While the technology is lowering barriers to entry in some occupations, it is raising expectations in others. As AI takes on a larger share of routine and technical tasks, human judgement and the ability to work effectively with AI are becoming increasingly important.
The report finds that job creation has been slower in occupations with the highest exposure to AI than in less exposed roles. Its impact, however, varies significantly across professions. In some fields, AI allows more candidates to perform tasks that previously required specialist expertise. In others, it removes routine work and shifts the focus towards more complex responsibilities, making entry requirements more demanding.
It also shows that employers are changing the skills they seek in AI-exposed occupations. New capabilities are appearing more frequently in job advertisements, while candidates who can use AI effectively are in growing demand and often receive higher salaries.
Occupations most exposed to AI
Of the 382 occupations included in the international ISCO-08 classification and analysed in the PwC report, 201 have high exposure to AI, while 181 have low exposure. The exposure score was calculated according to how many of the tasks specific to each occupation can be performed partly or entirely by AI.
Exposure is naturally higher in professions based mainly on office work, including cognitive and analytical tasks, and lower in occupations involving physical work or distinctly human and emotional abilities that are harder to automate.
The occupations most exposed to AI include roles in finance, such as financial analysts, financial advisers, finance managers, accountants, brokers, credit officers and tax officers, as well as professional services roles in areas including law, HR and communications.
Exposure is also high in engineering, IT&C, science and education, ranging from architects, engineers and software developers to physicists, mathematicians, sociologists and teachers.
Not all professions are affected in the same way
The impact of AI on these occupations differs from one case to another, leading to the emergence of two distinct categories of professions.
In some cases, AI takes over repetitive, administrative or time-consuming activities, leaving employees with tasks that require greater expertise, decision-making, evaluation or interpersonal interaction.
For example, in recruitment, AI can conduct the initial screening of CVs that meet certain criteria, while the recruiter focuses on interviews and salary negotiations.
These are AI-professionalised occupations, where technology removes routine tasks and raises the level of requirements. Professions in which AI raises the entry threshold are those where knowledge and specifically human abilities remain more important than the technical work that technology can perform.
This category includes public relations officers, recruitment agents, engineers, dealers and brokers, sales and marketing managers, PR managers and pharmacists.
In other cases, AI can perform some of the more complex activities included in a job description, leaving employees with tasks that require only basic skills.
For example, in an inventory management role, AI can maintain stock records, while the employee handles the physical movement of goods.
These are AI-democratised occupations, where technology lowers the barrier to entry and makes jobs accessible to a larger number of candidates. They include web technicians, software developers, accountants, credit officers, finance managers, store managers, financial analysts, lawyers and journalists.
According to PwC data, just over one-third of AI-exposed occupations are professionalised, while almost two-thirds are democratised.
Professionalised jobs are growing faster
Although democratised jobs are more numerous, professionalised roles appear to be gaining ground more rapidly. According to the PwC report, job openings in AI-professionalised occupations are growing twice as fast as those in democratised occupations.
This suggests that employers have a greater need for specialists capable of amplifying the results generated with the help of AI.
Professionalised occupations also tend to record stronger salary growth. Advertised salaries for these jobs have increased by 37 percent compared with 2021, against an average increase of 26 percent for democratised occupations.
Moreover, seven of the ten occupations with the fastest salary growth globally between 2019 and 2025 are AI-professionalised roles: research and development managers, up 81 percent; veterinarians, up 77 percent; chemists, up 63 percent; pharmacists, up 59 percent; power plant operators, up 56 percent; physicists, up 56 percent; and psychologists, up 53 percent.
The difference is reflected not only in salaries, but also in the structure of the skills employers require.
In professionalised occupations, AI removes the standardised part of the work and leaves more sophisticated activities at the centre. As a result, companies demand a broader set of expert-level, profession-specific skills, often linked to certifications or expertise that is difficult to transfer.
In democratised occupations, the emphasis shifts more towards broad, transferable skills that allow candidates from a wider range of professional backgrounds to enter the field.
Higher requirements in the early stages
AI is also accelerating changes in the skills profile required for entry-level roles.
According to the AI Jobs Barometer, entry-level roles in occupations most exposed to AI are seven times more likely to require skills previously associated with senior employees, including leadership, strategic thinking and data-driven decision-making, than roles in occupations with lower AI exposure.
Jobs most exposed to AI are also more likely to require abilities such as empathy, judgement and creativity.
These developments show that AI does not simply replace human work, but redistributes it. In some professions, technology broadens access and increases the number of opportunities. In others, it compresses the execution component and increases the share of activities requiring advanced skills, leading to higher salaries but also more demanding professional requirements.
AI skills are increasingly sought after and better paid
The report also shows that jobs requiring specific AI skills no longer represent a niche, but a rapidly expanding segment of the labour market.
Globally, the number of AI specialist jobs increased by 69 percent between 2024 and 2025 alone, while the total number of advertised jobs rose by 8.6 percent.
Requirements for these skills vary considerably between sectors. Around 20.4 percent of job advertisements in IT and computer science require AI-related skills, compared with 12.1 percent in marketing and PR and 8.8 percent in science and research.
At the opposite end of the spectrum, the share remains low in sectors such as transport, healthcare and hospitality, reflecting the more limited impact of AI in these areas so far.
In most professional fields, the majority of occupations requiring AI skills fall into the category of AI users, meaning positions in which employees use artificial intelligence tools in their day-to-day work without developing the systems themselves.
In many non-technical sectors, including marketing, sales and human resources, more than 95 percent of roles requiring AI skills are user roles.
Demand for AI developer positions, referring to those who build such systems, remains concentrated in IT and research.
Jobs requiring AI skills are better paid than those that do not, showing that employers increasingly reward the ability to use AI to produce concrete results.
In 2025, the salary premium for positions requiring AI skills compared with those that do not reached 88 percent in sales, 87 percent in customer relations, 75 percent in legal and compliance roles, and 50 percent in finance.
In IT, the gap was smaller, at 28 percent, suggesting that AI skills have already become more widespread in the sector.
Challenges for candidates and employers
For candidates and professionals, the message is relatively straightforward. In the age of artificial intelligence, employability increases when people can do two things at the same time.
The first is to use AI for the tasks it performs well. The second is to develop the abilities that technology cannot easily replace, including judgement, interpersonal skills, creativity, logic and the capacity to make decisions in complex situations.
For employers, however, the implications are more nuanced.
While expanding AI adoption has proved to be a challenge they can address, with results varying directly according to investment in tools and training, the question of the required skills mix is less straightforward.
AI professionalisation requires skills that are difficult, and in some cases impossible, to develop or acquire without a foundation in early and formal education and extensive practical exposure, whether these involve sophisticated cognitive skills or emotional abilities.
Moreover, beyond simply using tools to improve individual efficiency, both professionalisation and democratisation require changes to how work is organised and how tasks are allocated between people and technology in a structured and deliberate way. This is a complex, long-term exercise.
For public policymakers, the skills gap will clearly have a profound economic and social impact.
Without high-quality human resources in the areas that matter, the only remaining competitive advantage is cost, meaning people who are cheaper than AI. At the same time, without investment in at least basic AI skills, alongside relevant cognitive and emotional abilities, a significant share of the population will be left behind.
Author: Oana Munteanu, Director, Workforce Services Leader, PwC Romania