【AI x learning】 From Passive Use to Thinking Loops: Why You Must Learn to Ask Before Using AI

“It’s not that you don’t know how to use AI—you just don’t yet understand what kind of questioner you are.”

Published: 2025/12/18

As the first article in this series, it’s not about teaching you how to write prompts. Instead, it accompanies you in unpacking the journey of “Why am I still unsatisfied with AI’s answers?” Drawing from real user experiences, it explores the psychology and tone strategies behind questioning, helping you move from “being led by AI” to “leading AI with you.” This first article includes:

  • Questioning is not a technical issue but a reflection of mental state:
    • Some say, “I don’t know how to start asking.”
    • Some say, “I asked, but AI’s reply felt odd.”
    • Others treat AI as their best collaborator.
  • Different mindsets of questioners:
    • From “tone sprout type” to “tone design type,” analyzed step by step
    • Each type comes with everyday metaphors—you’ll likely see yourself in them
  • MIT research and psychology theories (SDT, EVT):
    • Not forced theory, but explanations of “why I ask this way”
  • Series preview:
    • Step by step, you too can begin collaborating with AI

Contents

Introduction
What Kind of Questioner Are You?
Types of Questioners
From Using AI to Training Your Thinking
Summary
References
FAQ

Introduction

In an age flooded with information, a well‑aimed question is proof that you haven’t given up thinking.
Is this you?

Hong‑An opened ChatGPT and typed a carefully considered opening line: “This is my original ad. Please transform it into a gentle yet firm brand narrative.”

He expected a rhythm and content he had never imagined, but ChatGPT’s reply felt like a rearranged version of his original ad copy.

“Was my tone too vague? Or did I ask the wrong question?” he wondered.

Next, Xiao‑Yin tried Copilot for her final book report. She asked: “Please write a 500‑word reflection on the book *Your Kindness Must Have a Sharp Edge*.”

Copilot produced a full essay. She felt efficient but sensed little personal involvement. She asked: “Can you make it sound more like my tone?”

Copilot generated another long essay. She noticed some adjustments but still felt uneasy: “Was my question not good enough? Did I really think it through?” In the end, she submitted it anyway.

Then Pei‑Cheng, a newcomer at work, was tasked with handling entertainment tax resubmission. Her boss said, “If you don’t know, ask AI.”

She stared at the input box and typed: “How do I file entertainment tax?”

Gemini listed documents and channels for filing, but it didn’t solve her problem. She needed “resubmission,” not initial filing. She remained confused.

Does this sound like your experience with AI?

The Consequences of Misusing AI

MIT Media Lab recently studied large language models (LLMs)[1]. Participants were divided into three groups: LLM group (using ChatGPT), search engine group (using Google), and brain‑only group (no tools). They wrote essays four times, with cross‑testing, brain activity analysis, NLP analysis, and interviews.

The study found that 83% of LLM users couldn’t recall what they had just written. Their brain connectivity was lowest, and they felt little ownership of the language. Vocabulary and themes were highly homogenized.

Results showed that overuse of AI reduces cognitive costs—effort in thinking, decision‑making, and memory. While it seems efficient, it harms long‑term thinking.

Researchers further suggested that prolonged low cognitive cost leads to “cognitive offloading,” eventually becoming “cognitive debt.” The effort you save now must be repaid later.

Nicola Jones in *Nature*[2] noted that even if people stop using AI, prior dependence may irreversibly lower brain activity.

Danish psychiatrist Søren Dinesen Østergaard[3] warned that generative AI’s fluent mimicry may overly accommodate users. Long‑term chatting could deepen delusions in sensitive groups and harm vulnerable populations.

Signs suggest that if we only wait for AI’s answers without thinking, one day we may truly be replaced. We might even worry about saying “thank you” to AI to survive.

“AI is scary. Should we stop using it?” you may ask.

From Questioning to Thinking

Not all users are like this.

Some open AI tools with confidence, clearly stating needs: “I’m organizing a topic. Here’s my draft. Please help me...” They quickly get text that matches their style and goals.

They refine by asking follow‑up questions, clarifying doubts, and pushing until tasks are complete. They then turn knowledge into teaching materials, reflections, or lessons.

They don’t just seek answers. They collaborate with AI. For them, AI is a co‑creator, but they remain in control.

Through AI, they solve problems once impossible and grow significantly in the process.

This system is the mental toolkit of “thinking‑type questioners.”

What Kind of Questioner Are You?

Before continuing with the rest of the article, this section presents a series of questions to help you examine your usual habits and states when asking questions or using AI.

This questionnaire is not meant to test whether you can ask questions. Instead, it helps you identify your tone participation state. Each option represents a rhythm of language and a psychological motive. Wherever you are now, the following articles will accompany you in finding the questioning style and rhythm that suit you.

Type: Not yet tested

1. When you face a problem you cannot solve alone and need to ask others or AI, what is your first reaction?

2. When interacting with AI or others, how do you usually ask questions?

3. In learning or work, what are your questioning habits?

4. When you open an AI tool, what is your most common first reaction?

5. After receiving AI’s reply, how do you usually feel?

6. When the answer is not as expected, what do you do?

7. When you need to ask in public, how do you feel?

8. Have you ever modified your questioning style to make AI’s reply closer to your tone?

9. After using AI to complete content, can you accurately recall its key points?

10. In team discussions, how often do you ask questions?

11. When using AI tools, do you feel you are learning?

12. Have you ever felt that AI’s replies were too accommodating, making you uneasy?

13. Have you ever felt your poor questioning led to inaccurate AI replies?

14. When asking strangers for help, how do you react?

15. When AI’s reply confuses you, what do you do?

The earlier set of questions wasn’t testing how often or how professionally you use AI. Instead, they were more like soul‑searching prompts, helping you see whether in dialogue you act as a bystander, participant, or leader. Your tone shapes the space of conversation—whether you’re being pushed along, exploring direction, or designing the entire process.

Note that the same person may shift roles depending on context. In social settings versus work, the outcomes can be very different.

Types of Questioners

Theoretical Foundations

In designing this framework, we drew on two psychological theories: Self‑Determination Theory (SDT)[4] and Expectancy‑Value Theory (EVT)[5].

These theories, from the perspectives of psychological needs and task motivation, help us understand AI users’ depth of language participation and control in dialogue.

By combining them, we can more precisely outline the mental profiles of different tone participants and design modules to support questioning and learning with AI.

Self‑Determination Theory (SDT)

SDT explains human motivation and personality development, distinguishing intrinsic from extrinsic motivation.

Intrinsic motivation means acting for enjoyment, challenge, or satisfaction. Extrinsic motivation means acting for rewards, recognition, or responsibility.

SDT emphasizes three basic psychological needs: Autonomy (feeling self‑directed), Competence (feeling effective), and Relatedness (feeling connected). When these are met, even external motives can transform into high‑quality self‑determined behavior.

Ryan & Deci further mapped motivation along a spectrum from “non‑self‑determined” to “highly self‑determined”:

  • Amotivation: no drive, no intention to act.
  • External Regulation: acting for rewards, punishment, or demands.
  • Introjected Regulation: acting to avoid guilt or maintain self‑image.
  • Identified Regulation: recognizing value and willing to act.
  • Integrated Regulation: aligning behavior with self‑values.
  • Intrinsic Motivation: acting for enjoyment, challenge, or satisfaction.

    • Thus, SDT doesn’t dismiss extrinsic motivation but stresses its internalization and quality. Motivation quality matters more than intensity.

    Expectancy‑Value Theory (EVT)

    EVT, proposed by John W. Atkinson in the 1950s–60s, explains motivation as a mix of Expectancy (“I can do it”) and Value (“It’s worth doing”).

    Eccles and colleagues later refined Value into:

  • Attainment Value: importance to identity.
  • Intrinsic Value: enjoyment and interest.
  • Utility Value: usefulness for future goals.
  • Cost: effort, anxiety, opportunity cost.

    • Expectancy also includes Efficacy Expectancy (ability to complete) and Outcome Expectancy (desired results). If either expectancy or value is zero, motivation approaches zero.

    EVT reminds us: motivation is the intersection of “I can do it” and “It’s worth doing.”

    Tone Sprout Users: Language Participation Not Yet Activated

    You may have said things like—

    “I was told to use AI, but it’s too complex. I don’t even know how to start.”

    “I don’t get the tone differences. As long as it produces results, that’s fine.”

    “I copied someone’s prompt, but the result felt off...”

    These aren’t mistakes. They show early awareness of your tone in AI dialogue. You’re beginning the journey of gaining control.

    Early users often feel passive, as if AI is uncontrollable. Sometimes they worry AI might judge them.

    Psychology calls this “unactivated motivation,” often driven by external pressure like rewards or punishment.

    It doesn’t mean lack of ability. It means autonomy, competence, and relatedness aren’t yet supported. Once they are, motivation can activate and external motives can transform into self‑determined behavior.

    EVT also reminds us: motivation arises from the overlap of “I can do it” and “It matters.”

    If you’re a tone sprout user, this series will guide you through clarifying purpose, starting phrases, and rhythm, helping you feel “I can do this” and “This helps me,” moving from bystander to participant.

    Tone Explorer Users: Intent Emerges, Rhythm Still Practiced

    You may have thought—

    “AI can solve this, but I’m not sure how to ask.”

    “AI answered, but something feels missing. I don’t know how to adjust.”

    “How can I make AI’s reply sound more like me?”

    These thoughts show you’ve taken the first brave step toward leading dialogue.

    Such users recognize tone participation’s importance and have intent, but still feel AI’s replies are hard to steer—like mismatched conversations.

    This often reflects EVT’s “high value × low expectancy”: knowing it’s worth doing but doubting ability. They may fear wasting time without results.

    In SDT, they have some autonomy and relatedness, motivating them to start leading dialogue rhythm.

    If you’re a tone explorer, this series will teach rhythm building, quality checks, and follow‑up skills, helping you move from “I want to ask” to “I know how to ask.”

    Tone Designer Users: Control Established, Strategy Building

    Experienced AI users may ask—

    “I know the problem I want to solve, but which question should I ask first?”

    “Is this enough? Should I keep asking? How do I extend my questions?”

    “Can AI really solve this? How do I confirm?”

    At this stage, AI feels like a daily partner, helping with tasks and workflow optimization.

    These users have control and are designing tone strategies—adjusting tone, reframing questions, challenging replies. They show integrated regulation (SDT) and high expectancy × high value (EVT).

    If you’re a tone designer, this series offers methods to design rhythm, check quality, and refine through follow‑ups, making AI a stronger partner.

    Tone Creator Users: Style and Rhythm Stabilized, Entering Creation

    “AI is my co‑creator!”

    “AI helps me grow and create beyond my limits.”

    For tone creators, AI is no longer just a tool but a collaborator across tasks.

    They’ve reached language creation, with stable control and style. Their motivation is intrinsic (SDT), and expectancy and value are fully integrated (EVT). Language becomes a medium of creation and growth.

    This series will share my own contexts, questioning methods, and projects solved with AI—projects still ongoing.

    Again, the goal is not ranking users but helping each find their rhythm, moving from bystander to designer.

    Whether you’re just starting, stuck, or aiming higher, tone support awaits. AI can shift from rigid tool to friendly collaborator.

    From Using AI to Training Your Thinking

    “Knowledge should not be poured in, but built through questioning, exploring, and constructing.” This is the core of Inquiry‑Based Learning[6][7].

    AI is a 24/7 partner. We can align inquiry steps—Questioning, Investigating, Explaining, Communicating, Reflecting—with AI dialogue and tone control to learn more effectively.

    Instead of just seeking answers, it becomes a process of knowledge construction:

  • Questioning: start by clarifying your thoughts with a first question.
  • Investigating & Explaining: check if your understanding matches AI’s, and whether its sources are reliable.
  • Communicating: confirm, extend, and refine through back‑and‑forth until goals are met.
  • Reflecting: analyze blind spots, overlooked areas, and better questioning methods.

    • Combining AI with inquiry‑based learning gave birth to this series. As the opening of “AI x learning,” we’ll explore themes such as:

  • How to organize your thoughts before questioning
  • How to help AI understand who you are, your tone, and your goals
  • Finding the rhythm and questioning style that suits you
  • Follow‑up questioning to evolve AI’s replies
  • How to design closure and timing for action
  • Caring for the psychological states that surface while questioning
  • Translating this module into non‑AI learning tools
  • Practice: Hard Drive Rescue Project

    • This series is not just about workflows or prompts (though we’ll share many). It’s about rebuilding the habit of inquiry‑based thinking and shaping your personal AI questioning method.

    Within this system, you’ll learn templates and modules such as:

  • Motivation Recognition Module: helps you know why you’re asking
  • Goal Anchor Generator: designs anchors tailored to your motives
  • Starter Phrase Generator: produces detailed opening prompts from your questionnaire
  • Dialogue Anchoring Rhythm: mechanisms to keep AI from running wild
  • Quality Check Process: ensures AI isn’t hallucinating
  • Semantic Follow‑up Flow: questioning techniques to request clarifications
  • Closure Checklist: sensing when to stop asking and start doing
  • AI Learning Flow: using AI for leaps in growth

    • One day, you’ll open AI without hesitation, no longer worrying about choosing Copilot, ChatGPT, or Gemini.

    Your first words will carry motivation. You’ll know what to explore, what to co‑create, and how to express your tone and adjust collaboration.

    Replies will no longer be passive answers. They’ll feel like “I participated” and “This is my way of thinking.”

    You’ll have rhythm, follow‑ups, closure, and emotional care in language. No more frustration when AI misses your intended answer.

    Dialogue will become a daily training ground for your linguistic sense.

    This is what you’ll look like after learning the module: someone with language sovereignty, designing thought through tone, and using AI to learn and grow. This series will help you reach that goal.

    Summary

    You’ll stop just typing inputs. Each question will engage your thinking. AI’s best reply is less important than your ability to ask.

    How do you want to start today’s dialogue? Try a short line—just to join the thinking.

    A good question doesn’t prove you’re smart. It proves you’re still thinking. But before you ask—do you really know why you’re asking?

    Next chapter, we’ll uncover the motives and purposes behind questioning.

    References

    [1] Zhang, Y., Wang, Y., & Picard, R. W. (2024). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task. arXiv. https://arxiv.org/abs/2506.08872

    [2] Jones, N. (2025, June 25). Does using ChatGPT change your brain activity? Study sparks debate. Nature. https://doi.org/10.1038/d41586-025-02005-y

    [3] Østergaard, S. D. (2025). Generative artificial intelligence chatbots and delusions: From guesswork to emerging cases. Acta Psychiatrica Scandinavica. Advance online publication. https://doi.org/10.1111/acps.70022

    [4] Ryan, R. M., & Deci, E. L. (2000). Self‑determination theory and the facilitation of intrinsic motivation, social development, and well‑being. American Psychologist, 55(1), 68–78. https://doi.org/10.1037/0003-066X.55.1.68

    [5] Wigfield, A., Tonks, S., & Klauda, S. L. (2009). Expectancy‑value theory. In K. R. Wentzel & A. Wigfield (Eds.), Handbook of motivation at school (pp. 55–75). Routledge. https://doi.org/10.4324/9780203879498

    [6] Spronken‑Smith, R. (2005). Experiencing the process of knowledge creation: The nature and use of inquiry‑based learning in higher education. University of Otago. Retrieved from Ako Aotearoa summary report

    [7] Pedaste, M., et al. (2015). Phases of inquiry‑based learning: Definitions and the inquiry cycle. Educational Research Review, 14, 47–61. https://doi.org/10.1016/j.edurev.2015.02.003

    FAQ

    It’s not that AI is wrong or you’re incapable. You just haven’t found the tone and structure that fit you.

    Yes, and it matters a lot. Tone affects how AI interprets you, your depth of participation, and the depth of its answers.

    Because you’re not just using a tool—you’re interacting with a language model. Your mindset directly shapes results. You must know why and how you ask to get the answers you want.

    This article is the series introduction. It includes modules the author personally uses and continues to refine. The first article clarifies user questioning states.

    Absolutely. This is written for those who’ve felt stuck but don’t want to give up. Tone is natural, metaphors are everyday, and jargon is avoided.

    Thank you for reading my article! Your support and encouragement fuel my creativity. If this piece inspired or helped you, please consider supporting me through the link above so I can continue sharing valuable content. Any amount is deeply appreciated. Thank you for your support and companionship—I look forward to sharing more meaningful and practical stories and experiences :)

    You may also be interested in these articles

    如何成為有效率的學習者？
    致，總是覺得時間不夠用的你-淺談時間管理
    【AI x learning】 提問不只是輸入，而是內在方向的外顯：找到屬於你的起跑線
    🌐 CN
    Copyright© Wakame, 2022