IT/ AI Seminar: Practical Presentation
Topics
Each topic includes five points you have to cover
1. How AI Works (Basics)
• What an AI/LLM is, in simple terms, with everyday examples
• How an answer is produced
• A short demo: same question, different context- different answers
• Typical limitations: why AI can be confidently wrong (hallucinations), examples
• Key terms to define: model, prompt, context, token, inference… at least 10
2. Prompts: How to Ask Better Questions
• What a prompt is and why clarity matters, examples
• A simple prompt template: task + details + output format
• How to improve a prompt in 2-3 steps
• Useful prompt types: summary, email/message, step-by-step plan
• A quick demo: a weak prompt vs. an improved prompt- an example from English Studies
3. AI with Documents (RAG): Answers from Your Materials
• Why a chatbot alone is not enough for course notes or policies
• RAG in one sentence: retrieve a relevant passage, then generate an answer
• Practical use cases: syllabus Q&A, study notes, university rules, FAQs
• What can go wrong: outdated sources, missing citations, irrelevant retrieval
• Find the requirements for writing a bachelor’s thesis
4. Checking AI Outputs: How to Verify Information
• Why verification is necessary (fluent text is not proof)
• Split an answer into checkable claims (names, dates, numbers, facts)
• Verify with at least two reliable sources (and link them)
• A demo: verify three claims from one AI answer
• How to write safely: label uncertainty and list what must be checked
5. Security When Using AI: Scams and Data Leaks
• Common threats: phishing, social engineering, and manipulated instructions
• What never to share with AI: passwords, personal data, private documents… a list of
• How to spot suspicious messages (3-5 red flags)
• A practical safety checklist: 8 rules for safe AI use
• A short case study: analyze one 'suspicious' message and propose actions
6. Privacy: Personal Data and Practical Settings
• What counts as personal vs. sensitive data (clear examples)
• Where data is collected: apps, browsers, accounts, permissions…
• Data minimization: share the minimum needed, anonymize when possible
• A practical audit: review permissions for two apps and adjust them
• Safe habits: remove identifiers, manage history, and use privacy settings…
7. Fairness and Ethics: When AI Can Harm People
• What bias is (one simple scenario) and where it comes from
• Where it matters: hiring, admissions, scoring, content moderation…
• How to test: compare outcomes across different user groups or cases
• A practical task: Choose one potentially biased sentence from a short text, rewrite it in a neutral way,
and explain why the original wording is unfair
• How to communicate ethically: transparency, limits, and human oversight…
8. AI for Study and Work: Practical Use That Makes Sense
• Typical helpful uses: summaries, outlines, brainstorming, language support…
• What humans must still do: judgement, verification, accountability…
• A demo: one task without AI vs. with AI (time and quality comparison)
• Academic integrity: when AI help is acceptable and when it is not
• Build a personal workflow: three steps for using AI responsibly
9. AI Images, Audio, and Video: Deepfakes and Verification
• What generative media means (text- image/voice/video)
• What a deepfake is and why it is risky, examples
• How to verify: source, context, and suspicious details/artifacts
• A practical checklist: steps for 'fake vs real' evaluation
• Ethics: do not spread unverified content; report and verify first… summary of rules
10. Cost and 'Heaviness' of AI: Performance, Time, and Energy
• Why AI costs money: compute, model size, and response length
• Why brevity helps: shorter prompts and controlled output formats
• Cloud vs on-device: speed, privacy, and availability trade-offs
• How to reduce cost/impact: summarise, reuse, limit queries, avoid waste
• A practical design: Which AI models are available, and how much do they cost?
11. AI: Current Trends and in the Future