Course resource
How Roles Are Changing
What is actually happening to job descriptions, based on what employers are hiring for rather than on prediction.
Read the job ads, not the think pieces
The most reliable signal about where work is going is what organisations are currently paying for. It is public, specific, and updated continuously.
Here are [N] current job ads in [MY FIELD]: [PASTE]
1. Which AI-related skills appear, and how often?
2. Are they framed as "must have" or "nice to have"?
3. What responsibilities appear now that would not have three years ago?
4. What has disappeared from these ads?
5. Which of these could I demonstrate today, and which not?
Use only the ads I pasted.
Question 4 is the one people miss. Tasks quietly vanishing from job descriptions tells you more than new buzzwords appearing.
The three shifts
1. From producing to directing. Roles that were "write the thing" are becoming "specify, review and own the thing". The output volume goes up; the judgement requirement goes up more.
2. From knowing to finding and verifying. Knowing facts is worth less. Knowing which facts matter, whether a source is credible, and what a wrong answer would look like is worth more.
3. From individual output to leverage. Value increasingly attaches to how much a person enables, not how much they personally produce. The person who builds the process that ten people use out-earns the person who is very fast individually.
What appears in ads now
Increasingly common across many fields, not just technical roles:
- Comfortable using AI tools in the workflow, stated as an expectation
- Ability to evaluate AI output critically, sometimes phrased as "quality assurance"
- Prompt or context design, often not named as such
- Workflow automation
- Understanding of AI governance and data handling, especially in regulated sectors
- Named as the person responsible for a tool or process
What is not going away
Worth stating plainly, because the discourse is noisy:
- Anything requiring accountability. Someone must be answerable.
- Anything requiring physical presence.
- Relationship work — clients, teams, negotiation, trust.
- Judgement where the situation is genuinely novel.
- Domain depth. AI is broad and shallow; expertise is narrow and deep, and combining them is where the value is.
Positioning yourself
The gap that is currently undersupplied is domain expertise plus practical AI capability. Not AI expertise alone — that is becoming common — and not domain expertise alone.
| My domain depth | |
| My AI capability, honestly | |
| Where they combine | |
| What I could demonstrate today | |
| The evidence I have for it |
That last row is the one that matters in a hiring conversation. "I understand AI" is unfalsifiable; "I cut our reporting cycle from three days to four hours, here is how" is not.
Building evidence
Three things, in order of usefulness:
1. A measured result in your current role. The strongest evidence there is. A before-and-after number from real work.
2. A documented build. Something you made, with a write-up including what broke. The failures are what demonstrate you actually did it.
3. Teaching it to someone. The fastest way to find out whether you understand something, and visible to others.
Certificates and course completions are the weakest form of evidence, including this one. The certificate from this course names your capstone and its measured result specifically for that reason.
The conversation with your employer
Once you have a result:
I automated [TASK]. It went from [BEFORE] to [AFTER].
I would like to spend some of that time on [HIGHER-VALUE WORK].
Can we agree what that should be?
Result first, then the ask. This is a materially different conversation from "I have been learning AI", and it produces a materially different response.
What not to do
Do not rebrand yourself as an AI person if your value is domain expertise. You will compete with people who have deeper AI backgrounds and abandon your actual advantage.
Do not chase job titles containing "AI". Many are unstable and some are marketing. The durable roles are existing roles done with more leverage.
Do not wait to be asked. In most organisations nobody will tell you to automate your own work. The people who do it anyway are the ones who end up owning the process.
Your review
| Quarter | New evidence | Where I used it | Result |
|---|---|---|---|