AI Jobs in 2028: Which Roles Grow and Which Shrink

There is a lot of noise about AI and jobs. Some of it is fear, some of it is marketing. This post sticks to what the large datasets actually say, and adds what I see from the delivery side, building AI systems that change how real teams work.

The big picture: more jobs created than lost, but a lot of churn

The World Economic Forum's Future of Jobs Report 2025, based on a survey of more than 1,000 large employers covering over 14 million workers, projects that 170 million jobs will be created and 92 million displaced between 2025 and 2030, a net gain of 78 million.

The same report says employers expect 39% of workers' core skills to change by 2030. Two-thirds plan to hire talent with specific AI skills, and 40% expect to reduce their workforce where AI can automate tasks.

Read those two numbers together. Net job growth does not mean your current job is safe. It means the total number of jobs grows while the mix shifts underneath it, and 2028 sits right in the middle of that 2025 to 2030 window.

Exposure is not the same as replacement

The International Labour Organization's 2025 update to its generative AI exposure index found that one in four workers worldwide is in an occupation with some degree of generative AI exposure. The ILO's conclusion is that, because human input is still needed, most of those jobs will be transformed rather than made redundant.

This matches what I see in practice. When I build an AI document processing pipeline, the job of the person who used to type invoice data does not vanish overnight. It moves to checking exceptions, handling the cases the model flags, and fixing upstream data problems. The task count drops. The judgement part stays.

Roles that grow through 2028

People who build AI systems

The WEF lists big data specialists, fintech engineers, and AI and machine learning specialists as the fastest-growing roles in percentage terms, with software and applications developers close behind.

LinkedIn's hiring data points the same way. In its Jobs on the Rise 2026 list for the US, AI engineer is first, followed by AI consultant and strategist, with data annotator and AI/ML researcher also in the top five. LinkedIn lists the most common AI engineer skills as LangChain, retrieval-augmented generation (RAG) and PyTorch.

People who deploy AI inside businesses

A model is not a product. Someone has to connect it to a company's documents, CRM and approval rules, then keep it working. That is why roles like AI consultant and forward deployed engineer are rising. I wrote about the second one in what a forward deployed engineer does, because it is my own title.

Non-tech roles that now expect AI skills

Indeed's Hiring Lab found that, by December 2025, 45% of data and analytics job postings mentioned AI, marketing postings rose from 8.4% to 14.9%, and HR postings doubled from 4.4% to 8.8%. The growth is not only in AI titles. It is AI showing up as a requirement inside ordinary jobs.

Roles that shrink through 2028

The WEF's fastest-declining list is dominated by clerical work: postal service clerks, bank tellers, data entry clerks, cashiers and ticket clerks, and administrative assistants.

The ILO data agrees on the category. Its 2025 update found that clerical occupations continue to have the highest exposure, and that media and web-related occupations saw their exposure rise as models improved at image, video and voice generation.

The pattern is simple. If most of a job is moving structured information from one place to another, by typing, copying, checking or routing, it is the easiest work to automate with today's tools.

Grow vs shrink at a glance

The tech job market is split

It would be wrong to read this as "tech jobs are booming." Indeed reported that at the end of 2025, total tech job postings were 34% below pre-pandemic levels, while tech postings mentioning AI were about 45% higher than in February 2020. Across all jobs, postings mentioning AI were 134% above February 2020 levels while the overall postings index was only 6% above.

So demand has moved, not simply grown. A general web developer competes in a smaller market than in 2021. A developer who can ship a reliable AI feature end to end is in a much tighter one. Employers feel that too: ManpowerGroup's 2026 survey of more than 39,000 employers found AI model and application development (20%) and AI literacy (19%) are now the hardest skills to find.

What this means if you are a worker

• Move toward judgement, away from transfer. If your job is mostly moving data between systems, learn the parts around it: exception handling, quality checks, process design. Those survive automation.
• Learn to direct AI, not just use it. Writing a good prompt is table stakes. Knowing when the output is wrong, and building a check for it, is the skill employers struggle to find.
• Technical people: go end to end. LinkedIn's data shows the most common prior roles for AI engineers include software engineer, data scientist and full stack engineer. Full-stack skills plus RAG and agents is a strong combination. I cover the path in how to become an AI engineer.

What this means if you run a business

• Plan for task redesign, not headcount cuts. The ILO's "transformed, not redundant" finding is a better planning assumption than replacement. Map which tasks in a role are automatable before deciding anything about the role.
• Start with the clerical layer. Document intake, data entry and routing are where AI pays off first and where risk is easiest to control. I covered where to begin in AI automation for small business.
• Budget for skills you will not find easily. If AI development is the hardest skill to hire, decide early whether to build that team, contract it, or both. My comparison of hiring an AI freelancer vs building an in-house team goes through the trade-offs.
• Follow the money as well as the job titles. Where AI spending goes shapes which skills companies hire for; the AI industry in 2028 breaks down the forecasts.

Work with me

I am a forward deployed engineer and CTO at Avenuor, and I help companies redesign workflows around AI without breaking what already works. If you want to know which tasks in your team are ready for automation, see workflow automation and AI agent development, or book a scoping call.