PSI - Issue 62

ScienceDirect Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 Available online at www.sciencedirect.com ScienceDirect Structural Integrity Procedia 00 (2022) 000–000 Available online at www.sciencedirect.com

www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia

Procedia Structural Integrity 62 (2024) 276–284

II Fabre Conference – Existing bridges, viaducts and tunnels: research, innovation and applications (FABRE24) Trustworthy AI for infrastructure monitoring: a blockchain-based approach Fabio Severino 1a , Andrea Canciani a , Claudio Felicioli b , Vincenzo Gervasi c , Andrea Pelosi c , Agnese Natali d , Walter Salvatore e , Simone Russo e a Geckosoft, Via San Lorenzo 6, 56127 Pisa, Italy b Traent, Borgo Stretto 3, 56127, Pisa, Italy c University of Pisa, Department of Computer Science, Largo Bruno Pontecorvo 3, 56127, Pisa, Italy d University of Pisa, Department of Industrial and Civil Engineer, Largo Lucio Lazzarino 1, 56122, Pisa, Italy e Inspectiondrone, Via San Raffaele 1, 20121, Milano, Italy Abstract In the field of Artificial Intelligence (AI), there is an increasing focus on enhancing trustworthiness especially in critical sectors such as in the management of civil infrastructure. This paper proposes the adoption of a framework based on Hybrid Distributed Ledger Technology (Hybrid-DLT) as a technological solution for improving trustworthiness. We detail three specific applications in the sector of critical infrastructure maintenance: Explainable AI (XAI) for risk classification, structural defects recognition, and real-time monitoring through IoT. The proposed approach employs tamper-resistant ledgers for tracking key processes such as dataset collection, model training, and inference generation, thereby ensuring non-repudiability for recorded actions and enabling auditability. We demonstrate how this strengthens the explainability mechanisms of AI models and enables the production of verifiable data lineage and certified inferences. Our framework can be applied to existing AI solutions, enhancing their trustworthiness. © 2024 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4.0 ) Peer-review under responsibility of Scientific Board Members Keywords: Trustworthy AI, Hybrid Blockchain, Machine Learning, Expert Systems, Risk Assessment, Infrastructure Monitoring Introduction The primary metric for evaluating Artificial Intelligence (AI) models has traditionally been predictive accuracy. Although this focus on accuracy has driven significant advancements, it is not sufficient for all application domains. Highly regulated fields, such as infrastructure maintenance, demand a different emphasis. These fields face II Fabre Conference – Existing bridges, viaducts and tunnels: research, innovation and applications (FABRE24) Trustworthy AI for infrastructure monitoring: a blockchain-based approach Fabio Severino 1a , Andrea Canciani a , Claudio Felicioli b , Vincenzo Gervasi c , Andrea Pelosi c , Agnese Natali d , Walter Salvatore e , Simone Russo e a Geckosoft, Via San Lorenzo 6, 56127 Pisa, Italy b Traent, Borgo Stretto 3, 56127, Pisa, Italy c University of Pisa, Department of Computer Science, Largo Bruno Pontecorvo 3, 56127, Pisa, Italy d University of Pisa, Department of Industrial and Civil Engineer, Largo Lucio Lazzarino 1, 56122, Pisa, Italy e Inspectiondrone, Via San Raffaele 1, 20121, Milano, Italy Abstract In the field of Artificial Intelligence (AI), there is an increasing focus on enhancing trustworthiness especially in critical sectors such as in the management of civil infrastructure. This paper proposes the adoption of a framework based on Hybrid Distributed Ledger Technology (Hybrid-DLT) as a technological solution for improving trustworthiness. We detail three specific applications in the sector of critical infrastructure maintenance: Explainable AI (XAI) for risk classification, structural defects recognition, and real-time monitoring through IoT. The proposed approach employs tamper-resistant ledgers for tracking key processes such as dataset collection, model training, and inference generation, thereby ensuring non-repudiability for recorded actions and enabling auditability. We demonstrate how this strengthens the explainability mechanisms of AI models and enables the production of verifiable data lineage and certified inferences. Our framework can be applied to existing AI solutions, enhancing their trustworthiness. © 2024 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4.0 ) Peer-review under responsibility of Scientific Board Members Keywords: Trustworthy AI, Hybrid Blockchain, Machine Learning, Expert Systems, Risk Assessment, Infrastructure Monitoring 1. Introduction The primary metric for evaluating Artificial Intelligence (AI) models has traditionally been predictive accuracy. Although this focus on accuracy has driven significant advancements, it is not sufficient for all application domains. Highly regulated fields, such as infrastructure maintenance, demand a different emphasis. These fields face © 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of Scientific Board Members 1.

1 * Fabio Severino. E-mail address: severino@geckosoft.it 1 * Fabio Severino. E-mail address: severino@geckosoft.it

2452-3216 © 2024 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4. 0 ) Peer-review under responsibility of Scientific Board Member s 2452-3216 © 2024 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license ( https://creativecommons.org/licenses/by-nc-nd/4. 0 ) Peer-review under responsibility of Scientific Board Member s

2452-3216 © 2024 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of Scientific Board Members 10.1016/j.prostr.2024.09.043