One of the most complex types of challenges businesses and society face is what are called “super wicked” problems. These are problems that often have multiple causes, there is no consensus regarding how to solve them and attempts at solutions can make the problems even worse. One super wicked problem businesses are currently facing is the widespread disruption of markets and ways of doing business being caused by AI. After providing an overview of super wicked problems and discussing “good witch AI” and “bad witch AI”, this article suggests that bad witch AI could be viewed as a super wicked problem and sets out a framework that could be used to address it.
Wicked Problems
Problems can be classified by how difficult they are to solve. Some problems have solutions that are well-known and can be solved with different types of resources. This does not mean that the problems are easy to solve because needed resources may be very difficult to obtain, but in many cases, the nature of a problem is understood, and there is generally wide agreement regarding how to solve it. For example, while a housing shortage is not easy to solve, the solution often involves building more houses which can be accomplished with land, construction, and financing.
Another type of problem is a “wicked” problem. These problems are much harder to solve because their causes may not be clear and there is no clear consensus regarding how to solve them or even if they should be solved at all. Wicked problems have the following characteristics:
- the true depth of the problem is not well understood until people try to solve them
- solutions to wicked problems are not right or wrong but rather represent degrees of improvement
- every wicked problem is essentially unique and there are no ready precedents that can be used as problem-solving models. The use of pre-existing solution approaches often makes the problem worse and may be a significant part of the problem.
- every solution to a wicked problem is a “one shot operation” where there is no opportunity to learn by trial and area
- wicked problems have no given alternative solutions
Super Wicked Problems
While these problems are difficult enough, they are not the hardest types of problems to solve. The most difficult problems to solve are known as “super wicked” problems. While super wicked problems have many traits of wicked problems, they also have several additional characteristics:
Time is running out. If wicked problems are not solved, they can cause significant damage, but in many cases, the depth of this damage is static or increases at a manageable rate. Because of this, wicked problems are often factored into “the way things are” and dealt with as an inescapable part of a system or environment. With super wicked problems, if the problem is not addressed, there is a real risk that its negative consequences may increase exponentially and cannot be reversed.
There is no accepted decision-making framework. Wicked problems may fall under the decision-making mandate of a particular country or body. For example, wicked problems in the area of military conflict are typically largely addressed by governments. With super wicked problems, however, there is no clear consensus regarding who is empowered to solve them. This makes it much harder to agree on a possible solution and implement it.
Stakeholder relationships are problematic. With wicked problems, stakeholders often have different views regarding how wicked problems should be solved. With super wicked problems, stakeholder relationships may be directly antagonistic, significantly complicating possible solutions. Stakeholders may be more concerned about winning the battle over adversaries than finding the best solution for all parties.
Impact of solutions. With wicked problems, attempted solutions, particularly with existing problem-solving frameworks, are often ineffectual. With super wicked problems, attempted solutions can make them worse or cause the nature of the problem to shift, requiring a significant change to the problem-solving approach.
Problem-solving perspective. Attempted solutions to wicked problems often do not fully consider the future. With super wicked problems, people are often highly focused on the short term, making it difficult to implement measures with lasting impact.
Good Witch AI
AI is currently causing a massive amount of market disruption. Some of the key areas of disruption are:
- data-driven decision making
- task automation
- experience personalization
- the application of Natural Language Processing;
- product and innovation acceleration.
From a business perspective, AI presents significant opportunities and risks. It has reduced the time it takes to perform many tasks from days to seconds, freeing up large amounts of time to perform other, higher-order tasks.
AI presents opportunities as well as risks.
Apart from time savings, it has broken down barriers to knowledge that have historically been very difficult for many people to surmount. Knowledge regarding every imaginable topic can be accessed instantly. Specialized areas of knowledge can be immediately accessed by laypeople and applied to solve complex problems.
One of the most positive components of AI is that it is a stepping-stone to even more sophisticated idea-generation and problem-solving approaches. AI can use AI to power the exploration of new solution frontiers based on massive amounts of data.
Bad Witch AI
Despite its many positive elements, AI also presents very real and large risks. Some of these risks include:
- eliminating large amounts of jobs
- making mistakes that go undetected in critical products and systems
- contributing to creation of system architecture that has vulnerabilities that are hard for humans to detect
- contributing to the creation of solution algorithms which are highly skewed because they are based on certain types of data rather than applicable data
- becoming a tool for criminal activity or harmful ends
- disseminating false, misleading, or hateful information
- becoming hard-wired into different types of systems to the point where it is difficult to completely shut down.
Bad witch AI could be viewed as a super wicked problem. Let’s consider bad witch AI within the framework of a super wicked problem.
Bad witch AI can be viewed as a super wicked problem.
Time is running out. To begin with, AI-driven market disruption is not linear. The impact of AI is speeding up at a very fast rate and, as models are trained on increasingly larger data sets and the mathematics of improving cost functions (which essentially means reducing the gaps between model predictions and model outcomes) become more sophisticated, it is likely that this will accelerate even faster.
No accepted solution framework. While there have been calls for the regulations of AI technology, there is no accepted authority for determining how it should be regulated and enforcing those regulations. This exacerbates the issue that, as time passes, AI market disruption is increasing exponentially.
Stakeholder relationships are problematic. Stakeholder relationships are problematic, and these problems will likely escalate as data scraping becomes more invasive, jobs are lost and the negative consequences of AI become more apparent.
Elusive solutions. Apart from deep and likely escalating stakeholder antagonism, AI solutions are elusive because AI can create new types of problems that require different solutions. Further, as AI becomes increasingly hardwired in operational and critical tasks, backtracking may cause damages that outweigh at least the perceived short-term risk of leaving the technology in place.
Responses discount the future irrationally. AI responses, by definition, discount the future irrationally because the impact of AI cannot be accurately forecasted. This means that all responses only address a part of what will rapidly become a much larger challenge.
A Framework for Addressing Bad Witch AI
While an analysis of how businesses can manage the tension between AI opportunities and risks, the following are several steps that businesses can take to manage bad witch AI.
Establish a comprehensive AI policy. Establish a comprehensive AI policy that clearly distinguishes between AI opportunities (good witch AI) and potential risks (bad witch AI), and outlines immediate steps for responding to adverse AI-related events. This policy should evolve with new insights and AI developments. Set up a regular review process, such as quarterly, to keep the policy aligned with the latest AI advancements.
Appoint an AI risk officer. Appoint a dedicated AI risk officer or task force responsible for implementing and updating the AI policy and action plan. Ensure that the AI risk officer collaborates closely with cybersecurity and compliance teams to ensure that AI-specific risks are comprehensively managed.
Communicate the AI policy to all stakeholders. Proactively communicate the AI policy to internal and external stakeholders, ensuring transparency in how AI is being used and the safeguards in place. Conduct regular briefings or workshops to educate stakeholders on potential AI risks and how the organization is addressing them.
Audit business model and systems for AI vulnerabilities. Conduct a thorough audit of the business model and all critical systems to assess potential vulnerabilities or adverse impacts AI could introduce. Regularly update audits to capture emerging AI threats and ensure contingency plans remain relevant.
Ringfence critical systems and data. Ringfence critical systems and secure sensitive data to safeguard integrity, implementing robust access controls, monitoring, and backup protocols. Consider using AI-specific data security solutions that detect anomalies and potential intrusions in real-time.
Apply design thinking to customer journey mapping. Utilize design thinking to map out customer journeys, focusing on how AI disruptions might alter customer experience and identifying new ways to create value during these shifts. Regularly revisit and refine customer journey maps to adapt to evolving AI impacts and customer expectations.
Key Article Points
- problems can be classified by how difficult they are to solve
- two categories of difficult problems are wicked and super wicked problems
- AI-driven market disruption can be viewed positively (good witch AI) or negatively (bad witch AI)
- AI risks can be viewed as super wicked problems
- companies should build a framework to protect themselves against potentially damaging AI disruptions.
Very fitting for this piece, the image for this article was generated by AI.