AI Job Titles Jumped 173 Percent in a Year, What to Ask Before You Take One

Positions with AI in the title increased 173 percent year over year in the first quarter, based on an analysis of 7 million job postings reported by CIO Dive. In the same period, businesses across all sectors cut around 26,000 tech jobs in July while hiring at technology sector companies rose by nearly 4,000. Competition for AI talent remains strong and expensive, in the framing of a Harvard Business Review analysis of the hiring landscape, which describes the largest technology firms in an intense scramble for AI researchers and engineers even while laying off thousands of other employees. At Metaintro, we track roughly 50 million job postings , and the gap between those two markets is the most important thing an AI candidate can understand before negotiating anything.

Why Is Tech Hiring Weak and AI Hiring Hot at the Same Time?

Because they are not the same market. Businesses across all sectors cut around 26,000 technology jobs in July, while hiring inside technology sector companies rose by nearly 4,000. Technology unemployment fell for a third consecutive month to 2.8 percent, below the national rate of 4.1 percent.

The sector data shows the same split. The information sector added 11.0 thousand jobs in July, according to Current Employment Statistics from the U.S. Bureau of Labor Statistics, yet it is down 81.0 thousand over the year. Professional and business services added 18.0 thousand in the month and 115.0 thousand over the year. A monthly gain inside an annual decline is what a market looks like when it is reallocating rather than growing.

Underneath the aggregate, demand is concentrated. Enterprises are pressing on with AI adoption, and the vice president of research at CompTIA said there is still strong demand for foundational roles across support, infrastructure and data as companies begin deeper integration of AI into technology strategies. That sentence is worth reading twice if you are not an AI researcher, because it says the adjacent roles are where the volume is.

What Are Employers Actually Paying for AI Skills?

At the extreme, more than most people believe. The largest firms are pursuing AI researchers and engineers with compensation packages reportedly worth millions of dollars, and one AI company department head captured the internal logic bluntly, saying that if the company is going to spend a billion dollars to build a model, ten million dollars for an engineer is a relatively low investment.

That is the top of a very steep curve and it is not the market most candidates are in. The broader picture from May 2025 national data puts software developers at an annual median of 135,980 dollars across 1,687,890 people, data scientists at 120,230 dollars across 262,440 people, and computer and information research scientists at 140,300 dollars across 37,200 people.

Notice the shape of those numbers. The highest median belongs to the smallest group, with only 37,200 computer and information research scientists nationally against 1,687,890 software developers. Scarcity rather than title is what produces the headline packages, which is why chasing an AI job title without the underlying scarcity rarely produces the pay that comes with it. Our guides to what machine learning engineers earn and to data scientist pay by experience and city break the ranges down further.

Which Questions Actually Matter Before You Accept?

Start with compute, because it is the constraint that decides whether you can do the work at all. Ask what compute budget the team has, whether it is committed or requested, and who approves an increase. A team without a settled answer is a team where your first year will be spent waiting rather than building.

Then ask about data. Ask what data the team has access to, how clean it is, who owns it and what approvals stand between you and using it. AI roles fail far more often on data access than on modelling ability, and the failure is invisible from the outside because job descriptions never mention it.

Third, ask what has actually shipped. The distinction between a team running experiments and a team with something in production is the distinction between a role that builds your track record and one that quietly does not. Ask what is live, how many users it touches and what happened the last time it went wrong. Our guide to how agentic AI skills are defining the next hiring wave covers what production experience is now worth.

Fourth, ask who owns the decisions. If a model recommends something and a business leader disagrees, what happens. The answer tells you whether the team has real authority or is a service function producing outputs that get overruled. That single question predicts job satisfaction in AI roles better than any compensation figure.

A fifth question is worth adding for anyone joining a team that is still forming. Ask how success will be measured in the first year, and listen for whether the answer is an outcome or an activity. Shipping a capability that changes a business metric is an outcome. Running a set number of experiments or evaluating a list of vendors is an activity, and activities are far easier to reorganise away. In a market where the same employers are cutting and hiring simultaneously, being attached to an outcome is the closest thing to protection an individual can arrange.

Should You Take the Startup or the Large Employer?

They fail differently, which is the more useful way to compare them. Large employers offer compute, data and stability, and the reported packages at the top end are genuinely large. The risk is scope. A narrow slice of a large system can be excellent for depth and poor for the breadth that makes you employable elsewhere, and reorganisations move quickly in a market where the same firms are laying off thousands while hiring aggressively.

Smaller companies offer breadth and ownership, usually at lower cash compensation. The risks are compute budgets, runway and the possibility that the AI work is a feature rather than the product. Ask directly how long the current funding lasts and what proportion of revenue depends on the AI capability, because those two answers describe your actual job security better than headcount does.

There is a third option that gets overlooked. Established companies outside the technology sector are hiring for exactly the foundational roles described in the industry analysis, across support, infrastructure and data. Competition is lower, the problems are often more interesting than outsiders assume, and the work tends to reach production faster because the business need is concrete. Our reporting on why AI engineers are in demand in the public sector and on the automotive AI hiring surge covers where that demand is landing.

How Do You Get Into AI Work Without an AI Title?

Move adjacent rather than aiming at the centre. The demand described in the industry analysis sits in support, infrastructure and data as companies integrate AI more deeply, and those roles are reachable from conventional engineering, analytics and operations backgrounds. Data engineering in particular is the most direct route, because every AI programme is bottlenecked on data long before it is bottlenecked on models.

Build evidence rather than credentials. One shipped project that solves a real problem, with a clear account of what you tried, what failed and how you measured the result, outperforms a stack of certificates. The reason is that employers hiring for AI are trying to identify judgement rather than familiarity, and judgement only shows up in the story of a decision.

There is a timing argument too. Foundational demand across support, infrastructure and data is being driven by companies moving from pilots into deeper integration, which is the phase where practical experience matters more than research credentials. That window favours people who already understand a business domain and are prepared to learn the tooling, rather than the reverse. It will not stay open indefinitely, because these roles standardise quickly once the integration work is done.

Use the internal route where it exists. Employers integrating AI into existing operations need people who understand the existing operations, which is an advantage an external candidate cannot replicate. Our guides to breaking into an AI manager role and to moving from coding into AI leadership both describe that transition, and how to become an AI ready coder covers the technical foundation.

Is the AI Title Itself a Risk?

It carries one, and it is worth pricing. A 173 percent year over year increase in postings carrying AI in the title tells you the label is being applied enthusiastically, which means a meaningful share of those roles will be conventional work with a new name attached. The way to test it is to ask what the person in this role did last quarter, in specifics. Vagueness is the tell.

The second risk is the reverse of the first. Titles that track a funding cycle can disappear when the cycle turns, which is why building capability rather than collecting a label is the durable strategy. The underlying skills of data handling, evaluation, deployment and judgement about where automation belongs survive any rebrand.

The third risk is concentration. Building a career entirely inside one company's tooling makes you extremely valuable there and less portable everywhere else. Given that the same firms cutting thousands of jobs are simultaneously bidding for AI specialists, portability is worth protecting deliberately even during a hiring boom.

What Does the Broader Market Mean for Your Timing?

It means negotiating leverage is real but uneven. Technology unemployment at 2.8 percent sits well below the national rate of 4.1 percent, which is a strong position for anyone currently employed in the sector. At the same time the information sector has shed 81.0 thousand jobs over the year, so the confidence should be specific to your skills rather than general.

The practical read is that this is a good moment to move for the right role and a poor moment to move for a marginal pay increase. The differentiators worth moving for are production exposure, data access and decision authority, because those compound into the next role. A twelve percent raise into a team that ships nothing does not.

If you are being recruited actively, remember that the intensity is concentrated at the specialist end and the volume is in foundational roles. Knowing which of those two conversations you are in prevents both undervaluing yourself and overplaying a hand. Our guide to how AI role searches have grown many times over covers how many people are now competing for the same pivot.

How Do You Evaluate an Offer Beyond the Number?

Separate the offer into four things that behave differently over time. Cash is the only one that is certain. Equity depends on an outcome you do not control. Scope determines what you can claim on your next application. Learning determines whether you are still competitive in three years. Ranking those four honestly before you look at any offer stops the largest number from doing all the deciding.

Then test the scope claim against reality. A job description promising ownership of a model in production is easy to write and easy to check, because you can ask what is live today and who maintains it. If the honest answer is that nothing has shipped yet, the role may still be worth taking, but you are joining a research effort rather than a production team and should price the risk accordingly.

Watch for the responsibility trap as well. Roles that carry accountability for AI outputs without authority over the data, the compute or the deployment decision are the ones people leave fastest, because every failure is attributable and every fix requires someone else's approval. Asking who signs off on a release surfaces this in one question.

Finally, weigh the team you would learn from. In a field moving this quickly, the single largest determinant of your capability in two years is the standard of the people reviewing your work. A slightly lower offer alongside people who will correct you is worth more than a higher one where you are the most experienced person in the room, and that calculation holds at every level of seniority.

What Does This Mean for Your Career Right Now?

Four moves. First, work out which market you are in. If your skills are scarce, the compute and data questions above are the ones that matter, and pay will follow. If they are foundational, the volume is genuinely there in support, infrastructure and data, and the fastest route in is adjacency rather than reinvention.

Second, ask the four questions before you accept anything. Compute, data, what has shipped, and who owns the decisions. An employer that answers all four crisply is rare and worth choosing even at slightly lower cash.

Third, keep your evidence portable. Document what you built, what you evaluated and what you rejected, in terms that make sense outside your employer's internal tooling. That record is what converts a boom time job into a durable career, and it is the same discipline that separates strong AI practitioners from enthusiastic ones.

Fourth, do not read the layoff headlines as the whole story. The same firms cutting thousands of roles are pursuing AI researchers with packages reportedly worth millions, while technology unemployment sits at 2.8 percent. The market is not shrinking. It is repricing, and the people who benefit are the ones who can describe precisely what they are worth to it.

Related Articles

• Japan's Tech Jobs Boom and AI Engineering Talent
• AI Skills Now Appear in a Large Share of Tech Job Postings
• The AI Skills Employers Actually Want
• Where Meta Engineers Land Next
• Defence Tech Is Hiring Engineers
• How to Negotiate an AI Salary
• AI Infrastructure Budgets and Tech Hiring
• What AI Salaries Mean for Housing Costs
• Entry Level Tech Hiring and the Rise of AI Roles
• AI Skills for Non Technical Roles

People Also Asked

Q: Are AI jobs actually growing or is tech shrinking?

A: Both are happening at once. Positions with AI in the title rose 173 percent year over year in the first quarter across an analysis of 7 million postings, while businesses across all sectors cut around 26,000 tech jobs in July and technology sector companies added nearly 4,000. The information sector is down 81.0 thousand over the year even though it added 11.0 thousand in July.

Q: What do AI specialists earn?

A: May 2025 national medians put software developers at 135,980 dollars, data scientists at 120,230 dollars and computer and information research scientists at 140,300 dollars. The headline packages reported at the largest technology firms run far above those medians and reflect scarcity at the specialist end, not the typical market.

Q: What should you ask in an AI job interview?

A: Four things. What compute budget the team has and who approves more, what data the team can access and who owns it, what has actually shipped to production, and who decides when a model recommendation conflicts with a business view. Those four answers tell you whether the role will build a track record or stall, which matters more than the salary band.

Future proof your career by choosing the AI role that builds evidence rather than the one with the best title. Metaintro tracks roughly 50 million job postings across specialist and foundational AI roles, and you can create a free account to see which employers near you are hiring for the work you actually want to do.