Przykładowa nazwa metody (pl)

Because it empowers students to defend and challenge ideas such as ChatGPT Contra rather than passively accept AI outputs, and helps educators foster critical thinking and evaluation — a core AI literacy skill increasingly important in dialogue, consulting, and R&D teams.

An AI Critical Thinking workshop is designed to give students hands-on experience evaluating AI outputs. Working in small groups, learners take on a persona (a ChatGPT critic) and generate prompts that address a specific scenario. They compare responses, look for logical flaws, hallucinations, ethical concerns, and value judgments. The goal of the workshop is not to reject AI but to build calibrated trust — students learn when to accept, when to challenge, and when to adjust an AI response.

Course: Intro to AI Management

Scenario: A class explores an AI announcement for a bank’s chatbot.

1. Each group asks ChatGPT: „Draft a customer refund letter for a bank customer who is close to churning.”
2. Groups analyze responses for empathy, tone, compliance and clarity.
3. They share top 3 red flags and rewrite one paragraph.
4. Final report: 1-minute „Response Critique” recorded by group.

Warm up & framing (5–10 min)

Pose the question: „How can AI both help and mislead us?” Collect a few quick responses.

Form teams & assign personas (5 min)

Teams of 3–4 with roles: prompter, critic, editor, presenter.

Prompting round (15 min)

Each team receives the challenge and drafts a prompt tailored to their persona and context.

Review round (10 min)

Students review, discuss and rank AI responses against a shared rubric: tone, accuracy, empathy, feasibility.

Deep dive on critique (15–20 min)

Choose the best case and unpack: what could go wrong? What assumptions did the AI make? What edge cases were missed?

Presentation & reflection (10 min)

Each group offers a two-minute pitch, reviewers give feedback and instructor closes with a synthesis: „What did we learn — and what should the AI do differently?”

• In mixed-experience groups where deeper AI literacy is emerging.
• In large classes, when the exercise is broken into asynchronous prep and synchronous critique.

• When learners lack foundational context — they may struggle to see AI errors.
• In fully virtual environments where liveliness of debate is limited.
• When learners have hard requirements about certainty over ideation — a design case may serve better.

• Students may lack AI evaluation vocabulary — encourage a shared critical framework.
• Time is short — set clear timers per phase and stick to them.

• Easier — Provide the rubric, prompts and pre-drafted responses.
• Harder — Add unknown scenario dimensions with 5 minutes of prep, then run the debrief on adversarial cases.

• Run a rehearsal prompt: show 1–2 examples with a fair critique. Sometimes learners freeze on the „attack”; scaffolds warm them up.
• Vary the scenario: student critiques thrive on real, spicy examples.
• Record everything: the video, the notes and the takeaways.
• Reserve buffer time: 15 minutes for feedback and student reactions.
• Debrief with the group and shared docs to preserve visible thinking.

Assessment is process-based. Highlight the rubric, feedback, hand-off, and include informal feedback more than „wrong or right answers.”

• Rubric-based observation of teamwork.
• Student journal reflection and self-scored contribution.
• Summary paper, group work and student written analysis on the AI critique challenge.

To have AI-driven Thinking benefit all learners and educators:

• Use clear time limits and visual timers; announce each transition.
• Provide written prompts and rubric so students can see the framework rather than remember it.
• Offer role choice — every student picks their comfort role before jumping into open discussion.
• Encourage recorded rehearsal for the presentation phase.
• Allow multiple participation styles: text-based, verbal, drawn.

These small structural tweaks let the method’s core benefits — critical AI literacy — arrive without leaving anyone behind.