AI Is Taking Some Jobs. Here Are the New Ones It Is Creating.

tobywins.ai  ·  June 2026  ·  6 min read

What does the actual job loss data say?

By now you have seen the numbers. Challenger, Gray and Christmas counted 87,714 positions eliminated by Artificial Intelligence in just the first five months of 2026. Goldman Sachs puts the net loss at around 16,000 US jobs per month. The World Economic Forum projects 92 million jobs displaced by 2030.

Those numbers are real. We track them at tobywins.ai and we are not going to minimize them.

But the same World Economic Forum report that projects 92 million job losses also projects 170 million new roles created by the same forces -- a net gain of 78 million jobs globally. That second number gets far less attention.

So here is what is actually growing, where the data comes from, and what it tells you about where to position yourself right now.

What does LinkedIn's job start data actually show?

LinkedIn published its annual Jobs on the Rise report for 2026 in January, based on an analysis of millions of jobs that its members actually started over the previous three years. Not projections. Not surveys. Real job starts.

The finding that got the most attention: Artificial Intelligence Engineer ranked as the number one fastest-growing job title in the United States. Artificial Intelligence Consultant came in at number two.

But the more interesting finding was further down the list. Data Annotator made the top five. So did Workforce Development Manager. New Home Sales Specialist. Healthcare Reimbursement Specialist. Physical Therapist.

In other words, the fastest-growing jobs are not all technical. Roughly half of the roles on LinkedIn's list did not exist 25 years ago -- but many of the ones gaining ground fastest are jobs that require human judgment, emotional intelligence, and trust. Things AI cannot yet replicate reliably.

The platform also flagged something worth noting for anyone thinking about their career: founders and independent consultants are rising fast. More people are building portable careers rather than depending on a single employer. That is not a coincidence. It is a rational response to uncertainty.

What is the new type of engineering role nobody had a title for three years ago?

In June 2026, O'Reilly published a piece by engineer Artur Huk that put a name to something companies are scrambling to hire for but struggle to describe in a job posting.

He called it Context Architecture -- the work of building the rules and guardrails that govern what AI systems are allowed to do before they write a single line of code.

The problem he identified is this: AI agents generate code fast. Very fast. So fast that companies are shipping systems that look functional on day one but become ungovernable within months. The AI was never told what it was not allowed to do. Nobody built the fence before opening the gate.

The solution requires people. Specifically, people doing jobs that did not have clear titles until recently.

Architects who write boundary documents instead of reviewing pull requests. Quality Assurance (QA) engineers who define threat scenarios before any code is generated, not after. Business analysts who specify hard acceptance criteria that get fed into the AI system as constraints. Developers who spend less time writing syntax and more time governing the generation process itself.

None of these are entry-level roles. All of them require deep domain knowledge. And demand for people who can do this work is outpacing supply significantly right now.

How large is the wage premium for AI skills?

PricewaterhouseCoopers published its 2025 Global AI Jobs Barometer -- a major research effort tracking how AI skills are affecting compensation across industries. The headline number: workers with advanced AI skills earn 56% more than peers in the same roles without those skills.

That is not a small premium. A 56% wage gap for doing the same job title with one added capability set is the kind of number that restructures entire career decisions.

Gloat, which analyzed the PwC data alongside its own workforce platform data covering millions of employees, added another finding: productivity growth has nearly quadrupled in industries most exposed to AI since 2022. The companies seeing the biggest gains are not the ones that replaced workers with AI. They are the ones that gave workers AI tools and then got out of the way.

That distinction matters. The roles with the strongest growth are not "AI instead of humans." They are "humans who know how to work with AI" -- and the market is paying a steep premium for exactly that combination.

What do these three trends add up to?

Pull back and look at all three data sources together and a clear picture emerges.

LinkedIn's job start data shows the market rewarding people who can build, consult on, annotate for, and govern AI systems -- and separately rewarding people in human-intensive roles that AI cannot replace.

The O'Reilly analysis shows a new layer of engineering work emerging around governance and constraint -- roles that require both technical fluency and the judgment to say no to a system that will otherwise say yes to everything.

The PwC research shows the financial case is not subtle. A 56% wage premium is a career-level signal, not a rounding error.

The common thread across all three is the same: the workers gaining ground are not trying to out-compute AI. They are doing the things AI still needs humans for -- setting the rules, cleaning the data, building trust, making judgment calls, and managing relationships.

So where does that leave you right now?

If your current role is heavily focused on tasks that follow predictable rules and produce consistent outputs -- data entry, document processing, routine analysis, scripted customer interactions -- the risk is real and worth taking seriously. We track it at tobywins.ai using government occupational data across more than 40,000 job title variants. You can look up your specific role for free, no sign-up required.

But if your instinct is to move toward AI rather than away from it -- to learn how these systems work, to position yourself as someone who can govern them, consult on them, or do the human work they cannot replace -- the data suggests the market is ready to pay significantly for exactly that move.

The question is not whether AI is changing the job market. It is which side of that change you end up on.