Practicing Questioning with AI
Questioning is a motor skill as much as a cognitive one: you only improve by practicing. Yet in real life, chances to practice are rare and risky (you don't get to "redo" a botched performance review). Conversational AI changes that: it offers an infinitely patient training partner, available on demand, able to play any interlocutor. This is Shift Kognition's signature angle — pairing psychology with AI to turn knowledge into skill.
Three complementary uses
flowchart TD
A["Practice with AI"] --> B["Simulate<br/>(role-play)"]
A --> C["Evaluate<br/>(feedback on my questions)"]
A --> D["Generate<br/>(prepare my questions)"]
1. Simulate a conversation (role-play)
Ask the AI to play a specific interlocutor and to not make it easy for you. You practice questioning and paraphrasing under realistic conditions.
Simulation prompt: "You play the role of Karim, a demotivated team member who feels his ideas are never heard. You're defensive and you answer briefly at first. I'm his manager and I want to understand what's going on in a one-on-one. Stay in character, don't give me the solution, and only open up if my questions are open and my paraphrasing is accurate. Start with: 'You wanted to see me?'"
2. Get feedback on your questions
After (or during) the simulation, ask for an analysis of your technique.
Feedback prompt: "Analyze the previous exchange solely in terms of MY questioning technique. For each of my turns, indicate: was it an open, closed, leading or fake question? Did I paraphrase before responding? Give me 3 questions I could have asked instead, and a score out of 10."
3. Prepare your questions in advance
Before an important conversation, have a bank of questions generated for you to sort.
Preparation prompt: "I'm preparing a discovery call with a prospect in the logistics sector. Suggest 10 open questions following the SPIN logic (situation, problem, implication, payoff). Avoid any leading or closed questions. Order them from broadest to most specific."
Using AI well as a coach
| Best practice | Why |
|---|---|
| Give the AI a precise role (age, mood, stakes) | The more defined the character, the more realistic the practice |
| Ask the AI to stay in character | Prevents it from "breaking" the role to help you too soon |
| Have it evaluate one dimension at a time | Targeted feedback (e.g. only your questions) is more actionable |
| Replay the same scene several times | Spaced repetition anchors the skill |
Guardrail: AI simulates, it does not feel. It's ideal for building the mechanics (question form, the reflex to paraphrase), but real empathy is tested with real humans. Use AI like a flight simulator: essential for training, never the final terrain.
Practical exercise
Run the "Karim" simulation above. Hold 10 turns without ever giving advice — only questions and paraphrases. Then run the feedback prompt and note the two most useful corrections for your next real conversation.
Summary
Conversational AI is an ideal training partner for a skill that demands practice. Three uses: simulate an interlocutor through role-play, evaluate the form of your questions and paraphrases, and generate a bank of questions before an important exchange. The keys: give a precise role, ask the AI to stay in character, target the feedback, and replay. Essential guardrail: AI builds the mechanics, but real empathy plays out with real humans.