Conceptual framework
At this moment of technological singularity, developing people and organizations that make AI their strategic co-thinker to address complex problems means taking their capacity to frame, understand, model, decide and implement solutions beyond anything ever possible in our history.
It all starts with a skill that can be learned: asking meaningful questions. Without it, we delegate. With it, we think better.
Three positions towards AI
The author's synthesis, building on Bevilacqua and colleagues (2026).
Absent
Does not use AI and is left out of what it now allows.
Delegates
Hands over the thinking to AI, not just the task. Works for AI.
Strategic co-thinker
Formulates meaningful questions and opens, with AI, a space to iterate reflectively with AI.
Five phases to solve a complex problem
| Phase | The question | What AI contributes as a co-thinker | If it is delegated to |
|---|---|---|---|
| FrameOpening up the problem | What is the real problem, the one that matters and not just the urgent one? | Suggests other framings and perspectives, and tests the frame. | The wrong problem gets solved efficiently. |
| UnderstandOpening up the problem | What causes it and whom does it affect? | Synthesizes sources, maps relationships and points out what is missing. | Information piles up without understanding. |
| Model solutionsOpening up the problem | What paths are there, beyond the obvious one? | Widens the space of options and combines ideas from distant fields. | The first plausible idea is accepted. |
| DecideSolving it | Which one do I choose, and what am I risking? | Argues against, simulates scenarios and exposes biases; judgment stays with the person. | What the machine proposes gets signed without taking responsibility. |
| ImplementSolving it | How do I carry it out, and how will I know it works? | Designs plans, anticipates failures and helps read the results. | Things get done without learning from the result. |
Where the framework comes from
The five phases synthesize established models of how people solve complex, ill-structured problems. Mumford and colleagues (1991), as synthesized by Mumford and colleagues (2010), describe eight processes of creative problem solving, from defining the problem to appraising the solution. Jonassen (1997), as cited by Kim and Clariana (2024), describes seven activities for solving ill-structured problems, which begin by setting the problem space and end by implementing and adapting the solution. Fischer, Greiff and Funke (2012) understand complex problem solving as acquiring knowledge about a system and applying it, and highlight five requirements: generating information, reducing it, building a model, deciding in a changing system and evaluating.
These models agree on two movements. First the problem is opened up: it is framed, understood, and alternatives are generated. Then it is closed: a decision is made and implemented. Throughout the whole cycle, metacognition is at work: the ability to monitor one's own thinking.
Where AI comes in
Bevilacqua and colleagues (2026) studied how AI changes the skills of top managers and describe an evolution from task delegation to co-thinking with AI (p. 8). They note that using it well requires asking the right questions, contextualizing problems and thinking critically about the results. Building on that work, this framework proposes three positions towards AI, “absent”, “delegates” and “strategic co-thinker”, and extends the view from executives to anyone who works, studies or decides.
Jonassen observed that solving well requires generating as many alternative representations of the problem as possible, and that this is precisely what students find hardest. That is where AI as a co-thinker can contribute most: in the movement that opens up the problem. That AI expands each phase is, for now, a working hypothesis.
On “modeling solutions”
In the literature on complex problems, building a model usually means representing the system, which here belongs to the understanding phase. In this framework, modeling solutions means designing and structuring solution paths before deciding.
References
- Bevilacqua, S., Ferraris, A., Matzler, K. and Kudej, M. (2026). Strategic leadership at high altitude: Investigating how AI affects the required skills of top managers. Journal of Business Research, 205, 115878.
- Fischer, A., Greiff, S. and Funke, J. (2012). The process of solving complex problems. The Journal of Problem Solving, 4(1), 19-42.
- Jonassen, D. H. (1997). Instructional design models for well-structured and ill-structured problem-solving learning outcomes. Educational Technology Research and Development, 45(1), 65-94. Cited from Kim and Clariana (2024).
- Kim, K. and Clariana, R. (2024). The influence of resource interdependence during problem solving in groups: Tracking changes in knowledge structure. Accepted manuscript, National Science Foundation public repository, https://par.nsf.gov/servlets/purl/10504626.
- Mumford, M. D., Mobley, M. I., Uhlman, C. E., Reiter-Palmon, R. and Doares, L. M. (1991). Process analytic models of creative capacities. Creativity Research Journal, 4(2), 91-122. Cited from Mumford and colleagues (2010).
- Mumford, M. D., Antes, A. L., Caughron, J. J., Connelly, S. and Beeler, C. (2010). Cross-field differences in creative problem-solving skills: A comparison of health, biological, and social sciences. Creativity Research Journal, 22(1), 14-26.