Not every AI-flavored job is an AI job. Sometimes the word is core to the work. Sometimes it is just sprinkled on top.
AI has become the parmesan cheese of job descriptions.
It gets sprinkled on everything.
Engineering roles, obviously.
Product roles, sure.
Marketing roles, sales roles, customer success roles, operations roles, finance roles, HR roles, legal roles, healthcare roles. Somewhere in the posting, the word shows up. AI. LLM. Automation. Copilot. Agentic. Machine learning. Generative AI.
Some of those jobs really are AI jobs.
Some of them use AI.
Some of them sell AI.
Some of them support customers who use AI.
Some of them mention AI because the company strategy deck mentions AI.
And some of them just want the word nearby.
Arthur C. Clarke wrote:
“Any sufficiently advanced technology is indistinguishable from magic.”clarke
That line still works. But job postings have created a smaller, more annoying problem.
Any sufficiently fashionable technology is indistinguishable from seasoning.
The word is not the work
The question is no longer whether a job mentions AI.
The question is what the posting means by it.
A job can mention AI because you will build models.
Or because you will use AI tools.
Or because you will sell an AI product.
Or because you will support AI customers.
Or because you will write content about AI.
Or because someone added “AI-powered” to the product description and now every job post inherited it.
Those are not the same job.
They are not even the same kind of exposure.
One requires deep technical fluency. One requires workflow judgment. One requires customer translation. One requires comfort with ambiguity. One may require nothing more than working at a company that would prefer investors not think it missed the moment.
That is why the label is not enough.
What the data showed
In the Roleworthy buzzword export, AI-related terms were not tiny edge cases. They were everywhere.roleworthy
Some of the biggest AI-related matches included:
| Term | Matching jobs |
|---|---|
| AI | 93,609 |
| Automation | 38,966 |
| LLM | 31,320 |
| Artificial intelligence | 16,627 |
| Agentic | 12,436 |
| Machine learning | 11,961 |
| Generative AI | 4,380 |
| Copilot | 2,730 |
The interesting part is not just the total.
It is where the terms show up.
For the keyword “AI,” the export found matches across engineering, sales, marketing, information technology, customer success, operations, product, finance and accounting, program, healthcare, data analytics, people, legal, risk, design, and more.
That does not mean every function has turned into an AI function.
It means AI language has moved into the general bloodstream of job descriptions.
Five kinds of AI job posts
A job seeker needs a better filter than “mentions AI.”
I would read AI postings in five buckets.
| Bucket | What it usually means |
|---|---|
| Build AI | You are creating models, infrastructure, products, or tooling |
| Use AI | You are expected to work with AI tools to do the job better or faster |
| Sell AI | You need to understand enough to explain, position, or sell the product |
| Govern AI | You are dealing with risk, policy, compliance, security, evaluation, or responsible use |
| Mention AI | The company wants the posting to sound current |
The last one is the tricky one.
It is not always bad. Sometimes a company is genuinely evolving and the job description is trying to catch up. But if AI appears only in the company boilerplate and never in the responsibilities, skills, metrics, or day-to-day work, then it may not be the role. It may just be the seasoning.
The outside research is more nuanced than the hype
The Anthropic Economic Index is useful here because it looked at real-world AI usage instead of just speculation. In a paper analyzing more than four million Claude conversations, Anthropic researchers found that AI use concentrated heavily in software development and writing tasks, but extended more broadly across the economy. They estimated that about 36% of occupations used AI for at least a quarter of their associated tasks, and that AI usage leaned more toward augmentation, 57%, than automation, 43%.anthropic
That is a much better frame than “AI will take everything” or “AI will save everything.”
In many jobs, AI is not replacing the role. It is becoming one more layer in how the role gets done.
The World Economic Forum’s 2025 report points in the same direction. It says half of employers plan to reorient their business in response to AI, two-thirds plan to hire talent with specific AI skills, and 40% anticipate reducing staff where AI can automate tasks.wef
That is not one story.
That is three stories at once.
Some jobs will build AI.
Some jobs will work with it.
Some jobs will be reshaped by it.
Some jobs will use the word before they know what it means.
What to ask when a job mentions AI
Do not be impressed by the word alone.
Translate it.
Ask:
- Is AI central to the job, or just part of the company description?
- Will I build AI, use AI, sell AI, govern AI, or support AI?
- What tools will I actually use?
- What skills are required beyond general comfort with AI?
- Does the role expect judgment, automation, prompt work, analytics, model evaluation, or customer translation?
- Is AI tied to compensation, performance expectations, or productivity targets?
- If AI is supposedly important, why does it not appear in the responsibilities?
That last question is useful.
If the posting says AI five times in the company overview and zero times in the actual role, read it accordingly.
The takeaway
AI is everywhere in job descriptions now.
That does not mean every job is an AI job.
It means AI has become a language companies use to describe the future they think they are moving toward.
Sometimes that future is real and specific.
Sometimes it is vague.
Sometimes it is exciting.
Sometimes it is just a little parmesan on top.
The job seeker’s job is not to be dazzled by the seasoning.
It is to figure out what is actually on the plate.