Executive summary

• Author: Cepilovs, Aleksandrs
• Pages: 5-6

AI IN THE OPERATIONAL MANAGEMENT OF LARGE-SCALE IT SYSTEMS  5

Executive summary

Over the last few years, the debate a round AI in the EU has picked up pa ce. In view of the implications of AI

across the economy and society, and also the benefits that can be reaped from the widespread adoption of AI

across all sectors of the economy, including the public sector, the EU has initiated the development of a

European approach to AI. This coordinated EU approa ch to AI aims to put the EU forward as the leader in

technology. It also aims to balance the advancement of technology and the need to protect and promote the

public interest. To ensure this balance, the development of trustworthy and secure AI in the EU should be

grounded in the common European values, such as respect of fundamental rights. In order to achieve these

objectives, the Commission has launched a number of initiatives, including additional funding for research and

innovation in AI, a coordinated plan on AI aimed at improving cooperation on AI in the EU, as well as the high-

level expert group on AI focusing on the development of recommendations on AI-related ethical, legal and

societal issues.

It is with this in mind that eu-LISA has been developing its approach to AI and the specific role the Agency can

play in advancing the adoption of AI in the EU. First and foremost, the Agency can support the development

and adoption of AI in the do main of borders, migration and security by, for example, supporting the necessary

computational infrastructure for the development and testing of AI tools for the key sta keholders. Second, the

Agency is well placed to benefit from the deployment of AI solutions in the context of the operational

management of large-scale IT systems, in particular focusing on the improvement of performance, stability and

the overall quality of the service it provides to stakeholders.

With the entry into force of the regulation establishing the Entry/Exit System (EES), eu-LISA has been mandated

to develop the new system, which incorporates a component for automated biometric matching, which will rely

on machine l earning techniques for biometric matching. The scope for application of AI is not only limited to

that, but includes a wide range of applications for AI which do not have an immediate effect on individuals. This

report therefore largely focuses on applications where ethical and legal considerations are either not relevant

or secondary; namely on those applications where the implementation of solutions relying on AI or ML

technologies can have an immediate effect on the performance of eu-LISA as an IT service provider.

First, significant performance improvements can be attained by deploying AI or M L solutions in the context

of IT infrastructure and service management (e.g. in application of performance monitoring, IT infrastructure

and network monit oring and diagnostics, IT event analytics or IT service desk operations), in particular by

increasing the availability of IT and network infrastructure, while at the same time reducing the workload of the

staff responsible for it (e.g. by taking over some of the repetitive routine tasks). Such efficiency gains pave the

way for devoting additional effort to further improvement of operations and services.

Second, conversational agents, virtual assistants and chatbots are another important class of AI tools in the

context of eu-LISA core business. Chatbots can be effectively deployed in a number of areas, including providing

training for future users o f the EES, providing support to the users of ETIAS, as well as integrating chatbots as

the first contact point for IT service desks. Use of chatbots or virtual assistants can improve the overall quality

of services delivered by eu-LISA, speed up the adoption of new systems by users and improve the quality of

information submitted by users using online systems.

Protection of IT systems from cyber threats is another domain where AI and ML can be put to effective use.

At the mos t basic level, protecting digital infrastructure relies on cyber threat intelligence in order to identify

vulnerabilities, as well as new attacks. Therefore, introducing automation to cyber threat intelligence using

machine learning is an important first step. The second level of application of AI in cyber security is in network

analysis and intrusion detection. And the next level is the creation of autonomous cyber security systems

capable of threat detection and response without the need for immediate human intervention.

Finally, machine learning can be effectively used to optimise the energy performance o f data centre