Understanding AI as a Tool

Understanding AI as a Tool

Why it matters: You’ll get better results when you treat AI as a tool you direct, not a magic oracle.

Goals

• Explain what modern AI can and cannot do well.
• Identify 2 tasks you should not delegate to AI (without safeguards).
• Set a simple “use AI / don’t use AI” rule for your work.

Key definitions

• Model: A system that predicts outputs from patterns in data; it can sound confident and still be wrong.
• Hallucination: A plausible-sounding answer that is not supported by facts or sources.
• Guardrail: A rule or constraint that prevents unsafe/incorrect outputs from shipping.
• Verification: A check that compares output to reality (sources, calculations, tests, policy).

Workflow (step-by-step)

1. Pick one recurring task (email, summary, outline, FAQ).
2. Define what “good” means (criteria + examples).
3. Decide what must be verified by you (facts, claims, numbers, compliance).
4. Write a baseline prompt with constraints and a checklist.
5. Run once, then iterate using critique + revision.

Example (good vs bad)

✅ Good: You use AI to draft a summary, then you verify key claims against the source and keep only supported points.

❌ Bad: You copy-paste an AI-generated answer into a customer email without checking facts, tone, or policy.

Checklist

• I can describe the expected output format and constraints.
• I know what I must verify (facts, numbers, citations, policy).
• I have a fallback plan if the model refuses or is uncertain.

Metrics / criteria

• Measurable: You can say whether the output is correct/complete (pass/fail or a score).
• Effort: You can produce the result in 15–30 minutes using the workflow.
• Clarity: Another person can run your prompt and get a similar outcome.
• Risk: High-risk outputs are reviewed by a human before sending/publishing.

Common mistakes + fixes

• Pitfall: Vague instructions. Fix: Add constraints, examples, and a success checklist.
• Pitfall: One-shot prompting. Fix: Iterate: draft → critique → revise → verify.
• Pitfall: No validation. Fix: Add QA steps, tests, and explicit “what counts as done”.

Practice prompts

• Draft a response, then list what must be verified and what sources would support it.

Student tasks

1. Apply the workflow to one real work task related to understanding ai as a tool.
2. Create a small artifact: prompt, checklist, rubric, table, draft, or decision note.
3. Run one quality check and record what changed after the check.
4. Write a 3-sentence reflection: what worked, what failed, and what you will reuse.

Useful external sources

• OpenAI prompt engineering guide: https://platform.openai.com/docs/guides/prompt-engineering — Prompt structure, iteration, constraints, and evaluation habits.
• NIST AI Risk Management Framework: https://www.nist.gov/itl/ai-risk-management-framework — Risk framing, governance language, and verification mindset.
• OWASP Top 10 for Large Language Model Applications: https://owasp.org/www-project-top-10-for-large-language-model-applications/ — Practical AI safety risks such as prompt injection, leakage, and insecure output handling.
• Google People + AI Guidebook: https://pair.withgoogle.com/guidebook/ — Human-centered AI design, user control, feedback, and failure modes.

Bibliography

• OpenAI prompt engineering guide. https://platform.openai.com/docs/guides/prompt-engineering
• NIST AI Risk Management Framework. https://www.nist.gov/itl/ai-risk-management-framework
• OWASP Top 10 for Large Language Model Applications. https://owasp.org/www-project-top-10-for-large-language-model-applications/
• Google People + AI Guidebook. https://pair.withgoogle.com/guidebook/