Independently researched from primary research, official UAE guidance, and current reporting. Checked on August 2, 2026.
The job interview now has an AI problem on both sides of the call. Employers are introducing voice agents to ask the questions, while candidates are using generative AI to prepare answers, complete tests, and sometimes receive live assistance during remote interviews.
AI can standardize interview information, while candidate-side AI is forcing employers to redesign how genuine skill is verified.
UAE employers, recruiters, jobseekers, HR teams, and vendors supplying automated hiring systems should pay attention.
Use AI for structured first-round information collection while keeping final judgment, exceptions, and appeals under human control.
Accents, speech recognition, technical failures, recording, accessibility, data retention, and undisclosed candidate assistance can distort the signal.
More employers are likely to combine automated screening with verified work samples, human review, and at least one controlled assessment.
Employers should publish one AI interview policy; candidates should ask what is recorded, scored, retained, and reviewed by a person.
High-volume employers and candidates who perform better in structured, consistent interviews receive the clearest direct benefit.
The study covered entry-level customer-service applicants in the Philippines, not the entire UAE labor market.
A new field experiment involving 70,884 applications found that applicants interviewed by an AI voice agent were 12% more likely to receive an offer. Human recruiters still reviewed the interviews and made every hiring decision, which makes the result more interesting than a simple story about replacing recruiters.
The Robius layer is the division of labor. AI may be useful at collecting comparable information, but trust is moving in the opposite direction at the verification stage. The employer still needs proof that the candidate, not an invisible copilot, can perform the work.
What the 70,884-Applicant Experiment Found
The study was conducted with PSG Global Solutions, a recruitment-process outsourcing company and Teleperformance subsidiary. Applicants for entry-level customer-service roles in the Philippines were randomly assigned to a human interviewer, an AI voice interviewer, or a condition where they could choose between the two.
The AI and human interviewers followed the same topic guidelines. After the interview, applicants completed standardized language and analytical tests. Human recruiters then reviewed the interview and test information before deciding whether to make an offer.
Offer rates rose from 8.70% in the human-interviewer group to 9.73% in the AI-interviewer group. The researchers also reported an 18% higher likelihood of starting the job and an 18% higher likelihood of remaining employed for at least one month, with positive effects still visible at four months.
The productivity measures available for hired workers did not show a meaningful decline. The authors attribute the improvement to what they call controlled variance: more consistent topic coverage and structure, while the AI still adjusted follow-up questions to the individual applicant.
The Result Is Stronger Than a Chatbot Demo, but Narrower Than a Universal Rule
This was a randomized field experiment linked to real hiring and employment outcomes. That makes it stronger than a vendor case study or a survey asking managers whether they like AI.
But the setting matters. These were high-volume, entry-level customer-service roles in the Philippines, where English fluency, communication, analytical ability, and shift flexibility were important. The result does not prove that an AI interviewer improves hiring for engineers, teachers, doctors, executives, or every multilingual UAE workplace.
There were also clear failure points. About 5% of applicants ended the interview because they did not want to speak to an AI, and the AI encountered technical difficulties in 7% of cases. The system was rated as less natural, even though overall satisfaction was similar.
The paper also found that 78% of candidates chose the AI interviewer when given a choice. Yet those choosing AI had lower language and analytical scores on average, showing why preference alone cannot be treated as proof of quality or fairness.
AI Is Now on Both Sides of the Interview
At the same time, companies are becoming less confident that a polished remote performance proves individual ability. Current reporting describes technology employers restoring in-person stages because candidates can use AI during coding tests, video calls, and take-home exercises.
That tension is already visible in the UAE. Our AI skills and UAE salary guide explains why employers increasingly want evidence of practical AI ability. The harder question is whether a candidate is demonstrating responsible tool use or outsourcing the skill being tested.
The debate over whether AI removes or reshapes jobs also changes the interview itself. Employers are no longer only hiring for what someone can produce unaided. They are trying to measure judgment, verification, tool choice, and the ability to recognize when AI is wrong.
The UAE Hiring Question Is Not AI or No AI
The Federal Authority for Government Human Resources has published guidance on evaluating AI used in talent assessment, and its newer Future Talents platform uses AI across parts of the talent journey. That does not create one binding rule for every private employer, but it shows that AI-mediated assessment is no longer an abstract issue in the UAE.
The practical question is which stage may be automated and which rights stay human. An AI can ask standardized questions, schedule candidates, transcribe answers, and organize evidence. A person should remain accountable for the final decision, unusual cases, accessibility adjustments, and challenges to an inaccurate score.
The recent 80,000-tech-jobs update showed why hiring claims need careful attribution. The same discipline belongs inside recruitment systems: record what the AI contributed, what the recruiter decided, and which evidence changed the outcome.
One AI Policy Should Apply to Both Sides
Many employers write a strict candidate rule while giving their own screening tools broad freedom. That is difficult to defend. A credible policy should explain both how the employer uses AI and what candidate use is allowed.
| Interview stage | Reasonable AI use | Control that keeps it credible |
|---|---|---|
| Application and scheduling | Summarize applications, answer process questions, arrange time slots | Human-accessible contact route and no automatic rejection from one weak signal |
| First-round interview | Ask structured questions, transcribe answers, organize comparable evidence | Disclosure, consent, accessibility option, audit trail, and human review |
| Skills assessment | Allow approved tools where the real job uses them | State permitted tools, require explanation, and verify authorship of the work |
| Final decision | Surface evidence and inconsistencies | Named human decision-maker, documented reasons, and a route to correct errors |
What Employers Should Publish Before the Interview
Candidates should know whether the interaction is recorded, whether voice or language features are scored, how long the material is retained, and whether their answers train any model. They should also know how to request a person or an adjustment when the automated route does not work for them.
Employers should say whether candidate AI use is prohibited, permitted for preparation, or allowed during particular tasks. A vague instruction such as “use your judgment” creates an avoidable dispute when the company later treats the same behavior as dishonesty.
Fraud controls matter too. The LinkedIn recruiter scam guide shows why candidates must independently verify the employer and interview link before sharing recordings, identity documents, or personal information with an automated system.
What Candidates Should Ask
Ask what the AI is doing, not merely whether AI is present. A transcription tool, an interviewer that chooses follow-ups, and a model that generates a suitability score create very different risks.
Keep a copy of the job description, the invitation, the stated AI-use rules, and any consent notice. During a tool-permitted assessment, be ready to explain your prompts, corrections, trade-offs, and final judgment. The ability to verify AI output is increasingly more valuable than pretending the tool was not used.
The UAE is already moving toward more agentic government services, so candidates and employers will encounter automated decisions in more settings. The durable skill is not avoiding automation. It is knowing where human accountability must remain.
The Bottom Line
The field experiment provides credible evidence that AI can improve one part of hiring: structured information collection at scale. It does not prove that every AI interview is fair, accurate, accessible, or suitable for every role.
The next hiring model is likely to be hybrid. AI gathers and organizes evidence, candidates use approved tools where the job demands them, and employers verify authorship, judgment, and real ability through controlled work and human review.
The best policy is symmetrical. Tell candidates what your AI is allowed to do, tell them what their AI is allowed to do, and keep a person responsible for the decision.
Sources
- Jabarian and Henkel / arXiv: Primary field experiment covering 70,884 applications, interview design, offer rates, job starts, retention, applicant responses, and study limitations. https://arxiv.org/pdf/2607.28222
- Business Insider: July 31 reporting on technology companies restoring in-person interview stages because of candidate use of AI during remote assessments. https://www.businessinsider.com/ai-job-interviews-in-person-nick-bloom-rto-2026-7
- FAHR: Official UAE guidance on the use and evaluation of artificial intelligence in talent assessment. https://www.fahr.gov.ae/en/blogs/fahr-the-use-of-artificial-intelligence-in-talent-assessment/
- FAHR: Official announcement of the AI-enhanced Future Talents platform used across parts of the federal talent-management journey. https://www.fahr.gov.ae/en/news/fahr-launches-the-ai-enhanced-future-talents-platform/
- Gulf News: UAE-focused reporting on candidate use of AI in remote interviews and the resulting challenge for employers. https://gulfnews.com/business/markets/ai-makes-remote-interviews-easier-for-job-seekers-but-harder-for-employers-1.500609110
Robius.news — Dubai, UAE — 2026 | Built to be first. Built to be trusted.
