they are!

Complicated:

Composed of elaborately interconnected parts; complex

Complex:

Characterized by a very complicated or involved arrangement of parts, units, etc.

This idea however, that these two things mean the same, is one of the biggest challenges we face in the business world today.

Why?

Because it allows business leaders to use the wrong type of approach to 'solve' the type of problem identified (assuming of course that they have correctly identified the problem).

Rick Nason, an associate professor of finance at Dalhousie University’s Rowe School of Business, explains the differences between these two types of problems clearly.

Complicated problems may be hard to solve (or may even seem unsolvable at a particular point in time) however they are addressable with a systemic or process driven approach. You are able to break down the problem to the individual components and apply rules, to see the output based on a particular input.

To put this into a context most of us in the business world are familiar with, complicated problems are those types of problems that can be addressed with an approach like LEAN1.

Complex problems however are actually quite different.

These problems are the result of networks of cause and effect, that cannot be individually teased out and addressed. A complex problem is not simply the sum of the constituent parts, they involve too many unknowns and interconnectedness which cause the separate components to behave in unpredictable ways.

A great example of a complex problem? Raising a child.
For those of you out there with more than one child, does it not blow your mind how different they are? Even though you swear you 'treat them the same', that somehow they react and behave differently to similar circumstances?


The challenge for most people, is that when they are faced with a problem, they default to a "complicated" problem solving model, irrespective of the actual nature of the issue in front of them.

The key difference that leaders need to take on is to "manage, not solve" when facing complexity.

Psychology and Behaviour

If the answer is so easy, that all we need to do is approach a complex problem by "managing it, and not solving it", why do most people not do it?

Cognitive biases.

Every day, we make hundreds of decisions. These decisions range in their scope and importance, from potentially the most important you have ever made in your life right down to where you will put something on your nightstand.

In order to cope with this constant requirement to make a choice, your brain has developed an array of cognitive biases.

Cognitive bias is a limitation in objective thinking that is caused by the tendency for the human brain to perceive information through a filter of personal experience and preferences. The filtering process is called heuristics; it’s a coping mechanism that allows the brain to prioritize and process the vast amount of input it receives each second. While the mechanism is very effective, its limitations can cause errors that can be exploited.2 

Cognitive biases are not necessarily bad, in fact, you probably would not be able to get through a typical day if these biases weren't working to help with the cognitive load; that does not mean however that some don't cause us challenges in the modern world.

It's important to remember that a lot of these patterns of thinking evolved and contributed to the survival of our species. We have been evolving for approximately 6 million years, with 'modern' humans dating back as much as 200,000 years; so even the last 2,000 years of 'doctrine' represents only 1% of our modern development.

A lot of these biases can be connected to our need for certainty. I have written about certainty before3 so I won't go into too much detail, certainty addresses our fundamental need for safety and control. It's the need for basic comfort, the need to avoid pain and stress, and also to create pleasure. Our need for certainty is a survival mechanism.

Here are a few of the biases that link (for me) to the ideas of complicated and complex and why our default is to start thinking about "solving" the problem (as opposed to managing it):

Ambiguity (uncertainty) aversion
Ambiguity aversion, or uncertainty aversion, is the tendency to favor the known over the unknown, including known risks over unknown risks

Availability heuristic
Availability is a heuristic whereby people make judgments about the likelihood of an event based on how easily an example, instance, or case comes to mind

Choice overload
Also referred to as ‘overchoice’, the phenomenon of choice overload occurs as a result of too many choices being available to consumers

Confirmation bias
Confirmation bias (Wason, 1960) occurs when people seek out or evaluate information in a way that fits with their existing thinking and preconceptions

Endowment effect
This bias occurs when we overvalue something that we own, regardless of its objective market value

Herd behavior
This effect is evident when people do what others are doing instead of using their own information or making independent decisions

Optimism bias
People tend to overestimate the probability of positive events and underestimate the probability of negative events happening to them in the future

Overconfidence (effect)
The overconfidence effect is observed when people’s subjective confidence in their own ability is greater than their objective (actual) performance

Present bias
The present bias refers to the tendency of people to give stronger weight to payoffs that are closer to the present time when considering trade-offs between two future moments

System justification
The tendency to defend and bolster the status quo. Existing social, economic, and political arrangements tend to be preferred, and alternatives disparaged, sometimes even at the expense of individual and collective self-interest

Weber–Fechner law
Difficulty in comparing small differences in large quantities


Those are A LOT of biases! I don't blame you if you skipped most of the list. The point I was trying to highlight was to introduce you to how many biases could be at play without you even realizing what is happening in your own head.

Business Impacts & Artificial Intelligence

You are probably wondering what then could possibly be the impacts to business, and especially to what a lot of people see as a 'saving grace' to these types of problems, AI.

With limited human capability to track and compute all of the different interactions of a complex system, surely the solution would have to be to get a machine to do it instead.

The major risk that I see really comes down to reliance.

As we integrate more and more of our workflows through massive AI systems to 'assist' the human component, we will most likely see people, teams and leadership get "lazy".

Giving up our critical thinking tasks in favour of a system developed / driven solution can create huge gaps in understanding for an organization (really getting and appreciating 'why' we do something).

A few more obstacles to be aware of:

• Data: Quality data is key for these systems to perform. Garbage in / garbage out
• Data Weighting: Systems we have now don't know how to distinguish good or relevant data from data or downward trending data
• Data Outliers: Tendency to exclude data 'anomalies' as inconsequential
• Lack of creativity / out of the box thinking
• Significant implementation costs (which may lead business to favour solutions to justify the spend)

The biggest question for me however is how agile and adaptable systems built on AI will be. Our current situation battling the CoronaVirus has shown us the devastating impacts of a complex problem. An AI driven system may have actually done a better job at predicting the spread earlier, made better suggestions as to when people should start social distancing and have already anticipated shortages in key PPE on the frontlines; or maybe the deliberate manipulation of the initial reported cases in China would have caused the system to do next to nothing ... who knows.

Another major challenge is the human capability for conceptual thinking. To make wild connections and to create something new. There have always been those in our societies that seem to be able to pick up on or create the newest trends. To know at some level that there is an opportunity for change, or that change is on the horizon and accelerate that is remarkable (and usually done with little to no hard data).

Conclusion

I believe that it is imperative for the business leaders of today (and tomorrow) to continue to leverage the human element. To take advantage of our ability for creativity, insight and empathy in order to truly create antifragile systems.

The CoronaVirus has shown us just how connected our world now is. We don't have the luxury to sit in one place and not care what is happening in other parts of the world.

Fear is a powerful motivator. There will no doubt be businesses that will favour machines over people, that feel that a system would not have made the same 'mistakes' that the people did and that we would have been better off.

Maybe.

At the end of the day however, there is nothing on this planet as adaptable as our species. Our ability to change, grow and evolve is unparalleled and should not be pushed aside so easily. The systems, the AI, the connectedness that we have created, these are tools and need to be used that way.

Let's focus our attention to better understanding the complicated systems we can control and the complex ones that we need to manage.

Create workflows and structures that are antifragile so that we don't find ourselves on the frontlines of the next pandemic with a message flashing across the screen simply stating, "cannot compute, insufficient data".