AI tools based on large language models are starting to reimagine how conversations unfold in organizations. Instead of just responding to questions, many systems now ask their own questions to clarify intent, explore context, or refine a task.

For example:

• WaLLM, a WhatsApp-based chatbot, doesn’t just answer queries. It offers follow-up questions, lists of trending and recent queries, and even a “Top Question of the Day.”
• OpenAI’s DeepResearch and tools like Manus are specifically designed to ask users questions when requests are ambiguous, rather than guessing and risking errors.

For managers, this means AI is moving from being a passive tool to an active conversational partner that can shape how issues are framed and which options are explored. This can be helpful for surfacing context and reducing misunderstandings, but it also means leaders need to be more intentional about how they let AI influence discussions and decisions.

Research comparing more than 1,600 executives with 13 leading language models shows that AI and human leaders tend to ask very different mixes of questions.

Key findings:

• AI systems often overemphasize interpretive analysis (e.g., “What does this mean?” “How do these pieces fit together?”).
• They tend to underweight productive and subjective questions (e.g., “What should we do next?” “How will stakeholders feel about this?”).

Because each type of question surfaces different information, this imbalance can:

• Create blind spots by overlooking execution details or stakeholder concerns.
• Skew discussions toward analysis at the expense of action and judgment.
• Subtly steer outcomes by framing what seems important and what gets ignored.

On top of that, the study found that models vary widely from one another. The questions asked by one system can be quite different from those of another, so tool choice directly affects how conversations are shaped.

The risk isn’t that AI asks questions; it’s that, if left unchecked, its questioning style can quietly redirect priorities and decisions away from what leaders actually care about.

Leaders can get value from inquiry-driven AI while managing risk by treating AI as a thinking partner, not a decision-maker. A few practical guidelines:

• Be deliberate about when and how you use AI questions. Use AI to broaden thinking or clarify ambiguity, but don’t let it define the full agenda of a discussion.
• Compare outputs across systems. Since the study found large differences across 13 leading models, it can help to run the same prompt through more than one tool and see what each one overlooks.
• Actively probe for missing perspectives. Pay special attention to areas AI tends to underweight, such as execution details, stakeholder reactions, and subjective judgments.
• Retain final judgment. Treat AI’s questions as inputs to guide your thinking, not as substitutes for leadership or accountability.

In practice, this means using AI to expand the conversation—for example, by surfacing additional angles or clarifying assumptions—while leaders remain responsible for prioritizing issues, balancing trade-offs, and making the final call.