← All workshops · Slides of workshop 1
The same eight statements, before and after workshops 1 to 5, to see what the participants think they learned. Its results feed the project report (Erasmus+).
Form link: TO BE COMPLETED (an online form, for example Google Forms or Microsoft Forms, and its QR code).
The same short form, twice: at the start of workshop 1 and at the end of workshop 5. Workshop 6 has no statement here: its exit card is enough. It does not repeat the exit cards, which check what people can say at the end of each workshop.
Form link: TO BE COMPLETED.
LR-K7Q2). They write it on the
back of their one-page sheet or in their phone. They answer "Before".⚠️ Important. Never keep a list of names and codes, and do not hand out the codes yourself. The code only links two answers of the same person, not the person.
💡 Tip. If you translate the form, keep the statements in the same order and the same scale of 1 to 4. Otherwise the answers of your group cannot be compared with the other groups.
1. Your code (example: LR-K7Q2: two letters for your town, then four letters or numbers you choose
and will remember). Use the same code before and after. Do not write your name.
2. When? Before the series / After the series
3. How much do you agree? One answer per line. 1 = Not at all · 2 = A little · 3 = Yes, mostly · 4 = Yes, I could explain it to someone
| # | Statement | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|
| S1 | For a chatbot I use, I can say what it reads before it answers, when, and who decides. | ||||
| S2 | I can write a system prompt (the instructions of a chatbot: who it is, its rules, its tone) that changes how it answers. | ||||
| S3 | I can explain three ways a chatbot gets its knowledge: all the documents at once, excerpts picked by a search engine (RAG), or a search it makes itself in the files. | ||||
| S4 | I can prepare a document for a chatbot so that each fact keeps its source and a person has checked it. | ||||
| S5 | I can tell when a chatbot needs a tool that reads live data (for example, today's fire danger level) and not a document. | ||||
| S6 | I can test a chatbot with the same questions after each change, and score each answer with clear rules. | ||||
| S7 | I can spot an answer, a figure or a source that a chatbot invented. | ||||
| S8 | I can build a chatbot with young people, in a tool with an Instructions box and a Knowledge box, and keep them safe while they use it. |
4. Optional. What worries you most about using a chatbot with young people?
5. Optional, after the series only. What will you try first with young people, and what do you still need for it?
Export the answers (CSV). For each statement, compare the "Before" and "After" answers of each code: only the codes that answered both times count. Then take the average per group. Results: TO BE COMPLETED after the series.
| Statement | What it is about | Workshop |
|---|---|---|
| S1 | the question of the series: what enters the model's context, when, and who decides | all |
| S2 | the system prompt | 1 |
| S3 | knowledge in the context, RAG, agentic search | 2 and 3 |
| S4 | curation of sources, the wiki | 3 and 5 |
| S5 | tools that read live data (MCP) | 4 |
| S6 | the test set, replay and the scoring rules |
all |
| S7 | the five personalities, the score 0 | 1 |
| S8 | rebuilding the chatbot with young people (WikiYouthBot) | 5 |