Delegate, Direct, or Do · STEM
Cybersecurity Analyst: what would you hand to AI?
Alert triage is automated on both sides of the fight; judging real threats is the work.
AI changes a lot of this job
Three bins
| Delegate | AI does it and a person spot-checks the result. |
|---|---|
| Direct | AI drafts; a person steers, fixes, and decides. |
| Do | A person does it. AI stays out. |
Four questions for every task
Every sort needs a one-line reason that uses at least one of these.
- What does a mistake cost? A typo in a draft is cheap. A wrong dose, a wrong verdict, or a live wire is not.
- Can a person check the output quickly? If checking takes as long as doing it, delegating saves nothing.
- Who answers for the result? Whoever signs, submits, or says it out loud owns it, whoever wrote the draft.
- Does it involve private data or judging people? Health records, student files, and decisions about a person need a person.
Grades 6-8: Sort it
Start with these tasks, then add 3-7 more that people in this job really do. Put each in a bin and write one line on why, using the four questions.
- Sort a night's worth of security alerts by severity
- Draft an incident summary for the IT director
- Decide whether to shut down a server that may be compromised
Some jobs land mostly in Do. That is a real answer, not a failure.
Grades 9-12: Brief it, then audit it
Sort the tasks as above. Then pick one Delegate or Direct task and write a brief with four parts: the goal, the context the AI needs, the constraints it must follow, and what done looks like.
No sample output for this career yet. Practice the audit on one from the same field: Data Scientist. The printable guide has all seven samples.
A live AI tool is optional, only if your district allows student use.
Close: one task you would never hand off
Which task would you never hand off in this job, and why?
Grades 6-8: one paragraph. Grades 9-12: a short argued response that answers the strongest reason someone might give for handing it off.
Go further
Capstone: Adversarial ML & AI Security: Attack the model, then defend it.
Curriculum: grades 6-8, Unit 6 “AI at Work” · grades 9-12, “Prompting and Context” and “Tools and Agents”