The complex passenger flows, demanding identification challenges and pressure-filled security conditions that are inherent in border control make the arena appear an interesting prospect for artificial intelligence-based solutions that can revolutionise critical processes.
Machine-learning technologies being developed could identify risk patterns at speeds way beyond humans’ capacity, and when tied with the powerful security offered by biometrics, AI’s potential to disrupt the world of borders expands even further.
What then is the future for the border guard? Will their decades of document security expertise and nuanced instincts be made rendered obsolete by a ‘border bot’?
Planet Biometrics caught up with Cyrille Bataller, Artificial Intelligence lead at Accenture, to learn about his team’s AI initiatives for borders and about his views on the importance of a “people-first” approach.
What kinds of role can AI play in border management?
Artificial Intelligence is a collection of technologies, powered by machine learning, big data and cloud, which enable IT systems to sense, comprehend, act and learn. AI enables machines to mimic human capabilities such as reading documents, recognizing people, understanding speech, answering questions in natural language… It encompasses various technologies such as machine learning, natural language processing, computer vision, and enables capabilities such as biometrics, gender and age detection, emotion recognition, language translation etc.
AI has existed in concept and academia for decades, but has really taken off since the mid-2000s thanks to the rise of big data and cloud, which enables machine learning at scale. AI enables automation and augmentation: automation of the most repetitive, well defined activities, thus freeing up people to focus on higher value add activities, and augmentation of people in these higher value add activities, helping them reach the right decision or outcome faster and be more productive. Future organizations will rely on teams of people and robots collaborating effectively to achieve more than either could on their own.
In border management, we’re already seeing AI in action, notably with self-service passport control gates that perform a complex, repetitive activity to examine each and every passenger at the border, under supervision from border guards, thus improving both security and facilitation at the border, enabling border guards to process larger volumes of passengers faster, while focusing on higher-level security responsibilities. It won’t be long, I think, until we also see swarms of automated drones patrolling land and sea borders using computer vision and other AI technologies to detect intrusions, and notably, hopefully, detect overcrowded migrant ships before tragedies happen. We will likely also see virtual agents handle customer queries on visas and border control procedures, for instance.
Another area in border management where AI will play a larger role is Risk Assessment. Some countries already utilize advanced passenger information (API) and passenger name records (PNR) to perform automated risk assessment and more will as terrorist activities do not abate. Similarly, knowing who a person is and what risk they present will also enable segmentation of travelers to expedite those with acceptable (low) risk and focus on the others.
Can you describe the new biometric and artificial intelligence (AI) technologies you are developing at Accenture? Tell me about the BME recently developed at your Labs which won the internal Innovation Award.
It’s very exciting to be recognized for creating innovative IP and assets that helps take Accenture into new fields. We have a portfolio of over 150 issued patents and pending patent applications generated in the area of biometrics.
The award-winning Biometric Matching Engine (BME) uses software to compare a variety of biometric identifiers, such as face, fingerprints, iris or voice, against large volumes of reference identity data. Based on this information, the software automatically creates a unique identification strategy to optimize the speed and effectiveness of database search for a matching record. Part of Accenture’s Unique Identity Service Platform, the Biometric Matching Engine can quickly produce rich analytical insights around individual identities across many different biometric modalities while improving user experiences, preventing fraud and streamlining interactions.
We started this work in the mid-2000s (at the Accenture Labs in Sophia Antipolis, France) right after the decision was made to introduce biometrics within travel documents like passports—a consequence of the 9/11 terrorist attacks. This is when we began looking at innovative ways to help clients improve security and also to facilitate border control.
Accenture’s unique identity offerings, primarily implemented in the security space, have been successfully used with the US-VISIT program, the United Nations High Commission for Refugees, India’s Aadhaar program and the European Commission’s biometric-enabled visa program.
Scaling biometric matching solutions has traditionally been at the expense of both accuracy as well as performance. Dynamic matching strategies allows scaling beyond what is otherwise possible without sacrificing accuracy or performance.
What are the kind of manual processing tasks (at borders) could be taken over by AI / robots?
Several manual inspection processes occur at multiple points from ‘curb to cockpit’ which may be enhanced through the use of AI technologies. Enhancement in the form of: (1) methods to validate breeder documents and identity claims – a la ICAO’s Traveler Identification Programme (TRIP), (2) allowing officers to focus on unknown or high-risk travelers by Risk Analytics of API, PNR, and potentially Social Media, (3) using API/PNR to access identity claims then applying passive biometric verification to validate the claims.
Extending the workforce with digital co-workers, while keeping a people first mentality. Combining existing biometric matching technologies with the latest in AI and machine learning capabilities allows for what Accenture calls ‘Emergent Identity’, whereby the more often a user or passenger interacts with the system the more the system learns about them and the faster identification becomes. This assists airport operators and border agencies to increase security and identification accuracy, while also speeding-up passenger flow.
Can you explain how AI or machine learning could specifically speed-up passenger flow (at airports / borders) while at the same time increasing security?
There is no doubt that the integration of new AI technologies into existing border and identity management systems will lead to a step-change in how passengers and citizens experience airports in the future – shortening waiting times, while helping to keep travelers, airports and national borders safe.
As I said above, some countries already utilize advanced passenger information (API) and passenger name records (PNR) to perform automated risk assessment and more will as terrorist activities do not abate. Similarly, knowing who a person is and what risk they present will also enable segmentation of travelers to expedite those with acceptable (low) risk and focus on the others.
Could these AI concepts be applied to other industries – what industries will AI impact most in the coming years?
AI has potential to be as disruptive as Digital is – impacting many different industries – many predict a substantial impact for what some call the second machine age, the third era of computing, the fourth industrial revolution (WEF). For instance, self-driving cars will drastically change taxis, logistics (truck drivers), but also other industries such as insurance, and it will change how our cities look (fewer car parks, less traffic) and our lifestyles as well.
Interactions with companies will also likely change, from cumbersome, frustrating service desk experiences to engagement anytime anywhere through any channel, 24x7, interacting with virtual agents AI has applicability to any industry that needs to process large numbers of customer requests which are process heavy. Industries such as financial service, travel and public service are most likely to adopt ‘cognitive agents’ to help manage service requests. For example, recently in the UK, a City Council announced it is to deploy an AI assistant named Amelia to help process citizen service requests.