AI-Enhanced Pedagogy: Future-Ready Education

Name: Formative Assessment and Feedback in the Age of AI
Price: 400 EUR

Why this is Erasmus+ eligible

Structured, multi-day programme (5 days) with a pre-defined schedule and learning outcomes — not a passive lecture format.

Built around interaction: participants engage directly with the trainer and with peers from other countries throughout the course.

Transnational by design — open to participants travelling from outside the host country, a core requirement of Erasmus+ mobility funding.

Aligned with the Erasmus+ horizontal priority: Digital Transformation.

The Erasmus+ Programme Guide names digital transformation as a horizontal priority, stating that "the Erasmus+ programme plays a key role in supporting citizens of all ages in acquiring the digital skills and competences they need to live, learn, work, exercise their rights, be informed, access online services, communicate, critically consume, create and disseminate digital education content" (Erasmus+ Programme Guide 2026, Part A – Priorities of the Erasmus Programme).

Source: Erasmus+ Programme Guide 2026, Key Action 1 — School Education Mobility.

Overview

Most classrooms already collect more evidence about learning than anyone acts on. A quiz result, a show of hands, a set of exit tickets — the information arrives, and then the next lesson happens the way it was already planned. Adding AI to that does not fix it. It makes the collecting faster.

The framework the week runs on has a shape worth knowing. Three questions — where the learner is going, where the learner is right now, and how to get there — cross with the three people who can act on them: the teacher, the learner’s peers, and the learner. That grid produces five strategies: clarifying what counts as success, engineering tasks that elicit evidence of learning, providing feedback that moves learners forward, activating students as learning resources for one another, and activating students as owners of their own learning. Dylan Wiliam and Marnie Thompson set it out in 2007, and the week takes one strategy per day.

AI enters at the points where it changes what this work costs in teacher time, and the week is equally clear about where it does not. It can produce a set of contrasting sample answers in minutes; building that set by hand used to take a teacher years. It can draft a quiz from a topic or a PDF in under a minute — but the distractors it writes are merely plausible, and a plausible distractor only tells you a student guessed. It can write thirty comments in the time it takes to write one, which is precisely the risk: in the largest review of feedback research, more than a third of feedback interventions left performance worse than no feedback at all.

Everything runs on one task you already set your own students. By Friday that task has success criteria that separate strong work from work that is merely adequate. It has a question whose every wrong option maps onto a misconception your class genuinely has. And it has three routines around it: feedback with response time written into the lesson, a peer review protocol, and self-assessment that still works when every student has an assistant of their own.

The honest scale of it: when Wiliam’s own programme was tested in a randomised trial across 140 English secondary schools, the gain was real, it was modest, and it took two years of teachers meeting monthly to produce. This is a week that starts that work, not one that finishes it — which is why the last session is spent building the group you will keep it going with.

Methodology

One task, all week

every exercise runs on a single assignment you already set your own students, rather than a case study. What you build on Monday is what you test on Friday.

Evidence before tools

each day names the pedagogical decision first and the software second, so the routine survives a tool changing its interface, its pricing or its owner.

Something written every day

five short documents you can use in your own classroom the following week — success criteria, a diagnostic question, a feedback rule, a peer protocol, a community plan.

Named tools, no allegiance

you work with quiz and polling platforms such as Kahoot, Quizlet or Mentimeter and with whichever general-purpose assistant your school permits. None of them is presented as the answer.

What you'll take home

Write learning intentions and success criteria for one of your own tasks, and test whether they actually separate strong work from middling work using three contrasting sample answers.

Build a diagnostic question in which every wrong option corresponds to a misconception your own students genuinely have, then read what the pattern of responses tells you.

Judge machine-drafted feedback against one test — does it tell the student what to do next, and does it leave the thinking to the student — and rewrite the comments that fail it.

Write response time into a lesson plan, so that feedback is something students act on rather than something they are handed.

Run a peer review round that meets the two conditions the research attaches to it: a shared group goal and individual accountability.

Teach a self-assessment routine that still does its job when every student has an AI assistant in their pocket.

Set up a monthly teacher learning community in your own school from a one-page plan with the first three agendas already written.

Daily programme

Warm-up and introductions, the grade you would give: everyone marks the same short anonymous piece of student work and posts a grade. The spread across the room is usually two or three grades wide. Introductions run off that disagreement instead of round the table, and the disagreement is the day's subject.
The difference between a learning intention and the activity that carries it: why stating the task at the start of a lesson is not stating the goal, and what changes when the goal is written first and the activity designed backwards from it.
Success criteria a student can actually use: working through Sadler's condition, that the student has to hold a notion of quality roughly like the teacher's, and testing each criterion for whether it separates good work from adequate work or merely describes the task.
Where AI changes the economics: generating a set of contrasting sample answers — one strong, one middling, one weak — for your own task in minutes, using the assistant your school permits. Building this set by hand used to take years of collected student work. Participants then edit what comes back, because a model's idea of weak work is not the same as your class's.
Output: for one task you actually set, a learning intention, success criteria that discriminate, and three contrasting sample answers you have checked and corrected yourself.

Warm-up, the misconception museum: in pairs, participants trade the one wrong answer their students produce every single year, and each tries to reconstruct the reasoning behind the other's. The collection becomes the raw material for the rest of the day.
The diagnostic question: designing an item where the answer tells you whether to move on or stop, and where each wrong option is attached to a specific, named misconception rather than being merely incorrect.
What the AI builders actually do, tested live: tools such as Kahoot or Mentimeter will generate a full set of questions from a topic, a PDF, a webpage or a slide deck in under a minute. Participants run their own topic through one, then compare the generated wrong options against the misconceptions collected in the warm-up. The gap between the two is the session's finding: the machine writes plausible distractors, and a plausible distractor only tells you a student guessed.
Running the item on the room: the group answers each other's questions live on a quiz or polling platform, then reads the response pattern together — not who got it wrong, which the platform reports, but what the pattern of wrong choices says about what they were thinking, which it does not.
Output: one diagnostic question on your own topic, every wrong option mapped to a named misconception, tested on the group and revised after.

What the feedback research actually found: Kluger and DeNisi's review of the feedback literature, in which more than a third of feedback interventions left performance worse than giving no feedback. The pattern behind the failures — feedback that pointed at the person rather than at the work.
Grades and comments together: Butler's studies, in which comments alone produced gains while grades alone and grades-with-comments did not, because the student reads the score first and their neighbour's score second.
Why this becomes urgent now: the constraint that used to limit feedback was the teacher's evening. AI removes that constraint, which means it is now trivially easy to produce far more of exactly the kind of feedback the research found made things worse.
The rewriting workshop: participants take machine-drafted comments on the sample answers built on Day 1 and rewrite each one against a single test — does it say what to do next, and does it leave the thinking to the student. Wiliam's own forms are worked through as models, including the one that returns the work rather than correcting it: five of these answers are wrong, find them and fix them.
The part that is not optional: booking the lesson time in which students respond to the feedback. A feedback routine with no response slot is a marking routine with better wording.
Where this course stops: what a school may automate in assessment, what pupils have to be told, and where their work goes are the subject of our separate course on AI-assisted learning. Today is only about whether the comment works.
Output: one set of comments in two versions, before and after rewriting, plus the response slot written into a real lesson plan.

The question that will be asked in your staffroom: if an assistant gives a student instant feedback at any hour, why run peer review at all. Participants argue it both ways before testing it.
Testing it: the same piece of work goes through a structured peer round and through an AI assistant, and the group compares what each noticed, what each missed, and what each left the student to do.
Why the giver gains: the student writing the feedback has to hold the success criteria against someone else's work, which is less emotionally loaded than holding them against their own — the part of peer assessment no assistant can take over, because the benefit is in the doing.
The two conditions the research attaches: a group goal, so that students work as a group rather than merely in one, and individual accountability, so there are no passengers. Both are designed into the protocol rather than hoped for.
Running two stars and a wish properly: practising the format on the Day 1 sample answers, including what to do when the wish is wrong, and how comments in student language differ from comments in teacher language.
Output: a peer review protocol for one of your lessons, with the criteria rewritten in your students' language and the accountability rule stated.

Students as owners of their own learning: setting up traffic lights and the routines around them, including the failure every teacher meets — students show green when they know they are lost — and the move that fixes it, which is asking the green cards to explain the idea to the yellow and red ones.
The problem an assistant creates here: a student can now get where am I right now answered for them, without ever making the judgement themselves. Teaching them to use it to check a judgement they have already made, and designing one routine that still works on that basis.
The honest scale of this work: Wiliam's own programme was tested in a randomised trial across 140 English secondary schools. The effect was positive and reliable, it was equivalent to a couple of months of additional progress, and it came from two years of teachers meeting monthly in small groups. It is worth knowing what size of thing is being claimed here, and what produced it.
Teacher learning community rehearsal: in groups of five, participants run one full cycle of the meeting they intend to hold at home — each names the single technique they are committing to, what they expect to go wrong with it, and what evidence they will bring back next month; the group's job is to hold each person to one specific next action rather than to agree with them. The cycle is the thing being taken home, not the notes.
The one thing you cannot do alone: drafting a single-page request to whoever controls meeting time in your school — what is being asked for, how long it takes, and what the first three meetings would cover.
Course evaluation and feedback survey; certificates.

Please note: sessions run from 09:30 to 15:00, Monday to Friday, with a break for lunch. Timings can shift a little to suit the venue and the group. If you'd like a specific topic covered, let us know when you apply — the confirmed day-by-day schedule is sent to registered participants ahead of the course.

Certificate: we issue the Certificate of Attendance to everyone who's present for at least 80% of the course — tracked with the daily sign-in sheet (see Certificate & assessment).

Equipment: a few of our courses ask participants to bring a laptop — check the day-by-day programme above, or ask us if you're not sure.

Need a formal invoice or proforma for your school's records? Ask us — we issue official invoices with the participant's name and course details.

Materials & tools

Wiliam and Thompson's five key strategies of formative assessment, used as the working structure of the week rather than summarised in an opening slide.

Black and Wiliam on classroom assessment — both the 1998 review and the shorter Phi Delta Kappan article that popularised it, kept distinct from one another.

Kluger and DeNisi's review of when feedback fails, worked through on Day 3 as the reason the week is cautious about producing more feedback.

Butler's studies on what happens when comments are handed over alongside a grade.

A worked template carried through the week: one task, its success criteria, its diagnostic question, its feedback rule and its peer protocol, filled in day by day and taken home completed.

Hands-on use throughout of quiz and polling platforms such as Kahoot, Quizlet or Mentimeter, of Google Classroom for setting and returning work, and of whichever general-purpose AI assistant your school permits.

Certificate & assessment

This is a Certificate of Attendance, not an exam-based qualification — recognition comes from completing the structured programme, not from passing a test. The real, lasting value is in the competencies listed above under What you'll take home.

Erasmus Mobility Academy

Certificate of Attendance

Formative Assessment and Feedback in the Age of AI · 5 days · 25 hrs

+ Europass Mobility document

Type your name above — clearly marked "SAMPLE" in the PDF, not an official document.

The real certificate is signed by your trainer and EMA's authorised signatory. Your own signature isn't on this document — it's captured separately, on the Europass Mobility document.

Need to verify a real certificate? Check it online using the serial number and participant's full name, or contact us directly — [email protected], OID E10304561.

Attendance

Tracked using a sign-in sheet with one signature per participant per day (paper or digital). This record is used to issue your Certificate of Attendance and Europass Mobility document. View sign-in sheet (PDF)

Output

Every participant leaves with a completed Dissemination & School Impact Plan — a step-by-step template for applying the course back at your school (see below).

Feedback

An end-of-course evaluation form is sent to every participant after the course ends — a chance to reflect on what you've learned and how you'll apply it, and it helps us keep improving the course for future participants.

Your paperwork

Keep your travel tickets, boarding passes and (if you used green travel) any related receipts — your school or National Agency may ask for these for their own Erasmus+ reporting.

How to prepare

Bring one task you actually set your students — the worksheet, the question or the assignment brief itself, not a description of it. Every exercise in the week runs on it.

If you can, bring two or three pieces of finished student work for that task with the names removed. Where you cannot, substitutes are generated in the room on the first day.

A laptop is more useful than a tablet, since most of the week is drafting and editing. The quiz and polling platforms run in a browser.

Set up your accounts before you travel rather than in the first session: whichever general-purpose AI assistant your school permits, and at least one quiz or polling platform such as Kahoot, Quizlet or Mentimeter.

Some schools block AI assistants on staff accounts. If yours does, sort out before you travel which account you will be working from, so the first session is not spent on sign-ups.

AI rules differ from one country and one school to the next, and they are moving quickly. Before you register, check what your own ministry and your own school currently permit staff to use with pupil work. We cannot make that call for you, and the week is far more useful when you arrive already knowing where your limits sit.

Practical information

Registration & documents

Not sure how registering works or what paperwork is involved? See our full Registration & Documents guide — from applying to your Certificate of Attendance, in one place.

Invitation letter

Need a formal invitation letter for your school or embassy? Request one — we send it within 2 business days. See a sample (PDF).

Learning Agreement

Prepared by our staff and sent to you and your school once your registration form is submitted — see what it looks like. See a sample (PDF).

Billing & invoicing

We issue official invoices/proformas with the participant's name and course details — ask us. Payment is due in full by bank transfer before the course start date. See a sample (PDF).

Airport transfer

Guidance on getting from the nearest airport to the venue is included in your welcome information.

Accommodation

We can recommend partner hotels close to the venue in the host city — ask us when you apply.

For your school / proposal

School impact

After completing this Artificial Intelligence course, the participating staff member can transfer outcomes directly to their school: a short briefing for colleagues sharing the key takeaways, applying relevant tools or strategies from the programme in their own role, and inclusion of the course materials in the school's professional development resources — turning one person's mobility into a wider institutional gain that supports the school's European Development Plan. Use the Dissemination & School Impact Plan below to fill in your own — it saves as you go and can generate a one-page summary for your coordinator too.

Prefer a blank, course-independent version of this template? Get it here →

After the course

Participants will receive their certificates (Certificate of Attendance + Europass).

Participants will complete a general course evaluation and event survey.

Participants will receive course materials, videos, photos, and PDF documents in both digital and print formats.

Follow-up support: an email group is created for idea exchange, self-evaluation materials are provided, and participants receive a follow-up check-in after 3 months — timed so that the teacher learning community planned on the last day has met at least twice by then.

Frequently asked

This course is an eligible training activity under Erasmus+ Key Action 1 (School Education Mobility) — the course fee, travel and subsistence fall within the eligible cost categories. Whether it is covered depends on your own project's approved grant agreement and your National Agency's rules; we provide all the invoices, attendance records and certificates your National Agency requires for reporting.

B1 is enough — you do not need to be fluent. Trainers speak slowly and plainly, every session is built around doing something rather than sitting and listening, and groups are mixed by nationality so people help each other out. A rough test: if you could follow a staff meeting held in English and ask a question in it, you will be fine here.

Absolutely — group bookings of 3+ from one school receive priority placement, and we can schedule a dedicated session for groups of 12 or more.

Our courses run on request at our centres in Antalya, İstanbul, İzmir, Mersin, Budapest, Bremen and Rome. Tell us your preferred city and dates and we will arrange a session.

Tuition, all course materials, both certificates (attendance + Europass Mobility) and a half-day guided tour of the host city with a local guide. Flights, accommodation and travel insurance are booked separately — we send a vetted hotel list.

You can move to any later session free of charge. Cancellations made 30+ days before the start date receive a full refund; a reduced refund applies closer to the date — see our full Cancellation & Refund Policy for the exact schedule.

The Erasmus+ Programme Guide names digital transformation as a horizontal priority, stating that "the Erasmus+ programme plays a key role in supporting citizens of all ages in acquiring the digital skills and competences they need to live, learn, work, exercise their rights, be informed, access online services, communicate, critically consume, create and disseminate digital education content" (Erasmus+ Programme Guide 2026, Part A – Priorities of the Erasmus Programme). Frame it in your own project's terms — your National Agency's current call will confirm exactly how it applies.

No. The week keeps the tools you already use and changes what you do with what they tell you. No coding, no prior AI experience, and nothing to install beyond a browser.

This one is about the assessment loop — what counts as success, what the evidence tells you, and what you do next. Our AI-Assisted Learning course covers judgement and governance: whether a task should be automated at all, and what pupils have to be told. Our course on AI in Personalized Learning compares specific tools for differentiating work between students. They are designed not to overlap.

Formative Assessment and Feedback in the Age of AI