If you are job hunting in Amman, elsewhere in Jordan, across the GCC, or for a remote role, you do not need another speech about AI replacing everyone. You need a map of which tasks are already changing, which claims are still early, and what to put on a CV this month. This guide is that map. It uses studies we opened and read, not recycled headlines.
Three things are inside. First, what recent Anthropic research actually measured about AI use and hiring. Second, how that lines up with the International Labour Organization and the World Economic Forum. Third, a practical section on skills, applications, and what is not worth panicking about. None of these sources is a forecast for the Jordanian labour market. Treat them as signals, then apply them to the role in front of you.
Jobs are bundles of tasks, and only some of those tasks are moving
The most useful recent work is not a single prediction. It is a series of measurements from Anthropic’s Economic Index, which studies anonymized Claude usage, plus a March 2026 labour-market paper that ties that usage to US occupations.
In Labor market impacts of AI: A new measure and early evidence (Maxim Massenkoff and Peter McCrory, 5 March 2026), the authors build a measure they call observed exposure. It asks a narrower question than “could a model do this someday?” A task counts more when it is theoretically feasible for a language model, shows up in real Claude use, is work-related, and is done in a more automated way. On that measure, computer programmers are the most exposed occupation they list, at 75% coverage. Data entry keyers are at 67%. Customer service representatives and, in the paper’s summary, financial analysts also sit among the most exposed. At the other end, 30% of workers are in occupations with zero coverage under their threshold, including cooks, motorcycle mechanics, lifeguards, and bartenders.
Theoretical possibility is much wider than current use. In computer and mathematical occupations, the paper says language models could theoretically touch 94% of tasks, while Claude’s observed coverage is 33%. Office and administrative occupations are at 90% theoretical capability. The gap is the point. Exposure on a slide deck is not the same thing as a tool already doing the work.
A companion report, Anthropic Economic Index report: Learning curves (lead authors Maxim Massenkoff, Eva Lyubich, and Peter McCrory, with Ruth Appel and Ryan Heller, 24 March 2026), looks at usage from 5 to 12 February 2026. About 49% of jobs had seen at least a quarter of their tasks performed using Claude. On Claude.ai, augmentation, where the person stays in the loop, increased slightly. The ten most common tasks were 19% of Claude.ai traffic in that February sample, down from 24% in November 2025, so use was spreading rather than sitting in a handful of coding chores. People who had used Claude for six months or more had a higher success rate in their conversations, about 10% higher in the report’s summary, and the association was not explained away by the task, the country, or the model.
What workers expect, and what has not shown up in unemployment
The labour-market paper’s early result is easy to misread in both directions. Using the US Current Population Survey, Massenkoff and McCrory find no systematic increase in unemployment for highly exposed workers since late 2022. They do find suggestive evidence that hiring of younger workers has slowed in exposed occupations. For workers aged 22 to 25, the job-finding rate in occupations with no AI exposure stayed around 2% per month, while entry into the most exposed jobs fell by about half a percentage point. Averaged over the post-ChatGPT period, that is a 14% drop in the job-finding rate versus 2022 in those exposed occupations, and the authors say it is only barely statistically significant. They do not find the same decrease for workers older than 25.
That is a hiring signal for early-career candidates in exposed office, coding, and analysis roles. It is not evidence that those professions have collapsed, and it is US survey evidence, not a count of vacancies in Amman or Riyadh.
Expectations are running ahead of that employment result. In Anthropic Economic Index report: Cadences (Maxim Massenkoff, Eva Lyubich, Szymon Sacher, Zoe Hitzig, Shaoyi Zhang, Ryan Heller, and Peter McCrory, 26 June 2026), Anthropic reports a survey launched in April 2026 and linked to usage for about 9,700 respondents. The authors are explicit that this is not a sample of the general public. Computer and mathematical occupations were roughly 30% of respondents, against about 4% of US employment.
Within that group, close to 6 in 10 chose a higher band for the share of their work they expect AI to handle in 12 months than for today. Over a third expect AI to be able to do most or nearly all of their work tasks next year. More than a third said it was likely or very likely that job responsibilities would change significantly. Only 10% rated losing their own job as likely or very likely. They were more worried about other people than about themselves, and over one third said the probability of a junior colleague losing a job in the next year was over 60%. Respondents were also more concerned about job loss in lower-income countries.
Early-career respondents said AI could already do the highest share of their work, and they were the most concerned about job loss. People with at least 15 years of experience put that share roughly 10 percentage points lower than people in their first year. The report’s respondents often pointed to judgment, context, and managing people as the work they did not expect a model to copy. Large majorities still reported productivity gains: 86% in speed, 82% in scope, and 69% in quality. 68% said they were learning more with AI, and 57% said it had made their skills more valuable.
The global picture is transformation, with clerical work out in front
Anthropic’s numbers describe one product’s users. A broader cut comes from the International Labour Organization and Poland’s National Research Institute. Their 20 May 2025 release, One in four jobs at risk of being transformed by GenAI, says 25% of global employment falls in occupations potentially exposed to generative AI, and 34% in high-income countries. Clerical jobs face the highest exposure. In high-income countries, jobs in the highest exposure gradient make up 9.6% of women’s employment and 3.5% of men’s. The ILO is plain that these figures are potential exposure, not recorded job losses, and that “transformation, not replacement, is the most likely outcome,” because many tasks still need a person.
Janine Berg, a senior economist at the ILO, put the tone in one line: “It’s easy to get lost in the AI hype. What we need is clarity and context.” That is the right standard for a job search in Amman or a remote application aimed at a Gulf employer. Hype tells you to rebrand yourself as an “AI expert” overnight. Context tells you which tasks in your current target role are clerical and repeatable, and which still depend on local knowledge, a client relationship, or a decision someone has to own.
Employers, surveyed separately, are already budgeting for both training and some cuts. The World Economic Forum’s Future of Jobs Report 2025, published 7 January 2025, draws on more than 1,000 employers representing more than 14 million workers across 22 industry clusters and 55 economies. Among those employers, 86% expect AI and information processing to transform their business by 2030. They expect 39% of workers’ existing skill sets to be transformed or outdated over 2025 to 2030. Analytical thinking remains the top core skill, with seven in ten companies calling it essential. The fastest-growing skills they name are AI and big data, then networks and cybersecurity, then technological literacy, alongside creative thinking, resilience, and curiosity.
Skill gaps are the barrier 63% of those employers name. 85% plan to prioritize upskilling, 70% expect to hire people with new skills, and 40% plan to reduce staff whose skills become less relevant. Half plan to reorient the business in response to AI, two-thirds plan to hire for specific AI skills, and 40% anticipate reducing the workforce where AI can automate tasks. In absolute numbers, they expect the largest declines in clerical and secretarial work, including data entry clerks. In percentage terms, the fastest-growing roles they name include big data specialists, fintech engineers, AI and machine learning specialists, and software and application developers. Those are employer expectations, not a vacancy count for Abdali or Dubai Internet City.
What to do with this if you are applying now
Use the research as a filter for your next application, not as a personality test.
- Split the target job into tasks before you rewrite your CV. List the recurring tasks in the job ad: drafting, reporting, data cleanup, customer replies, scheduling, coding, analysis, client meetings, site visits, approvals. The Anthropic and ILO findings both point at the first cluster, especially routine writing, data entry, and clerical support. Meetings, local compliance, and decisions that need context sit further away. Your CV should show strength in the second cluster and competent use of tools on the first.
- Show judgment, not a tool logo. A line that says “proficient in ChatGPT” does not match what these studies describe. The June survey’s heavier users still reported learning, and the March report ties better outcomes to people who have practiced. On a CV or in an interview, describe one piece of work you directed, checked, and would sign. Example shape: “Drafted the first pass of a donor report with an AI tool, then corrected the figures and the Arabic and English wording before a manager reviewed it.” Replace that with your real example. Do not invent one.
- Keep analytical thinking as the headline skill. The World Economic Forum’s employer survey still ranks it above tool-specific skills, with AI and big data rising fastest beside it. For a remote or GCC application, that means a work sample with a clear question, a source, and a conclusion, not a wall of generated prose. If you write in Arabic and English, say so. Cross-checking both is a human task these papers do not show a model replacing.
- If you are early in your career, apply more widely than the most exposed junior title. The only soft spot in the Anthropic hiring evidence is ages 22 to 25 entering highly exposed occupations, and even that result is tentative and American. Do not read it as “juniors will not be hired in Jordan.” Do read it as a reason to pair an exposed skill, such as basic data cleanup or support writing, with proof that you can talk to a client, run a shift, or own a small project. Junior colleagues were exactly who the June respondents worried about.
- For remote roles, compete on a reviewed artifact. Remote hiring already asks you to show work. The Cadences report finds that work conversations most often produce documents and reports. Send a short sample you edited yourself. Mention what you rejected from the draft. That is the difference between augmentation and pasting.
- For Amman and GCC roles, lean on context a model does not sit inside. Employer, regulator, family business, ministry process, hospital shift, NGO donor rule. The labour-market paper’s uncovered work is full of physical and in-person jobs, and experienced Claude users in the June survey pointed to trust and managing people. If your target role is nursing, teaching, site supervision, sales that depends on a relationship, or operations on the ground, do not retrain as a developer because a headline said so. Add technological literacy where the World Economic Forum employers say it is rising. Do not throw away the part of the job that is still scarce.
What not to panic about
- A whole occupation disappearing this year. The ILO’s index is about potential exposure, and Anthropic’s US comparison has not found a systematic unemployment rise in the most exposed jobs since late 2022. Roles can still shrink at the margin, especially clerical ones. That is a reason to move tasks up the value chain, not a reason to freeze.
- Needing to become a programmer to stay employable. Programmers are highly exposed in Anthropic’s measure because coding is a large share of real model use. Employers in the World Economic Forum survey also expect software roles to grow. Both can be true: the tasks inside the job change, and the job title does not vanish. If you are not aiming at software, technological literacy and careful use of AI on your own documents matter more than a new degree.
- US or global percentages as if they were Jordan’s vacancy board. Claude’s user base is not the Jordanian workforce. The June survey over-represents technical and management jobs. ILO exposure is higher in high-income countries than globally. Use the pattern, which is clerical and routine cognitive tasks first, and ignore anyone who quotes these studies as a local unemployment forecast.
- Using AI with the review step removed. The same research that shows speed gains also shows that experienced users iterate, and that people trust context and judgment as the hard part. An application full of unreviewed generated text is a weaker signal than a shorter one you can defend.
The calm version of the next few years is not “nothing changes” and not “your career is over.” It is that a growing share of tasks inside office, writing, support, analysis, and software jobs can be drafted or sped up, while hiring data has not yet shown a broad unemployment shock. For a job seeker in Amman, a GCC city, or a remote search, the advantage goes to the person who can name those tasks, use a tool on the repeatable ones, and still own the judgment, the relationship, and the local detail.
Sources
- Maxim Massenkoff and Peter McCrory, Labor market impacts of AI: A new measure and early evidence, Anthropic, 5 March 2026.
- Maxim Massenkoff, Eva Lyubich, and Peter McCrory, with Ruth Appel and Ryan Heller, Anthropic Economic Index report: Learning curves, 24 March 2026.
- Maxim Massenkoff, Eva Lyubich, Szymon Sacher, Zoe Hitzig, Shaoyi Zhang, Ryan Heller, and Peter McCrory, Anthropic Economic Index report: Cadences, 26 June 2026.
- International Labour Organization and NASK, One in four jobs at risk of being transformed by GenAI, new ILO–NASK Global Index shows, 20 May 2025.
- World Economic Forum, The Future of Jobs Report 2025 (digest), published 7 January 2025.
Leave a Reply