AI Prompting for Learning

Dimension: Pursuit · Type: Stage

Three principles for getting useful learning support from any AI assistant: be specific about your level, ask for structure, and push it deeper than the default answer.

Introduced by Tom Frohner (LinkedIn) at the AI for Your Career, Practical Tools and Prompts session of the UN Inter-Agency Career Week 2026, on 6 May 2026. Tom demonstrated the principles using LinkedIn Learning’s AI Coach, but emphasised that the same prompting approach works on any general-purpose AI tool.

The framework

The shift is from “tell me about X” to “teach me X at my level, structured, with the depth that beginners typically miss”. The principles work for one-off chats and for the instructions you write inside a custom AI agent.

When to use it

• Whenever you ask an AI assistant to help you learn or understand something.
• When you are stuck on a topic and a course feels too slow.
• When preparing for a discussion in an unfamiliar domain (a new sector, a new policy area, a new technical concept).

What you need

A specific learning goal: a topic, a skill, a concept you want to understand. Honest awareness of your current proficiency level. Access to any general-purpose AI assistant (ChatGPT, Claude, Copilot, Gemini), or LinkedIn Learning’s AI Coach if you have a Career Hub licence.

The three principles

1. Be specific about your level. Generic prompts produce generic answers. The AI does not know whether you are starting from zero or extending an existing knowledge base.

Weak prompt: “How can I get better at marketing?”

Strong prompt: “Explain digital marketing fundamentals to someone with no prior experience, using real-world examples.”

The level cue tells the AI what to assume, what to define, and what to skip. State it explicitly, even when it feels obvious.

2. Ask for structure. Default AI output is a list of tips or a long paragraph of overview. Neither is the most useful structure for learning.

Ask for:

• A learning roadmap, beginner-to-advanced.
• Books, free courses, and concepts named explicitly.
• Concepts explained “like a mentor would, walking me through it from first principles”.
• A timeline or sequence, with what to do in week one, week two, week three.

The structured output is also easier to act on. A roadmap with named resources beats a list of motivational tips every time.

3. Push it deeper. The default AI output covers what most people would cover. The compounding learning is in what most people miss.

Ask:

• “What would an expert know that a beginner wouldn’t?”
• “What are the common misconceptions in this topic?”
• “Where do most learners stall, and how do they get unstuck?”
• “If I had to teach this to a colleague in two weeks, what is the one thing I must understand by then?”

These questions force the AI to surface the non-obvious, which is where the gain lives.

Steps

1. Name your learning goal in one sentence. Specific, not “I want to learn AI”. Try “I want to understand prompt engineering well enough to build a career-development agent for myself within four weeks.”
2. State your current level. “I have never built an AI agent before but I have used ChatGPT for routine tasks for a year.” This anchors the AI’s response.
3. Ask for the roadmap with structure. “Give me a four-week roadmap with weekly milestones, named resources, and one concrete practice task per week.”
4. Push for depth. Once you have the roadmap, follow up: “What would an expert in prompt engineering say is the most overlooked aspect of this roadmap?”
5. Pair with practice. AI output is Education in the 3 E’s of Development. Without an Experience commitment (“where will I apply this next week?”) the learning will not move.

Worked example

Tom typed into LinkedIn Learning’s AI Coach: “Can you explain digital marketing fundamentals to someone with no prior experience?”

The AI Coach returned a short topic introduction, a build-your-marketing-technology roadmap, and three named beginner-level course proposals from the LinkedIn Learning library. The roadmap was structured as a sequence; the course proposals were tied to the level he specified.

Then he pushed deeper: “What would an expert know that a beginner wouldn’t?” The next response surfaced concepts the introductory roadmap had not named: marketing-attribution modelling, the difference between leading and lagging indicators in funnel design, why most digital-marketing teams misuse cohort analysis.

The same approach works on any general-purpose AI assistant. The point is in the prompting structure, not the platform.

Pitfalls

• Asking the broad question and accepting the broad answer. “Tell me about M&E” produces a generic overview. “Explain results-based M&E to someone with three years of programme management experience but no formal M&E training, with a focus on what most programme officers misunderstand about indicator design” produces something usable.
• Skipping the depth-push step. Most useful learning is one prompt past the default. The third principle is the one that produces real depth.
• Treating AI output as authoritative without verification. AI can be confident and wrong. For high-stakes learning, cross-check against a primary source: a published course, a textbook, a colleague who actually does the work.
• Using AI for learning without committing to practice. AI is the most efficient Education-dimension tool ever built; it is still only Education. Without Experience and Exposure, the learning does not move (see 3 E’s of Development).
• Confusing fluency with understanding. AI explanations make you feel like you understand. The test is whether you can apply the concept on a real task; if not, the comprehension was performative.

When not to use it

When the topic involves contested or fast-moving policy details, where the AI’s training data may be outdated or simplified. For sensitive technical or legal content (compliance, ethics frameworks, sanctions regimes), use AI as a starting point and verify against authoritative sources.

When you are practising a skill rather than learning a topic. AI prompting is for understanding; AI Roleplay for Skill Practice is for performing.

How I use it

Personal note pending. Davide to fill.

Related frameworks

• 3 E’s of Development, the broader frame in which AI prompting for learning sits as the Education dimension.
• AI Roleplay for Skill Practice, the companion approach for practising rather than learning.
• Four Prompting Principles, the broader prompting framework that applies to any AI use, not just learning.
• Career Gap to Sprint Workflow, the more specific career-development application of structured prompting.

---

Notes compiled by Davide Piga. Last updated 2026-05-09.