Etihad: using AI to read the airline's own safety reports
Etihad has a named AI safety-management application built with MBZUAI, alongside a broad data and analytics platform. Earlier touchless kiosk experiments have a different status and should not be mistaken for ongoing service.
The operating picture
Where text analysis enters safety management
Etihad Airways A6-APD Airbus A380-861 Sydney Kingsford Smith International Airport (SYD YSSY) (51404833592). Photograph: Mitchul Hope / CC BY-SA 2.0 (cropped to fit). Illustrates the airline, not an AI deployment.
Airline safety records contain large volumes of narrative reports. Finding patterns in that text is difficult when events are described in different words by different crews or departments.
Etihad says its partnership with Mohamed bin Zayed University of Artificial Intelligence introduced natural-language analysis into safety management in 2023. Separately, Databricks describes a broad Etihad lakehouse and natural-language analytics service, including a catering analysis example. Both need tests of analytical reliability, not merely counts of records ingested.
Selected implementations
The work behind the evidence
01 / Deployed
Finding a safety signal in an unstructured report
The operational problem
Safety teams need to identify recurring hazards while distinguishing meaningful patterns from accidental similarities in reports.
What the evidence reports
Etihad announced in December 2023 that it was using BERT, Azure and text analytics with MBZUAI to examine safety, maintenance and training data.
Operational significance
The tool has a specific analytical purpose within an accountable safety-management process.
Evidence still needed
No independent false-alert evaluation, hazard-detection improvement or completed safety outcome is documented.
Making the data estate usable through natural language
The operational problem
Teams can struggle to obtain consistent answers when operational and commercial data sit across many systems.
What the evidence reports
A Databricks customer account says Etihad consolidated data from more than 100 systems and provided Genie natural-language analytics. It describes a catering analysis reduced from hours to roughly two minutes.
Operational significance
The catering example is a defined task, although it is a provider account and not an airline-wide measure of AI performance.
Evidence still needed
The source lacks precise dates and independent verification of the time comparison.
A source can verify an announcement without verifying its projected outcome.
The wider register
Other tracked activity
Other records are shown separately, including non-AI technology and programmes at different stages of deployment. A register entry does not establish a measured AI benefit.
Etihad voice-activated and touchless kiosk trials with Elenium · Pilot
Etihad Airways and Elenium Automation showed a voice-activated self-service kiosk concept in March 2019; the airline's release of 20 March 2019 said it used AI and computer vision.
A year later the two went further. Etihad said on 6 April 2020 that it would trial touchless kiosks with voice control and health screening at Abu Dhabi from the end of April into May, to help identify travellers who might be medically at risk.
A source can verify an announcement without verifying its projected outcome.
Editorial assessment
Etihad's safety application is a substantial airline use case because the human team has a clear responsibility for acting on the finding.
It also raises a demanding validation question: which safety signals were discovered earlier, and how did analysts rule out misleading patterns?
Live data from the Aerostack register. Status reflects published evidence only. See Our approach for how records are graded.
Aerostack register · 3 tracked deployments
Aerostack tracks AI deployments across airports and airlines, with the published evidence behind each record. Photography is real aviation imagery. Record thumbnails may show the named airport or airline; other photographs do not.
_(51404833592).jpg)