Humans Fall in Love with Solutions—AI Can Help Fall in Love with Problems

Why augmenting problem exploration with artificial intelligence may be the biggest yet underused lever for innovation

One of the low-hanging fruits of using artificial intelligence to transform how we work is harnessing its power to help people solve problems faster, but most importantly, more creatively. I have written extensively about it (start here). For example, one of the main epiphanies of the last years has been how artificial intelligence can help critique our ideas, which is strengthening the idea flow at the end of its funnel - area where many people are often lacking either because of their skills or because they tend to avoid critiquing others work too directly.

Here, I want to go to the other end of the idea flow, the upstream part of problem-solving and creativity that, as we will see, is a key determinant of the quality of whatever happens downstream. I expand on an earlier article about how AI can help us discover which problems to solve. In this essay, we discuss what comes next, leveraging both current scientific understanding and practitioner experiences.

Innovation efforts often jump straight to brainstorming fixes, seduced by the “dopamine hit” of a clever solution. Innovation facilitators, for instance, know the struggle with keeping working teams focused on problem-exploration exercises, so that they do not end up paying lip service to it before moving on to the "real work". Yet theory and evidence remind us that the quality of the solution space is bounded by the quality of the problem space we first explore. Artificial intelligence now offers a practical, high‑return way to strengthen that front‑end work: accelerating, broadening, and systematizing problem exploration while keeping humans firmly in charge of purpose and judgment.

What the Research Already Tells Us

There is a reasonably extensive corpus of research on this. (To be frank, I would've expected more, but research is seemingly skewed the same way practitioners are—we focus more on solutions than on problems.)

Over the past 10–15 years, scholarly and managerial literature on innovation management has converged on the critical importance of problem-space exploration—the thorough investigation, framing, and (re)formulation of the problem itself—before moving into solution generation. Research on design thinking, creative problem solving, and strategic problem formulation demonstrates that teams and organizations that invest in clarifying and reframing the problem systematically produce more original and higher-impact ideas.

1. Time Spent Framing Correlates with Creativity
2. Multiple Problem Frames → Diverse Ideas
3. Deep User Insight Spurs Novelty
4. Reframing Techniques Lift Solution Quality
5. Reflection & Debate Outperform Rush‑to‑Closure
6. Problem/Solution Co‑Evolution Drives Breakthroughs

Across design thinking (Liedtka, 2015; Micheli et al., 2018), creative problem‑solving (Abdulla et al., 2020), and strategic management (Nickerson et al., 2012), the message is consistent: a well‑defined problem is half, or at least a big part of, the innovation.

Now let's break down the issue to identify places where we can solve it. A common framework for disciplined and thorough ideation, Design thinking’s Double Diamond highlights two macro phases: Diamond 1 – Problem Exploration and Definition; Diamond 2 – Solution Generation and Delivery.

Our focus is on the first diamond, where early framing determines everything that follows. Anecdotal evidence from AI‑assisted ideation projects conducted through the last few years suggests material gains in speed and depth of insight when humans partner with AI during this phase.

A few assumptions grounded on practice can guide us here.

The central hypothesis of this work, supported by early evidence collected using artificial intelligence-assisted ideation technologies and practices, shows that AI, used as a cognitive partner in Diamond 1, enables a more comprehensive and insightful definition of the problem space than unaided human work—ultimately yielding solutions that are both more novel and more useful.

AI, used as a cognitive partner in Diamond 1, enables a more comprehensive and insightful definition of the problem space than unaided human work—ultimately yielding solutions that are both more novel and more useful.

Humans still supply strategic intent and critical evaluation; the machine delivers rapid, wide‑angle exploration that would be prohibitively slow or narrow if done manually.

Why the AI-Human Partnership Works

Both humans and machines are bound by our capabilities (knowledge, logic) and incentives (hormones in the human brain, tokens in the machines). Unsurprisingly, human limitations reflect themselves in organizational barriers.

AI workflows, such as those in AI chats, are aligned to follow our instinct to move fast to solutions. They also risk so-called "institutional knowledge replication," that is, staying well within the "known-knowns" instead of venturing further.

However, they can be reconfigured by asking for more reflection time (and tokens) on the problem. The schemas below, part of an overall process illustrated in a previous article, show exercises that can help with that.

Consider the following examples, where artificial intelligence can help delve into the problem and take different perspectives. Similar opportunities can be unlocked in science and R&D, among other fields.

Once again, obtaining the best results requires synergy between artificial intelligence and human capabilities. People act as principled "system 2" (in Daniel Kahneman's terms) to the faster "system 1" thinking of the machines. The machines, especially if well configured and using the latest reasoning models, can also prevent us from often falling into our own System 1 thinking. (More on this here.)

Artificial intelligence, if designed around a user experience that supports, not substitutes, the human, can help here, for instance, through:

• Bias Buffering – AI can surface alternative framings that counter premature convergence.
• Perspective Multiplication – Large models digest cross‑domain data, exposing angles a single team may miss.
• Structured Decomposition – Algorithms break complex challenges into tractable sub‑problems.
• Multidisciplinary Scanning – AI links insights across industries, functions and sciences.
• Novelty Detection – Pattern‑spotting uncovers surprising anomalies worth reframing around.
• Knowledge Augmentation – Synthesized evidence bases raise the floor for non‑experts.

For more details and practical guidance, please review the previous essays.

Conclusion: Better Questions, Better Innovation

A decade of empirical work underscores a simple truth: the creative ceiling of any innovation effort is set early, when we decide what problem to solve. Humans tend to move to solutions too fast. Artificial intelligence, if instructed appropriately, doesn't suffer from the same bias or have the same dopamine hit. On the contrary, it could be incentivised to spend time understanding problems well.

As a result, if used well, with competent humans firmly in the loop, AI now gives organizations a scalable means to deepen that decision. By pairing human strategic judgment with machine‑driven exploration, teams can, among others:

• Surface hidden angles of attack, for instance, variables and stakeholder needs in hours, not weeks.
• Counteract cognitive biases that silently narrow the search space.
• Build a shared, evidence‑based understanding that accelerates alignment.

The result is a richer portfolio of solution avenues and, ultimately, more original and valuable solutions. Companies that cultivate disciplined, AI‑augmented problem framing are not just “doing design thinking faster;” they are upgrading the very substrate of innovation, ensuring they invest in solving the right problems before investing in solving them well.

This essay is part of a series on AI-augmented Collective Intelligence and the organizational, process, and skill infrastructure design that delivers the best performance for today's organizations. More here. Get in touch if you want these capabilities to augment your organization's collective intelligence.

References

• Abdulla, A., et al. (2020). Problem finding and creativity: A meta‑analytic review. Psychology of Aesthetics, Creativity, and the Arts, 14(1), 3–14.
• Baer, M., Dirks, K. T., & Nickerson, J. A. (2013). Microfoundations of strategic problem formulation. Strategic Management Journal, 34(2), 197–214.
• Carlgren, L., Rauth, I., & Elmquist, M. (2016). Framing design thinking: The concept in idea and enactment. Creativity and Innovation Management, 25(1), 38–57.
• Dorst, K. (2011). The core of “design thinking” and its application. Design Studies, 32(6), 521–532.
• Johanssen‑Sköldberg, U., Woodilla, J., & Çetinkaya, M. (2013). Design thinking: Past, present and possible futures. Creativity and Innovation Management, 22(2), 121–146.
• Liedtka, J. (2015). Linking design thinking with innovation outcomes through cognitive bias reduction. Journal of Product Innovation Management, 32(6), 925–938.
• McKilligan, S. & Creeger, S. (2018). Strategies to redefine the problem exploration space for design innovation. Proceedings of E&PDE 2018.
• Micheli, P., Wilner, S., Bhatti, S., Mura, M., & Beverland, M. (2018). Doing design thinking: Conceptual review, synthesis, and research agenda. Journal of Product Innovation Management, 35(5), 740–758.
• Nickerson, J. A., Yen, C. J., & Mahoney, J. T. (2012). Exploring the problem‑finding and problem‑solving approach for designing organizations. Academy of Management Perspectives, 26(1), 52–72.
• Seidel, V. & Fixson, S. (2013). Adopting design thinking in novice multidisciplinary teams. Journal of Product Innovation Management, 30(S1), 19–33.
• Studer, J., Daly, S., McKilligan, S., & Seifert, C. (2018). Evidence of problem exploration in creative designs. AI EDAM, 32(4), 429–444.