Gains:
- Being able to distinguish where in the PDR workflow (interview notes, psychoeducational material, career content, follow-up summary) artificial intelligence saves real time and where decisions such as diagnosis, guidance and psychological counseling are left to the expert, depending on the task risk level.
- Understanding that artificial intelligence can never make a decision in case of crisis and risk, that it is necessary to consult human experts and the necessary authorities, and that every output must be verified.
- Anonymization of student data within the scope of KVKK, professional confidentiality and PDR ethical principles and gaining the habit of choosing safe vehicles
A school counselor's day is filled with a load of writing and documentation that few people see: individual interview records, classroom guidance plans, psychoeducational brochures, parent information letters, scale reports, student tracking charts, group and commission minutes, career choice materials. Much of this work is repetitive and consumes hours at a desk; However, the real value of the profession lies in the face-to-face, trust-based relationship established with the student. Artificial intelligence (AI) — computer programs that can understand human language and produce text, outlines, plans, and summaries — is a powerful aid to speeding up this repetitive writing load. Throughout this module, we will learn to use AI not as a magic solution, but as a writing assistant that produces drafts, gives ideas, and edits.
But from the very beginning, let's put two sentences that stand above everything else in this profession: AI does not make decisions about the student; It does not diagnose, provide guidance, or replace a psychological counseling relationship. And even more critical: In situations of crisis and risk, AI never comes into play; There, human experts, family and competent authority are essential. These two principles are the security foundation on which the rest of the module is built.
This first unit lays three foundations: distinguishing where AI saves time in the PDR workflow and where it should stop; verifying each output; protecting student data.
Where does AI save time, where does the decision belong to the expert?
It is useful to think of PDR tasks in terms of a risk level. Risk level is how much harm it will cause to the student if a task is done incorrectly. In PDR, this risk is more severe than in most professions; because it is about a child's spiritual well-being, safety and future.
Low-stakes tasks that AI speeds up a lot: Outlining a classroom guidance activity, writing the first version of a test anxiety brochure, organizing scattered interview notes into an organized structure (after personal data is removed), drafting a parent newsletter, simplifying a career introduction text, generating ideas for a psychoeducational presentation. Here AI solves the “blank page” problem; You correct it, enrich it with your professional knowledge and give final approval.
High-risk tasks where the decision is left to the expert: Putting a diagnosis-like label on a student ("this child is depressed", "he has attention deficit"), clinically interpreting a scale result, directing a student to a profession, using definitive statements to a family, defining a behavior as "risk" and initiating action. These decisions affect the child's life; The responsibility and final say always belong to the expert. AI offers at most a point of view here, it cannot make decisions.
Forbidden area — where AI has never entered: Crisis and risk. Suicidal ideation, harming oneself or others, sexual/physical/emotional abuse, neglect, severe psychological symptoms. In these cases, we ask the AI "what should I do?" Asking is a waste of valuable minutes and trying to delegate responsibility to a program. The correct thing is to operate the emergency protocol, reach the authorized human expert and family, and notify the necessary authorities. We will cover this in detail in unit 9; But the reflex must be established from today.
Caution: "AI said" is not a justification. AI output is not a substitute for approval from a competent expert in any area that affects the student's mental well-being, safety, or future. An unverified AI output is like an unverified rumor.
Why is verification necessary? Hallucination and bias
AI produces text not by "knowing" but by "predicting the possible word". That's why sometimes it gives extremely fluent but actually incorrect information. This is called a hallucination: a scientific study that does not exist, an unspoken quote, a false statistic, a scale norm that does not exist, a fabricated guideline. For a PDR specialist, this is dangerous; Because incorrect information given in psychoeducation turns into a permanent misconception in the student, and wrong statistics given to a family turns into wrong anxiety.
The second problem is bias: AI can repeat social patterns in the data it is trained on; can produce biased output about gender, culture, socioeconomic status, or “achievement.” For example, it may skew a female student towards certain professions and a male student towards others; or it may frame a behavior as "naughtiness" in one student and a "serious problem" in another. Bias in PDR directly undermines the principle of justice and equality of opportunity.
Because of these two limits, we propose a simple validation discipline to apply to every output:
- Connect it to the source. Verify anything containing facts, statistics, quotes, rules, or diagnostic criteria from a reliable source (scientific study, government agency, professional organization document).
- Check again. Manually check a numerical or logical output (a table, a time plan).
- Professional filter. "Is this appropriate for this student, this age, this culture, this family? Is it likely to cause harm?" handle it with your own expertise.
- Take responsibility. The moment you approve it, that text is yours; Any mistakes are your responsibility.
Student data: privacy from the start
PDR records contain the most sensitive type of data. Student name, number, class are personal data; mental state, family situation, health information, a diagnosis, or a trauma are special personal data and require much higher protection. In Türkiye, KVKK (Personal Data Protection Law) regulates this; In addition, the PDR specialist has a strict confidentiality obligation arising from professional ethics. Simply typing a student's name, status, or a conversation record into a public AI tool means leaving the most sensitive data out of your control.
The rule is simple and uncompromising: Anonymize first. Remove identification information such as name-surname, school number, class-branch, parent's name and address; Use general expressions such as "Student A", "13-year-old student", "a teenager experiencing family tension". If possible, choose corporate tools that provide data processing assurance, provided by your school or ministry. We will deepen this principle in unit 10; Make it a reflex for now.
Tip: Before pasting a text into AI, search for name, number, institution and parent name with "Ctrl+F". If you find it, replace it with general phrases before sending. When in doubt: "Does this information identify the student?" ask; If your answer is yes, it will appear.
three mini cases
Case 1 — Made-up statistics. A guidance counselor asked AI for the "test anxiety rate in high school students" for a test anxiety seminar. AI gave a clear number like “73%” and a “research” name. When the teacher tried to look for the source, he saw that neither that place nor that study existed. Thanks to the verification step, he turned around presenting a fictitious number to 120 parents; instead he used a general statement whose source could be verified. Preparation took 6 minutes longer, misinformation was blocked.
Case 2 — Time savings. A PDR specialist was planning separate classroom guidance activities for 4 classes every month and spending approximately 4 hours on it. He started using AI as a draft generator: he had AI produce drafts of session plans based on the outcome and corrected them according to classroom reality. Time dropped from 4 hours per month to ~1.5 hours; He devoted the 2.5 hours he earned to individual meetings. The quality did not suffer because the final editing and filtering was done by the expert.
Case 3 — Crisis border. In one interview, a student talked about thoughts of harming herself. The expert's mind asks, "Should I write the situation to the AI and get suggestions?" It passed but stopped: it was a crisis situation. Without writing a single line to the AI, he executed the emergency protocol, informed the school administration and the family, and safely directed the student to the relevant specialist. He used AI only after the crisis, to edit the language of the minutes that recorded the incident anonymously.
Copiable templates
1) Separating the task by risk level:
Your role: WRITING assistant to a school counselor. Divide the following 8 PDR tasks into two lists: (A) low-risk tasks where the AI can confidently produce an outline, (B) high-risk tasks where the final decision must rest with the expert. Also, if a task falls into the crisis/risk area, mark it "PROHIBITED - human expert". Write a one-sentence justification for each task. Tasks: [paste your own weekly tasks here]
2) Validation check prompt:
Mark each sentence in the text below that contains a fact, statistic, quote, source, or scientific claim and add a note for each "from which source should it be verified?" Do not verify yourself; list only what needs to be verified.Text: [paste draft]
3) Anonymization pre-check:
Find each phrase in the text below that might directly or indirectly describe a student (name, number, grade, school, parent name, unique situation description) and suggest a more general alternative for each. Commenting content.Text: [paste text]
4) Supporting interview preparation questions:
Your role: PDR writing assistant. Diagnosis, guidance. I am preparing for a SUPPORTIVE meeting with a 13-year-old student (anonymous) about academic motivation. Suggest 8 non-judgmental, open-ended preparation questions. Add a warning at the top to "consult human professional/family" if signs of crisis are detected.
Weak prompt / Strong prompt
Weak:
A student of mine is unhappy, what should I do?
This prompt is context-free, may pose personal data risk, and pushes the AI into diagnosis/referral territory; produces generic, risky and useless output.
Strong:
Your role: WRITING assistant to a school counselor. Do not diagnose or give guidance. The following situation has been ANONYMOZED. Context: A 13-year-old student seems to have lost interest in class in recent weeks. Task: Suggest 6 non-judgmental, open-ended preparatory questions for a SUPPORTIVE interview with this student. If signs of crisis are detected, add a warning at the top to "consult human professional/family."
This demand encodes the role, limit, anonymity, format and crisis reflex from the beginning.
Common mistakes
- Pasting credentials. Entering the interview record or student name into the tool as is. Solution: anonymize first.
- Asking AI about the crisis. Spending minutes in case of risk. Solution: direct protocol and human expert.
- Using without verification. Passing on fabricated statistics/sources to the student or parent. Solution: connect to source.
- Waiting for diagnosis. From AI to "what's wrong with this kid?" to ask for a diagnosis. Solution: diagnosis and interpretation belong to the specialist.
- It means "AI said". Putting the responsibility on the vehicle. Solution: the final decision and responsibility belongs to the expert.
Table: Mission risk level and role of AI
Quest
Risk level
Role of AI
decision maker
Class guidance activity outline
low
Generates draft
Expert organizes
Psychoeducation brochure first version
low
draft + idea
Expert confirms
Editing conversation notes (anonymous)
medium
Format/language layout
Expert confirms
Scale result interpretation
high
Language presentation only
Expert reviews
Diagnosis / definitive guidance
high
Does not take part
expert
Crisis/risk response
forbidden
Does not take part
human expert + authority
In summary
AI is a powerful assistant that speeds up the repetitive writing load of PDR, but it is not the decision maker. Separate tasks by risk level; Use AI freely in low-risk draft work, keep high-risk decisions to yourself, leave AI out completely in crisis and risk. Verify every output, anonymise every data, always take responsibility.
Application task
Write down 10 of your own weekly PDR tasks in a list. Label each one “low risk / high risk / prohibited (crisis)”. Then, using the “strong prompt” pattern above, have an AI tool anonymously produce a draft of one of the low-risk tasks and put the output through the 4 steps of the verification discipline.
checklist
- [ ] I divided the task into risk level (low/high/prohibited).
- [ ] I anonymized the credentials before feeding the text to the tool.
- [ ] In a crisis/risk situation, I targeted the protocol and the human expert, not the AI.
- [ ] I have verified the facts, statistics, and sources in the output.
- [ ] I accepted that the final decision and responsibility lies with me.