Two AI Skills Will Matter Most in 2027. Pick Yours.
For the first time ever, ManpowerGroup’s 2026 Talent Shortage Survey, the largest of its kind, found AI skills harder to hire for than engineering or traditional IT across more than 39,000 employers worldwide. That alone is a big shift. The detail most coverage skips is that it is not one AI skill gap. It is two, ranked separately, and they point in different directions.
The Two Tracks, Defined Plainly
Employers are not short on one kind of AI person. They are short on two different kinds, and most career advice treats them as the same thing.
| Track | What it actually means | Who it fits |
|---|---|---|
| Building AI systems | Designing, training, and deploying AI models and applications | People headed toward engineering, data, or technical product roles |
| Using AI well | Applying AI tools with judgment inside an existing profession: marketing, finance, operations, design | Almost everyone else, in almost any career |
Neither track is the safer bet by default. Each is scarce for a different reason, and picking the wrong one for your actual goal wastes a year you will not get back before 2027 hiring cycles set in around whichever shape the market settles into.
Even inside the second track, the real premium is narrower than “knows how to use AI.” The people actually producing results are the ones who already understand a business deeply and have learned to direct AI with that context, catching the output that is technically correct but wrong for the specific situation it was meant to solve. That combination, business depth plus AI judgment, is harder to fake than either half alone, and it is exactly what most “AI literacy” training never gets to.
Why Both Beat Engineering and IT Now
|
That is not a small reshuffling. Two AI-specific capabilities now rank above the skill categories that defined hiring difficulty for the past decade. Engineering and IT did not get easier to hire for. AI skills simply became harder to find faster, which is a different kind of signal: it means supply has not caught up with a demand curve that moved quickly, not that the old skills stopped mattering.
Where This Shows Up Hardest in the Region
This split is not evenly felt, and it is worth looking at the Gulf and Jordan separately rather than as one regional story.
In the UAE, ManpowerGroup found 76 percent of employers report difficulty filling roles, above the global average of 72 percent, with the same two AI tracks topping the list there too, according to reporting in Computer Weekly.
Saudi Arabia is approaching the same shortage from the supply side. Its national cross-disciplinary data and AI curriculum, now required for every university undergraduate regardless of major, is a direct bet on building both tracks earlier, before students ever reach the job market.
Jordan is having a more specific version of this exact conversation. Jordan News recently framed the country’s challenge explicitly as a choice between producing AI users and AI builders, tied to the objectives in Jordan’s national AI strategy, and argued that Jordan’s real opportunity is training people across many professions, not only engineers, to work competently alongside AI. Separately, PwC’s regional CEO survey found Jordanian executives more confident in their ability to attract AI talent, 65 percent, than their peers elsewhere in the Middle East, 59 percent, or globally, 42 percent, a different measure than hiring difficulty but a sign the conversation is already live at the leadership level.
How to Pick Your Track
Ask yourself two questions before you commit a year of study or training to either direction.
- Do you want to be the person who builds the system, or the person in a specific profession who knows how to get real work out of it once it exists?
- When you picture the work day to day, are you drawn to models, data pipelines, and technical architecture, or to a business problem you already understand and want to solve faster and better?
Your honest answer to those two questions points at one track, not both. Trying to be credible at both at once, without picking a direction first, is how most self-taught AI learning ends up shallow in both places instead of strong in one. That is also why so many people feel busy learning AI without feeling more hireable for it: a scattered mix of a little model knowledge and a little tool familiarity does not read as depth in either track.
What to Build Before 2027
Whichever track fits you, a resume line that says “AI skills” proves nothing right now, because that phrase is exactly what got hard to verify. What separates a real candidate in either track is a specific piece of work: a model you built and can explain the tradeoffs in, or a real business problem you solved faster and better because you knew how to use the right tool and catch where it was wrong.
We built TechX’s training around that distinction on purpose. Whichever track fits you, the work is real, matched to actual industry roles based on the skills you are building and how you perform, not a simulated exercise. Someone with hands-on experience in that specific track looks at what you produced and tells you plainly whether it holds up, not just whether it ran. If you are deciding which track to build toward before 2027, explore TechX’s programs for aspiring professionals and career changers and tell us which direction you are weighing.
Sources
ManpowerGroup, 2026 Talent Shortage Survey, press release, February 26, 2026.
Computer Weekly, “The Middle East’s AI race has a people problem,” 2026.
Optimum Partners, “The Most Expensive Mistake in AI Hiring: More AI Engineers,” April 7, 2026.