Yes, when AI drafts from the hiring manager's intake notes rather than from a job title, separates real requirements from wishes, and a recruiter checks every requirement against the actual job before posting. Fairness comes mostly from honest requirements, not clever wording. The Arabic version is written for Arabic readers, not translated line by line.
#Key takeaways
- A job description drafted from a title alone copies the market's inflated requirements; draft from the intake conversation instead.
- The biggest fairness gain is cutting requirements the job does not need, such as degrees or years of experience nobody checks.
- Research suggests wording tweaks alone change who applies only slightly, so do not stop at a gender-decoder tool.
- Arabic and English versions are written separately from the same requirements, then checked by a fluent reader.
- For a few roles, a chat tool and the checklist below are enough; scale and many client formats are what justify more.
#Why are most job descriptions inaccurate?
A hiring manager asks for a role on Monday, the recruiter copies last year's ad or a competitor's, edits the title and posts it on LinkedIn, Bayt and the careers page. The requirements list grows with every copy. Applicants either self-select out or apply to everything, and screening gets harder as a result.
AI makes this faster and worse if it is used the same way: ask a model for 'a job description for an operations manager' and it returns the average of every such ad it has seen, inflated requirements included.
#Does gendered wording really change who applies?
Less than many tools suggest. A study of hundreds of thousands of postings, reported by MIT Sloan, april 2023, found that more neutral wording changed whether men and women applied only very slightly. Wording still matters for clarity and tone, but it is not where most of the fairness is won.
The larger lever is what the ad asks for. A requirement the job does not need excludes good candidates whatever the wording around it, and it also pushes screening, human or AI, to filter on the wrong thing.
#What is a practical playbook for ai-drafted job descriptions?
- run a structured intake: ask the hiring manager what the person will do in practice, what success looks like and which skills are needed from the first day.
- split the list: AI sorts the intake notes into must-haves, trainable skills and wishes, and the hiring manager confirms the split.
- draft from the notes: AI writes the description using only the intake notes and your house style, never a generic template.
- challenge each requirement: for every must-have, the recruiter asks how it will be checked; if nobody will check it, it goes.
- write both languages: Arabic and English versions are drafted from the same requirements, then read by a fluent reviewer.
- save the intake: the confirmed requirements become the screening criteria, so the ad and the shortlist match.
The last step links the ad to selection. Lewis Silkin, april 2024 notes that the EU AI Act treats AI used in recruitment and selection decisions as high-risk, and lists drafting job descriptions among the places generative AI already appears. Criteria written once and used for both is the simplest record of a fair process.
#What should stay with people?
The hiring manager owns the requirements, and the recruiter owns the challenge. AI can question a requirement, for example by asking whether a degree is needed for a role that the current holder does without one, but people make the call and the record says who.
#When is a chat tool enough?
If you post only a few roles at a time, a ChatGPT, Claude or Copilot plan your company approved, the six steps above as a checklist and a fluent Arabic reviewer cover it. No project is needed, and saying so is the honest answer.
More is justified when an agency writes for many clients with different formats, when hundreds of roles share a competency framework, or when ads must flow from the ATS to several job boards in both languages.
#Where does Stennir fit?
For most teams the work starts with AI consulting: Stennir's AI Opportunity Audit looks at your whole hiring flow and ranks five workflows, so you know whether job descriptions are worth building for or whether a checklist will do. If a build is worth it, AI implementation connects intake, drafting and posting to your ATS.
AI training teaches recruiters and hiring managers to run the intake and challenge requirements, in Arabic and English.
We are
Stennir helps talent teams decide where AI belongs in hiring and builds the parts worth building, with people owning every requirement.
We aren't
Stennir is not a job-ad writing app, and it does not decide what a role requires.
#What should a talent lead do next?
Take your five most-posted roles and, for each must-have, write down how it is checked at screening. Cut the ones nobody checks. Then read AI for operations and AI training for HR teams, and take the free readiness assessment.
#Common questions
Can AI write the Arabic version of a job description?
It can draft one, and the draft should be written from the agreed requirements rather than translated from English. Job titles, gender forms and formal register need a fluent reviewer, especially for Gulf audiences. Treat the Arabic version as a separate document with its own check, not as a copy.
Does an ai-written job description create legal risk?
The risk sits mostly in what the ad asks for and how it is used in selection, not in who typed it. Requirements linked to age, gender, nationality or family status need care in any jurisdiction. Ask your employment counsel to review your templates once, then reuse them.
What do you need from us to start?
A talent or HR lead, your current templates, a few hiring managers willing to run a structured intake, and the job boards you post to. For an agency, Stennir also needs your clients' format rules. From that, the audit shows whether a checklist or a build is the right answer.
