AI as a thinking partner
Use AI to explore options, challenge assumptions, and improve everyday work while keeping verification and final decisions in human hands.
AI can help you generate options, rephrase a question, or identify issues worth investigating. Its output is a starting point for your judgment, not a substitute for it. A useful workflow makes clear what the tool is helping with, what you will verify, and who makes the final decision.
Here is a practical approach for everyday work, using a fictional workshop-planning example.
1. Choose a bounded task
Start with something you can review: drafting an agenda, comparing possible discussion questions, or organizing non-sensitive notes. Avoid handing over an entire business decision without defining what a good result looks like.
Before opening the tool, write one sentence: “I want help with [task], and I will judge the result by [criteria].” For an agenda, those criteria might be a clear purpose, realistic timing, and an activity that helps participants apply an idea.
2. Give context and constraints
A vague request such as “Design a workshop” leaves many choices unexplained. A more useful starting prompt is:
I am planning a 60-minute introductory workshop for eight colleagues who want clearer project handovers. Suggest three agenda options. Include a practice activity and time for reflection. Use fictional examples. List your assumptions and the information you still need.
Then answer relevant follow-up questions. Describe the audience's experience, available time, intended outcome, and anything that must be avoided. Clearer context does not guarantee a good answer; it gives you a better basis for reviewing one.
3. Ask for alternatives before choosing
Request several approaches and compare their trade-offs. One agenda may emphasize discussion, another demonstration, and a third hands-on practice. Ask which audience each option assumes and where the timing might be unrealistic.
You can also ask: “What would make this plan fail?” or “Which parts are too ambitious for a beginner?” Treat the replies as additional suggestions, not independent validation of the original output.
4. Verify what matters
Separate ideas from factual claims. A proposed activity can be evaluated for usefulness; a quoted statistic, named source, product feature, or policy needs a reliable check outside the generated answer.
Use this review list:
- Is the information accurate and relevant to this audience?
- Are sources real, and do they support the claim made?
- Are important assumptions or limitations missing?
- Does the language imply certainty that the evidence does not justify?
- Can you explain and defend the choice without relying on the tool's confidence?
If a claim cannot be checked, remove it or clearly qualify it. For consequential decisions, involve the person with the relevant expertise and authority.
5. Protect information before entering it
Check your organization's rules and the selected tool's current data controls before sharing work material. Do not assume that every account or service has the same protections.
For an initial exercise, use invented names and fictional project details. Remove identifiers and confidential information from examples. If the task depends on sensitive material and you cannot confirm that its use is allowed, choose another task or seek the appropriate internal guidance.
6. Test, revise, and retain ownership
Choose one agenda, adjust it yourself, and try the practice activity with a colleague before running the workshop. Check whether the instructions are understandable and whether the planned timing is workable. Use that feedback to revise the plan.
Keep the final version and a short note explaining your choices. The person delivering the work remains responsible for its quality, suitability, and follow-through.
A fifteen-minute exercise
Use five minutes to define a low-risk task and write a contextual prompt. Use five minutes to compare three options. Spend the final five checking assumptions and selecting one small step to test. End by writing what you accepted, what you changed, and what still needs verification.
Your next step
Bring AI one clear task, ask for alternatives, verify the important details, and make the final choice yourself. The value comes from the quality of the thinking and review you bring to the exchange.
Explore related programs
- Practical AI at work ↗
Build a responsible AI workflow for everyday tasks: write clearer prompts, review outputs, and test useful applications with human judgment.