Weiser, A., & Von Krogh, G. (2023). Artificial intelligence and radical uncertainty. European Management Review, 20(4), 711–717. https://doi.org/10.1111/emre.12630
Summary
In recent years, artificial intelligence (AI) has become a game-changer, not only for data analysis and automation but also for how organizations make decisions under radical uncertainty. Radical uncertainty refers to situations in which it is impossible to know all the variables, possible outcomes, or even the scope of what might happen. Under these conditions, typical decision-making models (like assigning probabilities to outcomes) can be of limited use.
Bounded Rationality and AI
Herbert Simon introduced the concept of bounded rationality to describe how human decision-makers operate under imperfect information, limited time, and cognitive constraints. We cannot always optimize our decisions because we lack the ability or the data to consider every possible scenario. Instead, we do the best we can with the time and resources available.
Weiser and von Krogh suggest that AI can help relax some of these bounds. That does not mean AI will replace human judgment in every scenario; rather, it can augment our ability to handle complex and ambiguous decision situations by:
- Providing faster, more comprehensive scanning of data and patterns in the environment.
- Suggesting alternative solutions based on vast databases and combinatorial logic.
- Creating simulations or “imagined” scenarios, allowing humans to test ideas that they wouldn’t normally consider.
- Speeding up the feedback loop when experimenting with new strategies.
Even so, AI is not a silver bullet. It needs sufficient data to generate insights, and it also relies on human oversight—particularly when business owners, managers, and teams must factor in social and ethical considerations or weigh intangible data.
Core Decision-Making Stages
Weiser and von Krogh outline five sub-processes of decision-making and show how AI can be integrated:
- Agenda Setting
- Traditionally, a manager tries to spot issues or opportunities by scanning the internal and external environment.
- AI can help by sifting through large volumes of social media data, customer feedback, sensor data, and more, flagging emerging threats or unexplored opportunities much faster than a human could. For example, an AI system might detect a spike in negative customer sentiment about a new product, prompting a manager to act early.
- Problem Formulation
- Once an issue is identified, the next challenge is understanding its root causes and defining it properly.
- AI’s role could involve analyzing multiple data sources (e.g., competitor launches, macroeconomic data, user behavior) to clarify what’s causing a sales drop or supply chain disruption. This helps managers avoid biased assumptions and see a more accurate picture of the “problem space.”
- Identification of Alternatives
- In complex scenarios—especially under radical uncertainty—managers might not know all the potential solutions.
- AI can assist by using pattern recognition and generative models to present a range of possible actions. For instance, in drug discovery, AI can scan chemical properties to propose novel treatments. In a consumer-goods firm, AI might suggest new product variations based on market data. Importantly, AI can expand the solution search, but managers still need to assess which options fit their strategic goals and ethical standards.
- Selecting Alternatives
- Even if AI proposes a set of options, the final choice typically requires human judgment. Factors like corporate values, regulatory constraints, or stakeholder reactions may not be fully captured by algorithms.
- AI does add value, however, by simulating scenarios. A company considering a major investment can have AI generate different market or regulatory outcomes. In this sense, managers get a preview of how each possible path might play out, aiding more informed selection.
- Evaluation and Adaptation
- Decision-making doesn’t end once a plan is chosen. Under conditions of radical uncertainty, success often depends on rapid learning and adaptation.
- AI can streamline data collection and quickly analyze results so that leaders can pivot faster. For instance, real-time pricing algorithms can continuously adjust based on market responses, or A/B testing can be automated for marketing campaigns, giving near-instant feedback on performance.
AI in Team Decision-Making
The article draws connections to broader research on group dynamics. When AI is introduced as a “team member,” it might be perceived as having high technical ability but lacking the emotional and interpersonal qualities of human members. This can lead to trust issues or lower satisfaction if not managed carefully. Therefore, business owners need to think about how to integrate AI “opinions” into team discussions so that AI’s outputs add to the collective intelligence without undermining human collaboration.
Risks, Limits, and Legitimacy
AI has limitations. It relies on available data, which might be incomplete or biased, especially in uncharted territories where we have no historical precedent. Overreliance on AI can create a false sense of certainty if managers assume that an AI’s predictions are always accurate. Moreover, teams and stakeholders may question the legitimacy of decisions driven largely by opaque AI “black-box” models. Communicating transparently and ensuring human oversight helps maintain trust and accountability.
Possibility and Imagination
Interestingly, the authors highlight how AI can enhance creative, imaginative thinking in radical uncertainty. Instead of just crunching probabilities, AI can combine data in unexpected ways, leading to novel insights. Tools like generative AI (e.g., advanced language or image models) can propose out-of-the-box ideas. Still, a purely machine-driven approach might miss the intangible, human-driven leaps of logic—like emotional resonance or moral considerations—that can be vital in pioneering truly novel solutions.
Practical Takeaways
- Improved Scanning: AI tools can detect weak signals or patterns early, helping companies spot market shifts before they become crises.
- Structured Problem-Solving: AI can expedite root-cause analysis, saving time and clarifying the scope of issues to address.
- Expanded Solution Space: Generative algorithms can go beyond “obvious” solutions, proposing creative or unconventional paths.
- Scenario Modelling: By running multiple what-if simulations, AI can provide a range of plausible outcomes, helping managers weigh risks and trade-offs.
- Faster Feedback Loops: AI-driven analytics can rapidly confirm or disprove assumptions, making real-time adaptation more feasible.
Conclusion
Weiser and von Krogh argue that artificial intelligence has a unique ability to augment human intelligence when dealing with situations of deep or radical uncertainty. AI helps relax certain cognitive and resource constraints, providing faster, broader, and more flexible ways to gather information, generate alternatives, and adapt strategies.
However, they also stress that humans remain essential to the process. Ethical, contextual, and emotional judgments still demand human insight. AI’s power is most fully realized when managers understand its strengths and weaknesses—and design a decision-making environment that captures the best of both human creativity and machine-driven thoroughness.
10 Practical Insights for Business Owners and Managers
- Monitor New Threats and Opportunities Automatically
Employ AI-driven analytics to keep a constant pulse on external factors (e.g., social media sentiment, competitor filings, customer feedback) so you can spot market shifts or emerging risks early. - Use AI to Clarify—Not Replace—Human Judgment
Treat AI suggestions as a second opinion, not a substitute for decision-makers’ experience, ethics, and stakeholder understanding. - Leverage Data for Better Problem Definition
AI can sift through vast data sets to show you patterns and correlations, enabling a more accurate problem statement and keeping you from jumping to the wrong conclusions. - Expand Your Range of Solutions
Generative AI models, powered by large data sets, can propose non-traditional ideas. Encourage your team to evaluate these ideas creatively rather than dismiss them outright. - Run Multiple Scenario Simulations
In highly uncertain environments, let AI map out potential outcomes under different assumptions—economic changes, regulatory shifts, or competitor moves—to see how each scenario could play out. - Implement Rapid Experiments
Use AI tools to collect feedback on new initiatives quickly. For instance, set up A/B tests or real-time monitoring so you can pivot fast if the data suggests your approach isn’t working. - Balance Speed with Deep Analysis
AI accelerates data processing, but radical uncertainty sometimes requires deliberation. Know when to slow down for careful reflection on intangible risks or ethical concerns. - Ensure Transparency and Legitimacy
Black-box AI models can alienate employees and customers if they don’t understand them. Explain how AI-generated inputs influence final decisions to build trust and acceptance. - Cultivate an AI-Ready Culture
Train teams to interpret AI outputs, question them critically, and combine them with on-the-ground knowledge. This means fostering an environment where people welcome data-driven insights but keep a skeptical mindset. - Stay Alert to Biases in AI
AI is only as good as its training data. Periodically audit AI outputs to detect bias or skew. Combine machine suggestions with diverse human perspectives to maintain fairness and precision.
Key Takeaway:
AI is not a magic wand, but it offers robust analytical capabilities and “imaginative” potential that can help businesses navigate deep uncertainty. By combining AI’s strengths in data handling with human-driven ethics, adaptability, and creativity, organizations can significantly enhance the quality and speed of their decisions—even when facing unknown challenges.