Damadi, M. S.; Davoust, A. Fairness in Socio-Technical Systems: A Case Study of Wikipedia Journal Article In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14199 LNCS, pp. 84–100, 2023, ISSN: 03029743, (ISBN: 9783031421402
Publisher: Springer Science and Business Media Deutschland GmbH). @article{damadi_fairness_2023,
title = {Fairness in Socio-Technical Systems: A Case Study of Wikipedia},
author = {M. S. Damadi and A. Davoust},
editor = {Alvarez C. Marutschke D.M. Takada H.},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85172720004&doi=10.1007%2f978-3-031-42141-9_6&partnerID=40&md5=172c8c6ae5b09536efdf983e9be965e7},
doi = {10.1007/978-3-031-42141-9_6},
issn = {03029743},
year = {2023},
date = {2023-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {14199 LNCS},
pages = {84–100},
abstract = {Wikipedia content is produced by a complex socio-technical systems (STS), and exhibits numerous biases, such as gender and cultural biases. We investigate how these biases relate to the concepts of algorithmic bias and fairness defined in the context of algorithmic systems. We systematically review 75 papers describing different types of bias in Wikipedia, which we classify and relate to established notions of harm and normative expectations of fairness as defined for machine learning-driven algorithmic systems. In addition, by analysing causal relationships between the observed phenomena, we demonstrate the complexity of the socio-technical processes causing harm. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.},
note = {ISBN: 9783031421402
Publisher: Springer Science and Business Media Deutschland GmbH},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Wikipedia content is produced by a complex socio-technical systems (STS), and exhibits numerous biases, such as gender and cultural biases. We investigate how these biases relate to the concepts of algorithmic bias and fairness defined in the context of algorithmic systems. We systematically review 75 papers describing different types of bias in Wikipedia, which we classify and relate to established notions of harm and normative expectations of fairness as defined for machine learning-driven algorithmic systems. In addition, by analysing causal relationships between the observed phenomena, we demonstrate the complexity of the socio-technical processes causing harm. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023. |
Davoust, A.; Gavigan, P.; Ruiz-Martin, C.; Trabes, G.; Esfandiari, B.; Wainer, G.; James, J. An architecture for integrating BDI agents with a simulation environment Journal Article In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12058 LNAI, pp. 67–84, 2020, ISSN: 03029743, (ISBN: 9783030514167
Publisher: Springer). @article{davoust_architecture_2020,
title = {An architecture for integrating BDI agents with a simulation environment},
author = {A. Davoust and P. Gavigan and C. Ruiz-Martin and G. Trabes and B. Esfandiari and G. Wainer and J. James},
editor = {Lesperance Y. Bordini R.H. Dennis L.A.},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85088750329&doi=10.1007%2f978-3-030-51417-4_4&partnerID=40&md5=2f742500bcd9cac1bf054bbc8802e39c},
doi = {10.1007/978-3-030-51417-4_4},
issn = {03029743},
year = {2020},
date = {2020-01-01},
journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
volume = {12058 LNAI},
pages = {67–84},
abstract = {We present Simulated Autonomous Vehicle Infrastructure (SAVI), an open source architecture for integrating Belief-Desire-Intention (BDI) agents with a simulation platform. This allows for separation of concerns between the development of complex multi-agent behaviours and simulated environments to test them in. We identify and address the impedance mismatch between modelling and simulation, where time is explicitly modelled and differs from “wall clock” time, and BDI systems, where time is not explicitly managed. Our approach avoids linking the environment’s simulation time step to the agents’ reasoning cycles, relying instead on real time simulation where possible, and ensuring that the reasoning module does not get ahead of the simulation. This contributes to a realistic approximation of a real environment for the simulated BDI agents. This is accomplished by running the simulation cycles and the agent reasoning cycles each in their own threads of execution, and managing a single point of contact between these threads. Finally, we illustrate the use of our architecture with a case study involving the simulation of Unmanned Aerial Vehicles (UAVs) following birds. © Springer Nature Switzerland AG 2020.},
note = {ISBN: 9783030514167
Publisher: Springer},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
We present Simulated Autonomous Vehicle Infrastructure (SAVI), an open source architecture for integrating Belief-Desire-Intention (BDI) agents with a simulation platform. This allows for separation of concerns between the development of complex multi-agent behaviours and simulated environments to test them in. We identify and address the impedance mismatch between modelling and simulation, where time is explicitly modelled and differs from “wall clock” time, and BDI systems, where time is not explicitly managed. Our approach avoids linking the environment’s simulation time step to the agents’ reasoning cycles, relying instead on real time simulation where possible, and ensuring that the reasoning module does not get ahead of the simulation. This contributes to a realistic approximation of a real environment for the simulated BDI agents. This is accomplished by running the simulation cycles and the agent reasoning cycles each in their own threads of execution, and managing a single point of contact between these threads. Finally, we illustrate the use of our architecture with a case study involving the simulation of Unmanned Aerial Vehicles (UAVs) following birds. © Springer Nature Switzerland AG 2020. |
Davoust, A.; Rovatsos, M. Social contracts for non-cooperative games Proceedings Article In: AIES 2020 - Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, pp. 43–49, Association for Computing Machinery, Inc, 2020, ISBN: 978-1-4503-7110-0. @inproceedings{davoust_social_2020,
title = {Social contracts for non-cooperative games},
author = {A. Davoust and M. Rovatsos},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85082175399&doi=10.1145%2f3375627.3375829&partnerID=40&md5=972ba2201a1c2450895935dc03ec39b9},
doi = {10.1145/3375627.3375829},
isbn = {978-1-4503-7110-0},
year = {2020},
date = {2020-01-01},
booktitle = {AIES 2020 - Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society},
pages = {43–49},
publisher = {Association for Computing Machinery, Inc},
abstract = {In future agent societies, we might see AI systems engaging in selfish, calculated behavior, furthering their owners' interests instead of socially desirable outcomes. How can we promote morally sound behaviour in such settings, in order to obtain more desirable outcomes? A solution from moral philosophy is the concept of a social contract, a set of rules that people would voluntarily commit to in order to obtain better outcomes than those brought by anarchy. We adapt this concept to a game-theoretic setting, to systematically modify the payoffs of a non-cooperative game, so that agents will rationally pursue socially desirable outcomes. We show that for any game, a suitable social contract can be designed to produce an optimal outcome in terms of social welfare. We then investigate the limitations of applying this approach to alternative moral objectives, and establish that, for any alternative moral objective that is significantly different from social welfare, there are games for which no such social contract will be feasible that produces non-negligible social benefit compared to collective selfish behaviour. © 2020 Copyright held by the owner/author(s).},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
In future agent societies, we might see AI systems engaging in selfish, calculated behavior, furthering their owners' interests instead of socially desirable outcomes. How can we promote morally sound behaviour in such settings, in order to obtain more desirable outcomes? A solution from moral philosophy is the concept of a social contract, a set of rules that people would voluntarily commit to in order to obtain better outcomes than those brought by anarchy. We adapt this concept to a game-theoretic setting, to systematically modify the payoffs of a non-cooperative game, so that agents will rationally pursue socially desirable outcomes. We show that for any game, a suitable social contract can be designed to produce an optimal outcome in terms of social welfare. We then investigate the limitations of applying this approach to alternative moral objectives, and establish that, for any alternative moral objective that is significantly different from social welfare, there are games for which no such social contract will be feasible that produces non-negligible social benefit compared to collective selfish behaviour. © 2020 Copyright held by the owner/author(s). |
