Abeer Dyoub (Università degli studi di Bari "Aldo Moro", FAIR Spoke 6 - Symbiotic AI) present"Towards Ethical Risk Assessment with Fuzzy Rules in Symbiotic AI".
This presentation is part of the Virtual Young Poster Session of the FAIR 2024 General Conference.
For more information: [ Ссылка ]
Artificial Intelligence (AI) based Systems are expanding rapidly in all domains of life. They are entering our everyday life and performing tasks on our behalf. AI-based systems are increasingly engaging in close symbiotic relationships with humans, some examples of such symbiotic AI systems are digital twins, personal healthcare assistants, virtual avatars, etc. Symbiotic AI promises improved outcomes in various domains such as healthcare, education, and business. However, as the degree of symbiosis increases, so does the ethical risk. To ensure that these systems behave ethically and do not cause harm of any kind (physical, mental, violation of privacy, etc.), we need to find ways to assess the ethical risk (risk of causing harm), then choose the right action to mitigate that risk.
Nobody can precisely estimate possible ethical risks without a comprehensive understanding of all aspects of the risk system being studied. In practical scenarios, it is impossible to completely eliminate gaps in ethical risk assessment, leading to fuzziness (imprecision, vagueness, incompleteness, etc.). Therefore, it is essential to address and
manage the inherent fuzziness within ethical risk systems. In this work, we propose an ethical risk assessment (ERA) approach based on fuzzy logic. Fuzzy logic provides a flexible framework capable of capturing and processing vague, imprecise, and uncertain information, leading to more nuanced and comprehensive ethical risk assessments.
