Every few weeks there’s another headline about AI wiping out a profession. Then most people go back to work and wonder, quietly, whether they should be worried. The research doesn’t give us one clean number.
It does give us a more useful way to think about the problem: AI tends to change tasks before it eliminates entire jobs. That distinction matters.
Your job title is too broad
Take “marketing.”
One person spends most of the day drafting routine copy and resizing content.
Another manages clients, makes strategy calls, presents to executives, and decides what to do when the numbers make no sense.
Same industry. Very different exposure.
That’s why research based on task exposure can be more useful than a dramatic list of “jobs AI will replace.”
Routine digital work is more exposed
Work tends to be more exposed when it is:
- repetitive
- digital
- rule-based
- easy to measure
- easy to hand off in pieces
That includes parts of data entry, transcription, customer service, routine coding, and basic content drafting.
“Exposed” doesn’t automatically mean “gone.”
It means more of the work can potentially be accelerated, changed, or automated.
Judgment-heavy and physical work is harder to replace cleanly
Jobs built around physical presence, relationships, accountability, negotiation, or messy real-world judgment are harder to automate end to end. That doesn’t mean AI leaves them alone. It may still write reports, summarize records, draft messages, analyze data, or remove administrative work. The job can change a lot without disappearing.
A rough way to think about exposure
| Work type | General exposure | Why |
|---|---|---|
| Routine data entry / transcription | High | Repetitive, digital, rule-based |
| Entry-level customer service | High | Scriptable and high-volume |
| Junior coding / boilerplate writing | Moderate to high | Routine output is easier to automate |
| Marketing and content drafting | Moderate | Drafting is automatable; strategy and judgment less so |
| Skilled trades / in-person services | Lower | Physical presence and real-world judgment matter |
| Senior strategy / leadership | Lower overall | Context, relationships, and accountability matter, though many component tasks can still be assisted |
The labels are broad.
Your actual task list matters more.
So what should you do with this?
I’d make two lists. Tasks AI can already help with in your job. Tasks that still depend heavily on your judgment, relationships, context, or accountability.
That shows you more than a generic “AI risk score” for your title. Then learn the tools that are already touching your field.
Not because everyone needs to become an AI expert. Because it’s easier to adapt to a changing workflow when you know what the tool can and can’t actually do.
The research is still moving
Different research groups measure different things: exposure, employer expectations, productivity, employment changes, or forecasts.
Those aren’t interchangeable.
Some studies project substantial disruption. Others find little evidence so far of broad economy-wide employment losses tied directly to AI exposure.
That disagreement isn’t a reason to ignore the topic.
It’s a reason to be suspicious of anyone claiming they know exactly what your job will look like five years from now.
The useful question
Instead of:
Is AI taking my job?
ask:
Which parts of my job are becoming cheaper, faster, or easier because of AI — and what does that make more valuable?
That’s a much less dramatic question.
It may also be the one that actually helps.
Related Reading
AI Tools for Beginners: The Complete Guide to Getting Started
ChatGPT vs. Claude vs. Gemini: Which One Should You Use?
Sources & Last Updated
Last updated: August 2026.