TU Darmstadt / ULB / TUbiblio

Technology Assessment of Dual-Use ICTs - How to Assess Diffusion, Governance and Design

Riebe, Thea (2023)
Technology Assessment of Dual-Use ICTs - How to Assess Diffusion, Governance and Design.
Technische Universität Darmstadt
doi: 10.26083/tuprints-00022849
Ph.D. Thesis, Primary publication, Publisher's Version

Abstract

Technologies that can be used in military and civilian applications are referred to as dual-use. The dual-use nature of many information and communications technologies (ICTs) raises new questions for research and development for national, international, and human security. Measures to deal with the risks associated with the various dual-use technologies, including proliferation control, design approaches, and policy measures, vary widely. For example, Autonomous Weapon Systems (AWS) have not yet been regulated, while cryptographic products are subject to export and import controls. Innovations in artificial intelligence (AI), robotics, cybersecurity, and automated analysis of publicly available data raise new questions about their respective dual-use risks.

Dual-use risks have been systematically discussed so far, especially in the life sciences, which have contributed to the development of methods for assessment and risk management. Dual-use risks arise, among other things, from the fact that safety-critical technologies can be easily disseminated or modified, as well as used as part of a weapon system. Therefore, the development and adaptation of robots and software requires an independent consideration that builds on the insights of related dual-use discourses. Therefore, this dissertation considers the management of such risks in terms of the proliferation, regulation, and design of individual dual-use information technologies. Technology Assessment (TA) is the epistemological framework for this work, bringing together the concepts and approaches of Critical Security Studies (CSS) and Human-Computer Interaction (HCI) to help evaluate and shape dual-use technologies.

In order to identify the diffusion of dual-use at an early stage, the dissertation first examines the diffusion of dual-use innovations between civilian and military research in expert networks on LinkedIn, as well as on the basis of AI patents in a patent network. The results show low diffusion and tend to confirm existing studies on diffusion in patent networks. In the following section, the regulation of dual-use technologies is examined in the paper through two case studies. The first study uses a discourse analysis to show the value conflicts with regard to the regulation of autonomous weapons systems using the concept of Meaningful Human Control (MHC), while a second study, as a long-term comparative case study, analyzes the change and consequences of the regulation of strong cryptography in the U.S. as well as the programs of intelligence agencies for mass surveillance. Both cases point to the central role of private companies, both in the production of AWS and as intermediaries for the dissemination of encryption, as well as surveillance intermediaries. Subsequently, the dissertation examines the design of a dual-use technology using an Open Source Intelligence System (OSINT) for cybersecurity. For this purpose, conceptual, empirical, and technical studies are conducted as part of the Value-Sensitive Design (VSD) framework. During the studies, implications for research on and design of OSINT were identified. For example, the representative survey of the German population has shown that transparency of use while reducing mistrust is associated with higher acceptance of such systems. Additionally, it has been shown that data sparsity through the use of expert networks has many positive effects, not only improving the performance of the system, but is also preferable for legal and social reasons. Thus, the work contributes to the understanding of specific dual-use risks of AI, the regulation of AWS and cryptography, and the design of OSINT in cybersecurity. By combining concepts from CSS and participatory design methods in HCI, this work provides an interdisciplinary and multi-method contribution.

Item Type: Ph.D. Thesis
Erschienen: 2023
Creators: Riebe, Thea
Type of entry: Primary publication
Title: Technology Assessment of Dual-Use ICTs - How to Assess Diffusion, Governance and Design
Language: English
Referees: Reuter, Prof. Dr. Christian ; Nordmann, Prof. Dr. Alfred
Date: 2023
Place of Publication: Darmstadt
Collation: XII, 329 Seiten
Refereed: 7 November 2022
DOI: 10.26083/tuprints-00022849
URL / URN: https://tuprints.ulb.tu-darmstadt.de/22849
Corresponding Links:
Abstract:

Technologies that can be used in military and civilian applications are referred to as dual-use. The dual-use nature of many information and communications technologies (ICTs) raises new questions for research and development for national, international, and human security. Measures to deal with the risks associated with the various dual-use technologies, including proliferation control, design approaches, and policy measures, vary widely. For example, Autonomous Weapon Systems (AWS) have not yet been regulated, while cryptographic products are subject to export and import controls. Innovations in artificial intelligence (AI), robotics, cybersecurity, and automated analysis of publicly available data raise new questions about their respective dual-use risks.

Dual-use risks have been systematically discussed so far, especially in the life sciences, which have contributed to the development of methods for assessment and risk management. Dual-use risks arise, among other things, from the fact that safety-critical technologies can be easily disseminated or modified, as well as used as part of a weapon system. Therefore, the development and adaptation of robots and software requires an independent consideration that builds on the insights of related dual-use discourses. Therefore, this dissertation considers the management of such risks in terms of the proliferation, regulation, and design of individual dual-use information technologies. Technology Assessment (TA) is the epistemological framework for this work, bringing together the concepts and approaches of Critical Security Studies (CSS) and Human-Computer Interaction (HCI) to help evaluate and shape dual-use technologies.

In order to identify the diffusion of dual-use at an early stage, the dissertation first examines the diffusion of dual-use innovations between civilian and military research in expert networks on LinkedIn, as well as on the basis of AI patents in a patent network. The results show low diffusion and tend to confirm existing studies on diffusion in patent networks. In the following section, the regulation of dual-use technologies is examined in the paper through two case studies. The first study uses a discourse analysis to show the value conflicts with regard to the regulation of autonomous weapons systems using the concept of Meaningful Human Control (MHC), while a second study, as a long-term comparative case study, analyzes the change and consequences of the regulation of strong cryptography in the U.S. as well as the programs of intelligence agencies for mass surveillance. Both cases point to the central role of private companies, both in the production of AWS and as intermediaries for the dissemination of encryption, as well as surveillance intermediaries. Subsequently, the dissertation examines the design of a dual-use technology using an Open Source Intelligence System (OSINT) for cybersecurity. For this purpose, conceptual, empirical, and technical studies are conducted as part of the Value-Sensitive Design (VSD) framework. During the studies, implications for research on and design of OSINT were identified. For example, the representative survey of the German population has shown that transparency of use while reducing mistrust is associated with higher acceptance of such systems. Additionally, it has been shown that data sparsity through the use of expert networks has many positive effects, not only improving the performance of the system, but is also preferable for legal and social reasons. Thus, the work contributes to the understanding of specific dual-use risks of AI, the regulation of AWS and cryptography, and the design of OSINT in cybersecurity. By combining concepts from CSS and participatory design methods in HCI, this work provides an interdisciplinary and multi-method contribution.

Alternative Abstract:
Alternative abstract Language

Technologien, welche in militärischen und zivilen Anwendungen verwendet werden können, werden als Dual-Use bezeichnet. Durch den Dual-Use-Charakter vieler Informations- und Kommunikationstechnologien (IKT) ergeben sich für die Forschung und Entwicklung neue Fragen für die nationale, internationale und menschliche Sicherheit. Maßnahmen zum Umgang mit den Risiken bei den verschiedenen Dual-Use-Technologien, u.a. die Verbreitungskontrolle, Gestaltungsansätze sowie politischen Maßnahmen sind sehr unterschiedlich ausformuliert. So sind Autonome Waffensysteme (AWS) bisher nicht reguliert, während kryptographische Produkte Export- und Importkontrollen unterliegen. Innovationen im Bereich der Künstlichen Intelligenz (KI), Robotik, Cybersicherheit und der automatisierten Analyse öffentlich zugänglicher Daten werfen neue Fragen zu ihren jeweiligen Dual-Use-Risiken auf.

Systematisch wurden Dual-Use-Risiken bisher insbesondere in den Biowissenschaften diskutiert, die dazu beigetragen haben, Methoden zur Beurteilung und zum Risikomanagement zu erarbeiten. Dual-Use-Risiken ergeben sich u.a. dadurch, dass sicherheitskritische Technologien einfach verbreitet oder verändert, sowie als Teil einer Waffe verwendet werden können. Die Entwicklung und Anpassung von Robotern und Software benötigt deshalb eine eigenständige Betrachtung, die auf den Erkenntnissen verwandter Dual-Use-Diskursen aufbaut. Deswegen betrachtet die vorliegende Dissertation den Umgang mit solchen Risiken in Form von Verbreitung, Regulierung und Gestaltung von individuellen Dual-Use-Informationstechnologien. Technikfolgenabschätzung (TA) ist der epistemologische Rahmen für diese Arbeit, in der die Konzepte und Ansätze der Kritischen Sicherheitsforschung (CSS) und der Mensch-Computer Interaktion (HCI) zusammengeführt werden und dazu beitragen die Dual-Use-Technologien zu bewerten und zu gestalten.

Um die Verbreitung von Dual-Use frühzeitig erkennen zu können, untersucht die Dissertation zunächst die Diffusion von Dual-Use-Innovationen zwischen ziviler und militärischer Forschung in Expert*innennetzwerken auf LinkedIn, sowie anhand von KI-Patenten in einem Patentnetzwerk. Die Ergebnisse zeigen geringe Diffusion und bestätigen dabei in der Tendenz vorhandene Studien zur Verbreitung in Patentnetzwerken. Im darauffolgenden Abschnitt wird die Regulierung von Dual-Use-Technologien in der Arbeit durch zwei Fallstudien untersucht. Die erste Studie zeigt in einer Diskursanalyse die Wertekonflikte im Hinblick auf die Regulierung von Autonomen Waffensystemen mithilfe des Konzeptes Meaningful Human Control (MHC) auf, während eine zweite Studie als Langzeit-vergleichende Fallstudie den Wandel und die Konsequenzen der Regulierung von starker Kryptographie in den USA sowie der Programme von Geheimdiensten zur Massenüberwachung analysiert. Beide Fälle deuten dabei auf die zentrale Rolle der privaten Unternehmen hin, sowohl bei der Produktion von AWS als auch als Intermediäre zur Verbreitung von Verschlüsselung sowie als Überwachungsintermediäre. Darauffolgend untersucht die Dissertation das Design einer Dual-Use-Technologie anhand eines Open Source Intelligence Systems (OSINT) für die Cybersicherheit. Dafür werden als Teil des Value-Sensitive Design (VSD)-Framework konzeptuelle, empirische und technische Studien durchgeführt. Bei den Studien wurden Implikationen für die Forschung zu und die Gestaltung von OSINT herausgearbeitet. So hat die repräsentative Befragung der deutschen Bevölkerung gezeigt, dass Transparenz der Maßnahmen bei gleichzeitigem Abbau von Misstrauen mit einer höheren Akzeptanz solcher Systeme assoziiert wird. Zusätzlich hat sich gezeigt, dass Datensparsamkeit durch die Verwendung von Expert*innennetzwerke viele positive Effekt hat, nicht nur die Verbesserung der Performanz des Systems, sondern auch aus rechtlichen und sozialen Gründen vorzuziehen ist. Damit leistet die Arbeit einen Beitrag zum Verständnis von spezifischen Dual-Use-Risiken von KI, die Regulierung von AWS und Kryptographie, sowie der Gestaltung von OSINT in der Cybersicherheit. Durch die Kombination der Konzepte aus der CSS und den Methoden der partizipativen Gestaltung in der HCI stellt diese Arbeit hier einen interdisziplinären und multimethodischen Beitrag dar.

German
Status: Publisher's Version
URN: urn:nbn:de:tuda-tuprints-228491
Classification DDC: 000 Generalities, computers, information > 004 Computer science
300 Social sciences > 320 Political science
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Science and Technology for Peace and Security (PEASEC)
Date Deposited: 28 Jul 2023 12:09
Last Modified: 31 Jul 2023 10:50
PPN:
Referees: Reuter, Prof. Dr. Christian ; Nordmann, Prof. Dr. Alfred
Refereed / Verteidigung / mdl. Prüfung: 7 November 2022
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details