Activate Your Agency - When human agency meets AI agency

Name: Activate Your Agency - When human agency meets AI agency
Rating: 4.5 (1 reviews)

Decide what AI does, where it stops and what you check, then stand behind the result.

Created byDihan Zhang

Last updated 10/2026

English

What you'll learn

Frame AI-supported work by defining the intended result, human-owned standard, consequential judgment, and people involved.
Delegate deliberately by deciding what AI does, what it suggests, what a person must decide, and where AI must stop for review.
Verify what matters by identifying a decision-critical claim, selecting and completing the appropriate checking action, and making an evidence-informed decisi
Own the result and improve the workflow by protecting one necessary capability, making shared responsibility visible, involving the people with relevant autho

Course content

7 sections • 31 lectures • 1h 37m total length

Would you approve this report?3:23
Decide in thirty seconds whether to send an AI-written report, and note why: that honest call is your starting point.
Find what the report left out2:00
Compare the report with its sources and watch a target date become “confirmed”, the pattern to look for in any AI summary.
Why the report sounded so sure3:16
Trace the report's false certainty to its instruction, sources and workflow, then spot one overconfident sentence in your own week.
Lab: Choose your starting point2:24
Choose the supplied case or your own task, and record how it runs today as your honest current status.

Agency: the capacity to influence5:05
Only a snap judgement fits into four seconds; learn the four parts of agency that must happen earlier.
Tell human and AI agency apart4:31
Hand work to AI as you would to a colleague, keeping the intent, the checking and the accountability, and check what your tool can reach.
Start before the report exists4:36
Work backwards from the decision the report must support, and write the standard before drafting.
Lab: Frame your work1:21
Six minutes: write your Frame, starting from the person who needs to decide.

Turn “use AI” into decisions3:47
Split “use AI” into three decisions so your judgement goes where it matters, and start asking “for which part?”
Can do isn't should do3:53
What AI can do isn't what it should do, so learn to ask who gave it permission.
Stop and name who checks5:05
Turn “a human stays involved” into two lines: exactly where the AI stops, and who takes over.
Delegate the weekly report4:57
Decide what AI should do on the weekly report, where it must stop, and who decides next.
Lab: Draw your delegation line1:27
Ten minutes: pick your riskiest step and give it an AI role, a checkable stop point and a reviewer.

Think before AI drafts2:49
Write three lines before you open the tool, so the first fluent draft doesn't define what's good for you.
Build Goal, Guardrails and Proof4:26
Turn Frame and Delegate into a brief AI can follow, then add Proof: show sources and mark what's unconfirmed.
Brief AI on the weekly report3:20
Compare the original and briefed reports, and see the unconfirmed points stay visible when Proof asks for them.
Lab: Build and test your Working brief1:03
Ten minutes: build your Working brief, test it, and write one correction you'd actually send back.

You can't check everything3:11
You can't check every line, so let the cost of being wrong decide where your time goes.
Find the claims that matter2:04
Rank claims by one question: if this is wrong, what does somebody do differently?
Sort the other claims2:47
Sort the last two claims yourself, including a recommendation with no source.
Choose the right check4:28
Match each claim to a Quick review, Source check or Independent review using two yes-or-no questions.
Check the weekly report3:17
Run the checks and end each one with a decision others can act on: use, revise, hold or reject.
Lab: Verify one claim that matters1:14
Twelve minutes: check one claim that matters and record whether to use, revise, hold or reject it.

Own your part, not everything4:05
Own your part of the result, and pass the other decisions to the people who should make them.
Keep human capability3:53
Name the one thing you must stay able to notice, judge, explain or recover, however good the AI gets.
Collective agency: own it together3:24
See who else holds the decisions, and ask each person for the one thing only they can do.
Lab: Turn your check into action1:04
Eight minutes: turn your check into next steps, with a clear action beside every name.

Put it all together3:42
Follow one delivery date through your whole Workcard and watch the four steps work together.
Write one next-use rule3:36
Turn your Workcard into one rule (“When…, before…, I will…”) small enough to use on a busy day.
Lab: Place your next-use rule1:11
Nine minutes: check your Workcard for gaps, then write your rule and put it where the work happens.
Activate your agency now2:08
Use your rule this week; if the moment slips past, fix the rule and keep iterating.

Requirements

Knowledge worker who already uses generative AI and understands basic practices such as prompting.

Description

This course contains the use of artificial intelligence (AI). It serves professionals who use AI in their work and answer for the value of the work that goes out under their name. You'll learn to direct, check and stand behind work done with both human and AI agency, using a supplied case or one real task of your own, and finish with a completed Agency Workcard and one rule for the next time that work comes round. The four steps (Frame, Delegate, Verify and Own) are a practical synthesis of published research on human agency (Bandura, 2001), appropriate reliance on automation (Lee & See, 2004) and human-centred AI that keeps people in control (Shneiderman, 2020). The course also draws on Martin Seligman's argument in Agency (Seligman, 2026) that resources alone don't move people forward; people also need to believe their actions can make a difference. I'm Dihan Zhang, a practitioner and researcher in AI transformation, organisational change and capability building, with more than 15 years of workplace experience. I built this course the way it teaches: human agency set the direction and made the calls, while AI agents helped with production. Your feedback will shape the next version, so please tell me what works and what doesn't.

How the course works

About two and a half hours in total: roughly 90 minutes of video and an hour of guided exercises.
One practical example taken through four steps: a simulated project report we supply, or a safe task of your own.
Seven short labs in which you build your Agency Workcard.

No technical background or particular AI tool needed.

References

Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26.
Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80.
Seligman, M. E. P. (2026). Agency: The psychological history of human progress. John Murray One.
Shneiderman, B. (2020). Human-centered artificial intelligence: Reliable, safe & trustworthy. International Journal of Human–Computer Interaction, 36(6), 495–504.

Who this course is for:

The intended learners are general knowledge worker who are not studying AI as a researcher or completing an academic course. They are trying to produce credible work under ordinary conditions such as deadlines, limited attention, incomplete information, organizational rules, and pressure to work faster.

Activate Your Agency - When human agency meets AI agency

What you'll learn

Explore related topics

Course content

Opening: let's work this out together4 lectures • 11min

Frame: decide what the work is for4 lectures • 16min

Delegate: decide where AI stops5 lectures • 19min

Working brief: tell AI your decisions4 lectures • 12min

Verify: check what could change a decision6 lectures • 17min

Own: act on what you found4 lectures • 12min

Complete the loop: one rule for next week4 lectures • 11min

Requirements

Description

Who this course is for: