Proper Prompting for School Psychologists

Key Points (For Readers on the Go)

  • When people say “AI isn’t very good,” the problem is almost always how it’s being used.

  • The most important AI skill for psychologists is proper prompting.

  • Prompting is a back-and-forth supervisory process, not copying prompt libraries.

  • Large language models (LLMs) work best when treated like a capable intern.

  • You don’t need a paid course—high-quality free resources exist.

Why This Matters for Psychologists (and Other Providers)

I’m honestly surprised it’s taken me this long to write directly about prompting, given how often it comes up in training and how important it is.

In nearly every AI training I give, someone says:

“I tried AI, and honestly, it wasn’t very good.”

Almost without exception, the issue isn’t the tool—it’s the interaction.

People often give AI a vague instruction, expect a polished professional product in one shot, and are disappointed when the output misses the mark.

Large language models (LLMs) should be treated like interns: capable, fast, and helpful—but in need of clear guidance, feedback, and oversight.

Let’s Retire the Term “Prompt Engineering”

You’ll often hear this skill called prompt engineering. For educators, psychologists, and allied health professionals, that framing creates unnecessary barriers (and makes it sound harder than it is).

You do not need:

  • Technical expertise

  • Coding skills

  • A massive prompt library

You do need to communicate clearly, revise thoughtfully, and apply professional judgment.

That’s not engineering - That’s supervision.

Start With This Mental Model: AI as an Intern

If you handed a task to a graduate intern and said:

“Draft an email about an upcoming FBA.”

You would expect:

  • Clarifying questions

  • A (bad) rough first draft

  • Revisions

AI works the same way.

Good prompting is not about getting the perfect output on the first try. It’s about guiding the process.

Strategy 1: Provide Context (Role-Based Prompting)

Ineffective prompt

“Draft an email about an upcoming FBA.”

Improved prompt

“You are a seasoned school psychologist drafting a sensitive email to a 3rd-grade teacher and a guardian about a student who is struggling with emotional regulation. Draft a brief, jargon-free email outlining the purpose of the FBA, emphasizing a collaborative approach, and requesting three potential meeting times.”

This works because you clearly defined:

  • Persona (who the AI is)

  • Audience

  • Tone

  • Purpose

  • Constraints

Exactly how you would orient an intern (with the exception of persona) before asking them to communicate on your behalf.

Strategy 2: Be Specific

Vague prompts produce vague results.

Ineffective prompt

“Give me interventions for ADHD.”

Effective prompt (from the slides)

“Research three Tier 3 evidence-based interventions for a 9th-grade student with documented ADHD (Inattentive type), focusing specifically on reducing task initiation difficulties in their Biology class. Format as a bulleted list suitable for presentation during an IEP team meeting. Provide URLs for your sources.”

Specificity improves:

  • Relevance

  • Accuracy

  • Efficiency

Just like with an intern, unclear instructions lead to unusable work.

Strategy 3: Build Through Conversation (Don’t Start Over)

This is where many users get tripped up.

You do need to establish a persona first.
That initial framing still matters.

However, once the persona is set, you do not need to restate:

  • Persona

  • Audience

  • Tone

  • Constraints

Unless you are doing so for emphasis or correction.

Why This Works

LLMs retain conversational context. Once you’ve oriented the “intern,” you can move straight into revision and refinement.

Initial prompt (persona + task)

“You are a school psychologist. Generate five legally sound accommodations for a high school student with Dyslexia in written expression, aligned with IDEA guidelines. Format as a table and provide source URLs.”

Follow-up prompt (refinement only)

“Now rephrase accommodation #3 so it’s clearly understood by a parent.”

No restarting.
No re-explaining the job.

That’s how supervision works in real practice.

Strategy 4: Ask AI If It Has Questions

One of the most underused prompting techniques is inviting clarification before output.

Example

“I need help drafting measurable IEP goals for a student with executive functioning difficulties. Before you generate the goals, do you have any questions?”

The AI may ask about:

  • Grade level

  • Priority skills

  • Setting

  • Baseline data

That exchange saves time and improves accuracy—just like a competent intern checking assumptions.

Why Prompt Libraries Fall Short

Prompt libraries can be useful exposure tools, but they often:

  • Encourage copy-paste behavior

  • Create false confidence

  • Break when context changes

Professional practice is contextual. Prompting should be, too.

Learning how to guide, revise, and supervise matters far more than collecting templates.

Free, High-Quality Prompting Resources

You don’t need a paid course to learn this skill. If you want structured guidance, these are the resources I recommend most often:

I receive no remuneration from any of these sources.

Bonus: Tell AI to Cite Its Sources (Make It Bring Receipts)

One simple instruction dramatically improves AI usefulness and safety:

Tell it to cite.

When you ask an LLM to provide URLs or sources:

  • Claims are replaced with traceable information

This mirrors what you would expect from an intern:

“Don’t just tell me—show me where you got this.”

Example

Instead of:

“List evidence-based interventions for ADHD.”

Use:

“List three evidence-based Tier 3 interventions for ADHD relevant to a secondary school setting. Provide URLs to peer-reviewed or authoritative sources for each.”

Citations don’t make AI correct.
They make it auditable.

Final Takeaways

  • If AI seems ineffective, the issue is often prompting—not capability.

  • Prompting is a professional communication skill, not a technical one.

  • Establish a persona, then iterate—don’t restart.

  • Skip prompt libraries; learn the conversation.

  • Asking AI to cite sources improves transparency and oversight.

AI Use Disclosure
Portions of this post were drafted with the assistance of an AI writing tool and revised by the author for accuracy, clarity, and professional judgment.

Previous
Previous

AI and Skill Development in School Psychology: Are We Sacrificing Mastery for Speed?

Next
Next

Thanks to AI, You Should Be Generating (At Least) Two Psychological Reports Per Student/Patient