Demonstrate three bounded tasks with inputs and correction rules.
You will reproduce practical document, note, learning, planning, and comparison exercises while checking results against sources.
Before you begin
Choose a reversible, low-risk task and remove private information. Keep the original notes or source beside you for comparison.
Good first uses
AI is useful for turning rough notes into a checklist, explaining an unfamiliar term at a chosen reading level, proposing meal ideas from ingredients, or comparing options against criteria you provide. These tasks are easy to review and do not require the model to act on your behalf.
A prompt pattern
State the outcome, context, constraints, and format. For example: “Turn these notes into a five-step plan. Keep every deadline unchanged. Put uncertainties in a separate list.” Then inspect the result against the notes. Clear source boundaries reduce invention.
Protect information
Do not paste passwords, authentication tokens, private identification numbers, confidential work documents, or someone else’s sensitive information into an AI service. For cloud tools, review retention and training settings. For local tools, confirm that plugins, telemetry, and integrations do not send content elsewhere.
Keep responsibility
AI can suggest; you decide. Verify appointments, prices, travel restrictions, medication information, and contractual language with authoritative sources. Do not let convenience turn an unverified draft into an automatic action.
Summarize with evidence
Use a document you are allowed to process. Ask for five bullet points and require a supporting quote or section reference for each. Compare every bullet with the original and delete unsupported material. Then ask what uncertainty the summary omits. For long documents, process labeled sections and combine only after checking. The result is a reviewable outline, not a replacement for reading material whose details or legal meaning matter.
Organize notes without losing facts
Request categories such as decisions, actions, questions, and dates. State that wording may be cleaned up but names, dates, quantities, and commitments must not change. Compare line by line and put uncertain items in an “Unclear” section rather than allowing guesses. This works for study notes and planning because the original remains the source of truth and every transformed fact can be traced back.
Learn actively and verify
Ask for a plain-language explanation, analogy, counterexample, and five questions. Answer before requesting feedback. Verify factual claims with official documentation, a textbook, or another authoritative source. For programming, request a tiny example, run it in an isolated project, and explain each line yourself. Learning improves when AI creates practice and feedback rather than doing all the thinking.
Plan without unsafe automation
Provide the goal, constraints, deadline, resources, and decisions already made. Request milestones, dependencies, risks, and questions—not automatic calendar changes. For comparisons, define criteria and mark missing data. Independently check current prices, availability, rules, and specifications. Never let an unverified answer send messages, make purchases, change accounts, or commit files. Keep human approval before actions affecting others or difficult to reverse.
Select a reviewable everyday workflow
Decision: Use local processing for suitable private drafts or a permitted hosted assistant, while keeping external actions manual.
Relevant options: ChatGPT, Claude, Gemini, and Microsoft Copilot are hosted products; LM Studio can use local models. Account and data controls differ.
Worked example: Turn non-sensitive notes into a checklist that quotes the note supporting each item.
Failure mode: The tool may invent a deadline, omit an item, or treat a suggestion as a confirmed appointment.
Validation: Compare every line with the notes; keep email, calendar, purchases, and submissions outside model authority.
Useful next step: Save one prompt template with an uncertainty section and manual approval.
Official sources checked 2026-10-24: ChatGPT · Claude · Gemini · Microsoft Copilot · LM Studio. Product availability, support, terms, pricing, and licenses can change. These are vendor or project sources; no independent benchmark is claimed.
Three reviewable mini cases
- Checklist: “Book room; ask Sam for slides; review Friday.” Preserve all items and invent no other deadline.
- Comparison: “A costs 20 and works offline; B costs 12 and needs internet.” Use only those criteria and mark missing facts unknown.
- Learning: explain a supplied paragraph using only that paragraph and quote support.
These are synthetic tests, not claims that a specific model produced a result.
Authoritative references
Reference scope checked 2026-10-27. These sources support the concepts and risk controls; product-specific behavior still requires the nearby vendor documentation and a dated test.
Your next step
Run one case, compare every statement with the input, and record omissions separately from inventions.
Editorial provenance
Nyfir Studios is the publisher and editorially responsible organization. AI tools assisted with drafting and structured review for this candidate; factual claims, test labels, links, and retained evidence require human review before deployment. No named employee, credential, or unrecorded hands-on result is implied.