Diligence: Creation, Transparency, and Deployment
Take responsibility at every stage: choose and use AI thoughtfully, be honest about AI's role where it matters, and own the output once it leaves your hands.
“The AI Did It” Is Never the Answer
If a colleague’s report contains an error, “my intern wrote that part” doesn’t get them off the hook. They put their name on it. AI works the same way. Diligence is taking responsibility for what you do with AI and how you do it, and the framework breaks it into three stages:
- 1 Creation diligence
Choose tools and use them thoughtfully
- 2 Transparency diligence
Be honest about AI's role where it matters
- 3 Deployment diligence
Own and verify what you share
Creation Diligence: How You Work With AI
Creation diligence is about the choices you make while working:
- Which tool, with which data? Use tools your organization or school approves. Personal free accounts may handle data very differently from enterprise tools with data agreements.
- What are you sharing? Don’t paste confidential business information, personal data about others (health, finances, performance reviews), or anything you’ve promised to keep private into a tool not approved for it. When in doubt, remove identifying details or leave it out.
- Whose work is it built on? Be careful asking AI to imitate a specific living creator’s work, or to rework copyrighted material you don’t have rights to use.
- Who could be affected? Consider whether AI’s known weaknesses, like bias in how it describes groups of people, could cause harm in this particular task.
Transparency Diligence: Being Honest About AI’s Role
Transparency means being honest with the people who have a stake in how the work was made. The right level depends on context:
| Context | Typical expectation |
|---|---|
| School assignments | Follow your instructor’s policy exactly. If none is stated, ask before using AI, and disclose how you used it. |
| Published writing, research, journalism | Disclosure is often required by the publication or institution. Check its policy. |
| Client or customer deliverables | Follow contracts and your organization’s policy. Clients may have a right to know. |
| Internal routine work (formatting, a first draft you rewrote) | Often no formal disclosure, but follow team norms and never misrepresent. |
| Anything where the reader assumes it’s personally yours (a condolence note, a reference letter) | Think hard. The value of the message may depend on it being genuinely yours. |
The test: would the people affected feel misled if they knew how this was made? If yes, be transparent, or rethink the approach.
Deployment Diligence: Owning the Output
Deployment diligence covers what happens when AI-assisted work leaves your hands: sent, published, submitted, or used to make decisions.
- Verify before you ship. Apply your discernment checklist, scaled to the stakes. The higher the stakes, the more thorough the check.
- You are accountable. Errors, bias, or plagiarism in AI-assisted work are yours to answer for.
- Agency raises the bar. If you deploy an AI that acts for you, like a chatbot or an automated workflow, you’re responsible for its ongoing behavior. Monitor it, and give people a way to reach a human.
Common mistake: adding “This was generated with AI” to a document and treating that as permission to skip verification.
Transparency and deployment diligence are separate duties. A disclosure tells readers how the work was made. It doesn’t make errors acceptable. You still own the accuracy of what you share.
- Pick a piece of AI-assisted work you’ve produced recently, or the output from an earlier Try It in this course.
- Answer in writing:
- Creation: Which tool did you use? Was it approved for this kind of data? Did you share anything you shouldn’t have?
- Transparency: Who has a stake in this work? Would they feel misled if they knew AI was involved? What disclosure, if any, is expected?
- Deployment: What did you verify before sharing? What would happen if a specific claim was wrong?
- Look up your organization’s or school’s AI policy. If you can’t find one, write down whom you’d ask.
- Note one change you’ll make to how you work, based on this review.
1. You need help summarizing customer complaints that include names and account numbers. What's the most diligent approach?
Creation diligence means matching data to approved tools and removing identifying details when you can. Temporary chats and 'please forget' requests don't change the data terms of an unapproved tool.
2. Which question best captures transparency diligence?
Transparency is about honesty toward people with a stake in the work. Their reasonable expectations, and any formal policies, determine what disclosure is needed.
3. A team deploys an AI assistant to answer benefits questions for employees. Who is responsible when it gives wrong answers?
Deployment diligence means owning what you put into the world, including AI that acts on your behalf. That includes monitoring and human escalation paths.
Up Next
The last lesson brings all four competencies together into a personal AI practice you can use every day, and a statement of how you’ll use AI that you can share with your team, manager, or instructor.