From ChatGPT to L5: Artificial intelligence competency enters job requirements with a graded level
Listen to this article
Read by Anchor
In a job posting for an Account Executive in Amman, between a line that asks for two years of experience in the telecommunications sector and a line that asks for fluency in Arabic and English, a third requirement appears in an odd format on regional ads: “Artificial intelligence competency, minimum L2 to L3”. No degree is required, nor any programming skill. Just a level, written in the same way language proficiency levels are listed on a résumé.
The posting comes from Unifonic, the Saudi cloud-telecommunications company that now offers its platform as an AI-powered customer-experience solution. Reading its open job board on 12 September 2026 shows that the line is not a one-off typo: it recurs in a Strategic Account Manager posting in Riyadh at the same level, L2 to L3, and appears in a Software Quality Engineering Manager posting in Cairo at a much higher level, L5.
The two levels demand completely different things.In the commercial role, the posting defines the minimum as a practical, basic understanding of AI tools with the ability to apply them in sales tasks, and it names the expected use cases as prospect research, tailoring outreach messages, extracting insights from a CRM system, and tracking the opportunity pipeline. The tools listed are the ones people actually use: ChatGPT, Microsoft Copilot and Gemini. In the quality role, the L5 level corresponds in the posting to an explicit requirement to understand agent-based feature testing: hallucination detection, grounding, guardrails, fallback behavior and resistance to prompt injection.
That is a wide gap under a single heading. A reader who sees the phrase “AI competency” and assumes it means using ChatGPT in their work would be correct for one position and completely wrong for another on the same company’s board.
The Riyadh posting contains a detail worth pausing over. The role manages government accounts, financial institutions, telecom companies and senior clients in organized sectors, and it simultaneously asks its holder for a practical level of AI-tool use to analyze account growth. In other words, the required competency has migrated from engineering teams to become a condition for anyone sitting in front of a government client. By contrast, the higher level, L5, is attached to a role that demands ten years in development or testing, seven years in quality roles, and three years in test-team management, whereas the L2 to L3 level is attached to a role in Amman that requires only two years of experience. The grade is therefore tied both to the job and to the holder’s position on the responsibility ladder.
The measurement aspect matters more than the names of the tools.When a company turns AI competency into a numbered level, it does the same thing it already does with language: it moves the skill from a claim on a résumé to something you can be asked about in an interview and have your answer measured. The phrase “AI-savvy” does not correspond to any level. The answer that matches a level is a specific task you completed with a specific tool, and what changed in its outcome.
The volume of evidence deserves precise wording. This is a single company’s job board, with the line appearing in three open postings today and in a fourth posting that has closed. It is not a regional policy, nor a standard adopted by the Gulf labour market, nor an indication that other firms will follow suit. It is, however, a written, public and verifiable practice at a Saudi company that hires in three countries: Saudi Arabia, Jordan and Egypt. The question worth tracking in the coming weeks is whether the line will appear in other firms’ postings in the same sector, and whether any of them will publish a clear definition for each level instead of just the number. A undefined requirement is hard for anyone to prepare for.
Your practical step is smaller than overhauling your career path. Open your résumé to the AI line if it contains one, and replace it with a task: the tool, the work you performed with it, and a measurable result. Then reread the job posting you are targeting, and pinpoint exactly what it asks of you: to use AI, or to evaluate its outputs. The two demands lead to two different interviews, and preparing for them is not the same.