For many, the number of recent aviation accidents making headlines already this year has been fair cause for concern. In January, a Bombardier CRJ700 operated by American Airlines collided with a U.S. Army Sikorsky UH-60 Black Hawk helicopter whilst the Bombardier aircraft was attempting to land at Ronald Reagan Washington National Airport and crashed into the Potomac River. Sadly, there were no survivors of the 64 people on board the CRJ700 nor the three occupants of the Sikorsky UH-60 helicopter.
February brought us similarly worrying headlines, with a Bombardier CRJ900 operated by Endeavor Air crashing and overturning during landing at Toronto Pearson International Airport. Thankfully and miraculously, in this latest incident, all 80 people on board survived.
Despite these recent incidents, it remains true that accidents like these are extremely rare. In the U.S., for example, publicly available NTSB data shows a clear downward trajectory in the number of fatal aviation accidents during the years leading up to 2025.
(Graph courtesy of ESi)
This highlights the continuing year-on-year trend of safety advancements across all forms of private and commercial aviation. That is particularly evident when you also consider that the overall numbers of flights and passengers each year have also significantly increased and are expected keep on growing. According to IATA, the number of industry-wide air passenger journeys are expected to more than double from the 2019 level to reach 7.8 billion by 2040.
Nevertheless, despite all the advancements being made, major accidents still occur and claims arising from them will continue to follow.
In more complex cases, particularly those that have entered the litigation process where the facts of the incident are either disputed or are yet to be determined, insurers may decide to appoint expert accident reconstruction and investigation firms to help support these claims. These firms can add huge value to the claims process by providing insurers with technical expertise and objective analysis critical for determining the causes and circumstances surrounding an accident. In litigation, their analysis can accompany the findings of official accident investigation reports, such as those produced by the NTSB or AAIB. However, whilst the rules under ICAO Annex 13 dictate that an official accident investigation report is not focussed on assigning blame or determining legal liability, an independent accident reconstruction and investigation firm will often be able to objectively analyse a case and apportion liability in support of legal cases for their clients.
The supporting roles of accident reconstruction firms can empower insurers and their clients to make more informed and beneficial long-term decisions. This is a great attribute to have on your side, particularly when the stakes can be so high.
Insurers and Accident Reconstruction Firms
Global Aerospace and other insurers have a history of working alongside one such valued accident reconstruction firm called ESi. Formed in 1987, they are an engineering and scientific investigation and analysis specialist who have handled and overseen a broad range of accident reconstruction projects across many areas, including in aviation.
Global and ESi have worked closely to support our common clients for a host of litigated aviation cases across multiple jurisdictions around the world. This has ranged from complex aircraft product and component cases in the U.S. to a remarkable flight path and weapon trajectory analysis and reconstruction animation produced as part of an expert submission into court supporting our client’s case following the downing of Ukraine International Airlines Flight 752 on 08 January 2020 in Iran.
Their expertise can also add value and contribute in other ways, including developing standards and procedures that make the aviation industry safer, providing expert testimony in aviation-related litigation and sharing their knowledge and findings with the wider industry.
Understanding Accident Reconstruction Analysis—Past and Present
In the past, accident reconstruction firms had two main sources of technical information: the wreckage and radar. And that is if they were lucky.
In a typical accident today, there is an abundance of information available. Firms like ESi have the ability and expertise to utilise vast arrays of different data sets to investigate all aspects of an accident. Available data can include on board avionics, flight data, cockpit recordings, environmental factors and GPS data to name but a few.
Nowadays, it is not uncommon to have videos of an incident, either from CCTV footage or videos taken by civilian passers-by from all different angles, which may then be uploaded onto social media sites for all to see. When combined with the abundance of other raw data they have available at their disposal, ESi can turn photographs and videos into usable configuration via a process called photogrammetry.
This specialised technique maps and creates 3D computerised models of objects, structures and landscapes. ESi combine and use this complex density of data to provide insurers with an independent and largely objective analysis of the circumstances surrounding a particular accident. This can then usually be presented visually in the form of reconstruction animations.
ESi’s vast experience in this area is often utilised in litigation to support defence arguments during the claims process, as well as providing expert witness testimony. Identifying and helping to apportion responsibility from these accidents is a vital tool that these reconstructions can offer to insurers as they can conduct and create independent analyses of aircraft accidents.
AI and the Future of Accident Reconstruction
We are all now armchair experts on the topic of AI. In 2025, everybody has heard of it. Everyone is talking about it. But do we really know what it is, and do we know its true potential?
Whilst AI has seemingly endless possibilities to assist in all facets of life, the primary focus of this article is how it might advance accident reconstructions in the future. Could it soon be possible to perfectly recreate an accident using AI? Today, even the most basic AI platforms can identify anomalies quickly from complex data sets, resulting in faster conclusions and recommendations. But how could this benefit accident reconstruction scenarios and is it reliable? Here are some potential benefits of AI in accident reconstruction:
Time and cost reduction
AI programmes can process enormous volumes of data to reach conclusions at seemingly instantaneous speeds, a pace far quicker than their human counterparts could even dream of.
Processing data quickly would carry with it the additional benefit of being more time- and cost-effective for insurers. Whilst current accident and reconstruction projects typically cost in the region of USD50,000, AI provides an opportunity to drive that cost down.
Current reconstruction projects and animations also typically take weeks or months to be produced. Theoretically, an AI tool could reduce this process to a matter of days, hours or perhaps even minutes. At reduced time and cost, these reconstructions and animations would become more economically viable and accessible, even for lower-value claims, possibly making them more prevalent amongst litigation submissions or even at the pre-litigation stage if circumstances allow. However, looking solely through the lens of costs alone is perhaps too narrow a view. Holding the key to AI’s future will be the quality, accuracy and trust in its output.
Enhanced trend analysis
AI can also use complex data to spot trends that a person might miss. This area is of great interest as machine learning algorithms can detect anomalies that traditional methods may have overlooked.
As described in detail in Nanduri & Sherry’s 2016 paper on anomaly detection in aircraft data, traditional algorithms typically look for deviations from normal values, whereas machine learning can identify unusual patterns in monitored signals to detect any anomalies or issues, like finding a metaphorical needle in a haystack.
This capability could prove extremely valuable in accident analysis where AI may be able to directly compare flight data recorder data with cockpit voice recorder recordings and other acoustic signal data to provide clearer insights into the chronology of events leading up to an accident.
As it develops, AI will also likely find and introduce new ways of analysing data, providing deeper insights that can be visualised.
More detailed visualisations
Stakeholders and fact finders have long benefited from examining the internal workings of aircraft components—for instance, understanding how a complex hydraulic system functions or how hot and cold airflow through a turbocharger affects its lubricating oil properties. Detailed visualisations will continue to hold significant value. AI will enhance our efficiency in creating them and generate more nuanced and insightful understandings to enhance these visualisations and improve safety-related learnings post-accident.
(Image courtesy of ESi)
Removal of confirmation bias
As humans, we can suffer from subconscious confirmation bias. That is when information that seems to support a preconceived conclusion is used as the primary focus during an investigation. Investigators are trained to mitigate confirmation bias; however, to remove it entirely from their conclusions is nigh on impossible. AI does not suffer from the same ’human flaw.’ It can review all available data without holding preconceived ideas of what might have caused an accident. Therefore, it has the potential to remove all elements of bias from its conclusions, thus offering more accurate results.
Challenges AI Will Face in Accident Reconstruction
Although AI offers many opportunities, its use may also present many challenges, including:
Trust
The real question and challenge that remains to be answered lies in whether the aviation industry is ready to use AI in sensitive cases such as an air accident. Not only will public trust need to become more established with the use of such tools, but equally as vital will be the need to establish trust and support from within regulatory bodies and legal authorities themselves for AI’s conclusions to be made admissible in litigation. Will they deem these conclusions to be safe and reliable?
Artificial hallucinations
Current AI models are known to frequently embed and produce random falsehoods within its generated content. They can contain false or misleading information presented to the reader or viewer as fact. This issue has already caught some out, including a now infamous personal injury lawsuit in New York brought against Avianca airline in which the claimant lawyer used ChatGPT to write their argument brief. Avianca’s lawyers and the Judge judiciously discovered that the decisions and legal precedents summarised in this brief did not exist. The reason for this? AI had fabricated them all. Whilst this falsehood may carry with it a somewhat humorous element, the real-life connotations of this show precisely why there remain significant hurdles for current AI models to overcome before they can be trusted enough to be used in litigation or accident reconstructions.
Potential lack of repeatability
Whilst AI does not suffer from confirmation bias, it requires high-quality data inputs for its conclusions to be sufficiently accurate and reliable. By virtue of its nature, AI continually learns from its inputs. Reasonably, a court’s criteria for the admissibility of AI accident analysis into legal proceedings may rely on the repeatability of results.
Given that AI tools constantly learn from new data inputs, the answer to a question posed today may differ from the answer it gives tomorrow. Whilst these answers may not be wildly different from one another, the lack of pure repeatability may provide obstacles to regulatory authorities and legal court systems affirming the use of AI in accident reconstructions and analysis.
Establishing trust, a regulatory structure and a standard for incorporating AI into investigations and litigation are, therefore, crucial.
Key points and AI’s Potential Beyond Accident Reconstruction
AI has the potential to become a hugely valuable partner in many aspects of aviation accident reconstruction and analysis, although this will not be without its challenges and an approach of cautious integration into established processes would be advisable. Still, AI holds significant promise for improving the accessibility, promptness, efficiency and accuracy of aviation accident reconstruction and analysis investigations, ultimately leading to more reliable conclusions and providing enhanced support for insurers and their clients during the claims process. To reach its full potential, the focus of AI development must be on building trust among all stakeholders, which will only be possible if its outputs become more consistent, repeatable and with the risk of artificial hallucinations entirely removed.
Some experts go further to wonder if AI can be a tool for accident prevention. It is no secret that AI’s potential is vast. Consequently, its use within an aviation context will likely not be limited to accident reconstruction and analysis. This article has focused on its potential post-accident supporting role, with its ability to deliver insights buried within a mountain of information. However, it can be predicted that in the future, it will also play a pivotal role in identifying and recommending safety measures that mitigate or eliminate root causes of aviation accidents altogether.
Contributors:
ESi – https://www.engsys.com/areas-of-expertise/transportation/aviation
Special thanks to Timothy P. Jung , Ph.D., P.E. and Charles Fox, Ph.D. for their contributions to this article.
References:
Nanduri, A. & Sherry, L., Anomaly Detection In Aircraft Data Using Recurrent Neural Networks (RNN), Center for Air Transportation Systems Research (CATSR) at George mason University (GMU), Fairfax, Virginia, https://www.researchgate.net/publication/303885965_Anomaly_detection_in_aircraft_data_using_Recurrent_Neural_Networks_RNN
https://condonlaw.com/2023/06/in-landmark-case-involving-attorneys-who-submitted-fake-court-decisions-generated-by-chatgpt-federal-judge-sanctions-the-attorneys-and-dismisses-action-as-time-barred-under-the-montreal-convention/
James Hopkins, Senior Claims Adjuster
Global Aerospace
James Hopkins joined Global Aerospace in 2015. He was promoted to Claims Adjuster in April 2018 and then to his current position of Senior Claims Adjuster in 2023. James holds the CII Certificate in Insurance.
About Global Aerospace
Global Aerospace has a century of experience and powerful passion for providing aviation insurance solutions that protect industry stakeholders and empower the industry to thrive. With financial stability from a pool of the world’s foremost capital, we leverage innovative ideas, advanced technology and a powerful synergy among diverse team members to underwrite and process claims for the many risks our clients face. Headquartered in the UK, we have offices in Canada, France, Germany and throughout the United States. Learn more at https://www.global-aero.com/
Global Aerospace Media Contact
Suzanne Keneally
Vice President, Group Head of Communications
+1 973-490-8588
Source: Prodigy.press
Release ID: 1389040
