Start with Questions, not with Answers:
In the age of agentic AI, humans are being pushed upstream: from completing tasks to being able to define them with precision, and from producing answers to designing the right questions to spark action.

In most classrooms today, the student who finishes fastest while producing the “right answer” is the one rewarded. But what happens when an AI agent can do that same thing within seconds? The skill that seemed most valuable – in this case, efficient production – suddenly becomes a cheaper commodity, especially in an AI-driven economy.

The importance of knowing things will never lose its intrinsic value, but in the age of AI, there is a shift taking place in terms of which skills will add the most value in the near and far future. In the case of producing the right answer efficiently, the skill’s economic value, at least, seems to be shifting. It’s not necessarily about what you know, and in its place, valuable human contribution may be moving toward skills like being able to craft the right questions in order to identify and frame problems creatively. After all, “advances in AI have caused a seismic shift from a world in which answers were crucial to one in which questions are. The big differentiator is no longer access to information but the ability to craft smart prompts.”1

This is why “encouraging students to ask questions is not new but has been increasingly important as students interact with artificial intelligence systems.”2 It’s why Mount Vernon School, an organization whose mission is to be “a school of inquiry, innovation, and impact,” recently partnered with IBL expert, Trevor MacKenzie, to infuse moments of inquiry, joy, and wonder in every PK-12 classroom on its Atlanta day campus. When there’s abundant access to answers (some trustworthy, some not so much), what we need are students who know how to ask critical, novel questions.

Building off our previous article, “Learning in the Age of Agentic AI: When Machines Execute, What Becomes of Mastery?” which discussed the shift from task execution to task stewardship, the shifts we’re paying attention to here are as follows.

  1. A shift from the importance of answer production to an emphasis on question architecture.
  2. A shift from doing the work to being able to define and frame the work

The Slow Elevator Problem & the Mirror Solution:

Asking questions often helps us reframe a problem and identify new ways forward. Yet, answers are what we seem to be demanding in terms of student outputs in traditional school contexts.

Thinking back to our modern day classroom where the student who produces the correct answers the fastest receives the reward, certain biases are being formed among the minds of the learners. It’s why “we often overvalue answers [in the real world] and undervalue the framing of questions.” Yet, we now live “in a world where AI generates abundant answers, [meaning] this bias can be costly. [However,] we continue to reward knowledge holders, even when the value has shifted to those who can define the problem in the first place…”3

Innovation expert and consultant, Thomas Wedell-Wedellsborg, once shared a story he called the “The Slow Elevator Problem” in a classic HBR article, “Are You Solving the Right Problems?”

In the article, he asks us to imagine being the landlord or owner of an office building where tenants are complaining that the elevator is too slow and that it needs to be fixed. The shared assumption is that the elevator is slow.

Install a new lift! Upgrade the elevator motor! The solutions couldn’t be more obvious, or so it seems.

However, when you bring the problem to some building managers, they suggest a simple fix that reframes the entire question or problem. By putting up mirrors where people had to wait, tenants would be more likely to lose track of time due to distractions by their own likeness. Mirrors were the simple solution once the problem was framed differently.

“Creative solutions,” writes Wedell-Wedellsborg, “nearly always come from an alternative definition of your problem.”4 Alternative definitions of the problem enable us to discover and explore new “solution spaces.” By reframing the problem as one about the wait being intolerable instead of the elevator being too slow, there is space for alternative, novel solutions (such as putting up mirrors or playing relaxing music). Dr. Zorana Ivcevic Pringle, creativity scholar and writer, supports this understanding in her work, The Creativity Choice, claiming that alternative definitions and reframings are exactly what spark humans’ capacity to be creative.

Creative problem-solving, in such cases, “involves the generation of novel and valuable ideas. Novel solutions are original ideas that depart from existing knowledge, and valuable solutions are useful ideas that can be implemented to yield economic and social returns.”5 But in order to have “novel ideas,” one must architect and frame questions creatively because “breakthrough ideas rarely start with answers. Instead, they start with better questions… In environments defined by uncertainty, [people] cannot simply rely on existing answers. Instead, they must develop the ability to frame the right questions.”6

One reason this shift is happening is because agentic AI is so fast at producing answers and executing tasks. Another reason is that AI is not as creative, novel, or effective at posing questions that get us to “depart from existing knowledge.” Human cognition “is inherently forward-looking and theory based, enabling humans to transcend data and prediction to generate new data and observations and conduct experimentation.”7 AI models, on the other hand, “still tend to occupy more incremental search spaces that are proximal to existing solutions rather than exploring highly novel spaces, possibly because of the models’ training methods on past data, or formal rationality.”8 Humans are inherently creative in the sense of producing truly novel content, whereas AI is good at scanning previously known things and recombining them into new, coherent patterns.

The constraint is in the questioning rather than producing answers, and for this reason, humans have much value to add in an area where humans are being pushed upstream. “When it’s easier than ever to generate answers and insights,” writes Sangeet Paul Choudary, “value migrates to the upstream work of framing the right inquiry… Curiosity, the ability to frame the right question before seeking answers… become[s] increasingly valuable as the ability to generate answers becomes commoditized by AI.”9

This is why Khan, Fisher, and Frey emphasize that, in the age of agentic AI, “students need to develop both cognitive skills and personal dispositions that support inquiry and exploration. Key skills include asking thoughtful questions, observing carefully, identifying patterns, making connections between new and prior knowledge, and tolerating ambiguity.”10 When humans combine with AI, many of these “cognitive skills” become “fusion skills” – such as what Daugherty and Wilson call intelligent interrogation, or, the skill of “knowing how best to ask questions about AI, across levels of abstraction, to get the insights you need.”11 Both authors go on to claim with confidence that such “fusion skills” are now table stakes for current and future entry level workers.

Defining the Work – The Creative Importance of Problem Finding:

The shift from answer production to question architecture points to where human value can be found in both the current and future workplace and where we need to cultivate competency in our current K12 students. And the value of questioning relates to so much more than simply finding answers or solutions. “Getting answers,” as one author writes, “is just one of the many reasons we have to ask questions.”12

We also ask questions to spark creativity and to discover, frame, and even reframe problems in novel ways that no machine could have imagined.

In Dr. Zorana Ivcevic Pringle’s book, The Creativity Choice: The Science of Making Decisions to Turn Ideas Into Action, the author makes the case that creativity has been misunderstood as predominantly about problem solving when, in reality, the most crucial aspect may be the process of problem finding, which involves noticing what’s overlooked, asking novel questions, and reframing one’s approach, much like what we saw with the “slow elevator problem.” Other academic literature not only supports this but places it in the relevant context of artificial intelligence: “The creative process generally encompasses stages such as problem identification, idea generation, and idea implementation. Problem identification is an early stage of the creative process, when individuals gather information, set objectives, and define specific challenges that need creative solutions. In human-AI collaboration, problem identification is an essential skill that the human partner needs to initiate prompts for the AI to answer.”13 According to Dr. Pringle’s work, problem finding is an integral part of what we call “the human advantage,” because our ability to creatively find and frame problems is driven by our emotions and by our embodiment, meaning machines lack that capacity when quickly producing solutions. The elevator problem, for instance, started with an emotional experience, which in turn sparked human ingenuity through the act of empathic questioning and problem framing.

When thinking about the impact of agentic AI, problem finding is one of the necessary components for being able to not just do the work but to be able to define the work, especially when operating collaboratively in human-AI hybrid teams. Problem finding also relates to the shift from answer production to questioning because, if AI produces answers on demand, the human who notices the right question – who sees what’s missing, what’s assumed, what’s unexplored – becomes the real creative force.

What this means is that the future of human value is not simply better problem solving, but better problem finding and better problem defining. Creativity begins in noticing, naming, reframing, and constructing the problem itself. It’s not enough to ask a question; the skill is also in how you frame it, what constraints you set, and what you choose to explore versus close down.

Risks and Opportunities in the Age of Agentic AI:

When examining current K12 practices, most of our assessment systems still reward problem-solving speed and answer correctness, which leaves little room for creative output and expression. But what would it look like to reward problem-finding and question quality instead? It’s an important provocation when we consider the following:

If school is still over-optimized for answer production, procedural completion, and compliance with predefined tasks, are we inadvertently preparing students for the zone of work AI is already rapidly absorbing? The question is not whether foundational knowledge still matters. Of course it does. The question is whether school is cultivating enough framing, judgment, inquiry design, and process transparency on top of those foundations.

There are risks though when exposing learners to advanced AI tools. The “productive friction” of noticing, exploring, reframing, and sitting with uncertainty is exactly what builds durable creative capacity, but agentic AI can remove that friction entirely if we let it.

So how do we measure what matters in ways that sustain the productive friction:

  • What if we assessed students on the quality of their questions, not just their answers?
  • What if “defining the work” were an explicit, taught, practiced skill — not just the teacher’s job?
  • What if we treated problem-finding as a discipline, not a warm-up activity?
  • What if students were not only asked to complete a task but to design the task themselves?

Students who do well in this new AI-mediated environment are not the ones who produce the most; they’ll be the ones who know what’s worth producing.

Agentic AI is pushing human value into new areas and new skill domains, whether that be upstream or downstream. It’s a shift from task execution to task stewardship, from answer production to question architecture, from doing the work to defining the work, and it’s not about more output; it’s about better framing. Not toward faster answers, but toward wiser questions. Not toward solving predefined problems, but toward noticing and creatively constructing the right problems in the first place.

Resources

  1. Chevallier, Arnaud, Frédéric Dalsace, and Jean-Louis Barsoux (2024). “The Art of Asking Smarter Questions.” Harvard Business Review. https://hbr.org/2024/05/the-art-of-asking-smarter-questions
  2. Khan, Salman, Douglas Fisher, Nancy Frey et al. (2026). Teaching Students to Use AI Ethically and Responsibly: Exploring AI with Intentionality, Curiosity, and Care, Corwin (Fisher & Frey), p. 56.
  3. Choudary, Sangeet Paul (2025). Reshuffle: Who Wins When AI Restacks the Knowledge Economy. Sangeet Paul Choudary, p. 197.
  4. Wedell-Wedellsborg, Thomas (2017). “Are You Solving the Right Problems?” Harvard Business Review. https://hbr.org/2017/01/are-you-solving-the-right-problems
  5. Boussioux, Leonard, et al. (2024). “The Crowdless Future? Generative AI and Creative Problem-Solving.” Organization Science 35(5): 1589–1607. https://pubsonline.informs.org/doi/epdf/10.1287/orsc.2023.18430
  6. Hirst, Peter (2026). “Why Asking Better Questions May Be the Most Important Leadership Skill in the AI Era.” MIT Sloan School of Management. https://executive.mit.edu/blog/why-asking-better-questions-may-be-the-most-important-leadership-skill-in-the-ai-era.html
  7. Boussioux, Leonard, et al. (2024). “The Crowdless Future? Generative AI and Creative Problem-Solving.” Organization Science 35(5): 1589–1607. https://pubsonline.informs.org/doi/epdf/10.1287/orsc.2023.18430
  8. Ibid.
  9. Choudary, Sangeet Paul (2025). Reshuffle: Who Wins When AI Restacks the Knowledge Economy. Sangeet Paul Choudary, p. 197.
  10. Khan, Salman, Douglas Fisher, Nancy Frey et al. (2026). Teaching Students to Use AI Ethically and Responsibly: Exploring AI with Intentionality, Curiosity, and Care, Corwin (Fisher & Frey), p. 80.
  11. Daugherty, Paul R. and H. James Wilson (2024). Human + Machine: Reimagining Work in the Age of AI. Harvard Business Review Press, p. 249.
  12. Lauritzen, Pia (2025). “How To Ask Better Questions in a World Drowning in AI Answers.” Forbes. https://www.forbes.com/sites/pialauritzen/2025/08/21/how-to-ask-better-questions-in-a-world-drowning-in-ai-answers/
  13. Habib, Sabrina, Thomas Vogel, Xiao Anli, and Evelyn Thorne (2023). “How Does Generative Artificial Intelligence Impact Student Creativity?” Journal of Creativity 34 100072. https://doi.org/10.1016/j.yjoc.2023.100072, p. 6.
LinkedIn
Mount Vernon Ventures © 2026