Digital Economy Dispatch #222 -- Why AI is about Better Questions not Better Answers

In the current wave of AI, much attention has been focused on the generative capabilities of AI systems—their ability to create text, images, code, and other content that appears remarkably human-like. However, this fascination with AI's creative outputs may be obscuring a more fundamental transformation: the evolution in how we formulate and pose questions to these systems. The true revolution in AI might not lie in its ability to generate answers, but in how it compels us to become more sophisticated in asking better questions.

The Importance of Quality Input

Consider the current enthusiasm surrounding generative AI tools. Organizations are rapidly deploying these technologies, excited by their potential to generate content, analyze data, and automate decisions. However, as with any powerful tool, the quality of output is inherently dependent on the quality of input. The sophistication of our questions determines the value of the answers we receive.

This becomes particularly evident in enterprise settings, where the complexity of business challenges demands more than just computational power. My experiences from the past (pre-AI) with developing inventory management system illustrates this point. Despite the sophisticated algorithms and extensive data analysis capabilities, in practice we found that they failed to deliver optimal results because the initial questions posed to the system did not adequately account for local variations, cultural nuances, and infrastructure limitations. Our focus was too often misplaced on tweeking the algorithms to deliver better answers rather than ensuring we guided users toward asking better questions.

Such concerns are amplified with the deployment of AI-based systems. Here, the importance of better questioning extends beyond algorithmic issues to include 2 further key areas: Contextualizing data and building better prompts.

In the first case, in an era of "data-driven everything", organizations must resist the temptation to view data in isolation. Raw data, no matter how abundant, provides limited value without the context that comes from asking the right questions. This context-aware questioning becomes particularly crucial when dealing with AI systems trained on internet-scale datasets, which may harbour inherent biases and inconsistencies.

In the second, it is particularly evident with generative AI applications, where the quality of prompts—essentially, the questions we pose to these systems—directly determines the value of the generated content. Organizations rushing to adopt large language models often discover that simply having access to sophisticated AI isn't enough; the art of crafting effective prompts is crucial. Poor questions invariably lead to poor outputs, regardless of the underlying model's capabilities.

So, for example, when users ask vague questions like "Write content about our product" instead of specific, contextual queries like "Write a technical comparison of our product's energy efficiency features against industry standards, focusing on data from the last two quarters," the resulting content often lacks focus, accuracy, and practical utility.

Moreover, the challenge extends beyond just adding details. Effective prompting requires understanding the context, limitations, and capabilities of the AI system. Organizations frequently discover that their initial excitement about generative AI's potential is tempered by the realization that they need to invest significant effort in learning how to ask questions that elicit useful responses. This includes understanding how to provide relevant context, specify desired formats and tone, and include necessary constraints—all elements of what now is called "prompt engineering".

AI as a Question Refiner

This focus on “better questions” can lead to an even more interesting conclusion. Perhaps even more significantly, generative AI's most valuable contribution may lie not in its ability to generate answers, but in its capacity to help users refine and clarify their questions.

This is particularly evident in professional services contexts, where clients often struggle to articulate their exact needs. In legal services, for example, a client typically approaches legal professionals with broad, imprecise questions containing a mix of issues cloaked in details of specific personal circumstances. The real value lies not in generating an immediate response, but in using AI to help explore and clarify the underlying context to extract the key elements of the situation: What specific actions led to this consideration? What documentation exists? What outcomes are actually desired?

In these scenarios, AI systems can serve as sophisticated question-refinement engines, helping to disambiguate complex queries through iterative dialogue. Rather than attempting to generate definitive answers, the AI can help surface important considerations, identify missing context, and guide users toward more precise and meaningful questions. This approach transforms the traditional client-professional interaction from a simple question-and-answer exchange into a more nuanced exploration of the actual problem space.

This reframing of AI's role—from answer generator to question refiner—has profound implications for many domains such as professional services. Law firms, management consultancies, and other advisory services are beginning to recognize that AI's greatest value may lie in its ability to help clients better understand and articulate their own needs before human experts become involved. This not only leads to more efficient use of professional expertise but also often results in better outcomes as the true nature of the problem is more clearly understood from the outset. Expensive human-based insight can then be targeted where is adds the most value.

New Competencies for Leaders

For leaders and decision-makers, this shift in perspective from focusing on delivering answers to asking better questions demands a new set of competencies. The ability to formulate incisive questions becomes as important as the technical skills needed to implement AI solutions. This includes:

1. Developing a deeper understanding of the context in which questions are being asked, including cultural, ethical, and operational considerations.

2. Recognizing the limitations and potential biases in both the questions being asked and the data being used to generate answers.

3. Creating frameworks for evaluating the quality and relevance of questions before committing resources to finding answers.

4. Fostering an organizational culture that values thoughtful questioning as much as efficient answer generation.

With this in mind, the path forward requires a fundamental reconceptualization of how we approach AI implementation. Rather than viewing AI primarily as an answer generator, we must see it as much as a tool for developing better questions. This might mean spending more time in the problem-definition phase, engaging diverse stakeholders to ensure questions account for various perspectives, and regularly revisiting and refining the questions we ask of our AI systems.

The Future of AI and Questioning

The future belongs to leaders who understand that the quality of their questions—not just the quantity of their data or the sophistication of their algorithms—will determine their success in leveraging AI effectively. While the generative capabilities of AI are indeed remarkable, they should not overshadow the more fundamental transformation in how we approach problem-solving. The real challenge—and opportunity—lies in asking better questions.