Safety challenges of ai in autonomous systems design – solutions from human factors perspective emphasizing ai awareness

Hannu Karvonen*, Eetu Heikkilä, Mikael Wahlström

*Corresponding author for this work

    Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

    5 Citations (Scopus)

    Abstract

    Artificial intelligence (AI) is a key technology that is utilized in autonomous systems. However, using AI in such systems introduces also several safety challenges, many of which are related to complex human-AI interactions. In general, these challenges relate to the changing roles of the people who interact with the increasingly autonomous systems in various ways. In this paper, we consider a set of practical safety challenges related to applying advanced AI to autonomous machine systems and present solutions from the perspective of human factors. We apply the novel concept of AI awareness (AIA) to discuss the challenges in detail, and based on this, provide suggestions and guidelines for mitigating the AI safety risks with autonomous systems. In addition, we briefly consider the system design process to identify the actions that should be taken to ensure AIA throughout the different systems engineering design phases. The theoretical research presented in this paper aims to provide the first steps towards considering AIA in autonomous systems and understanding the human factors perspective viewpoint on AI safety for autonomous systems.

    Original languageEnglish
    Title of host publicationEngineering Psychology and Cognitive Ergonomics. Cognition and Design
    Subtitle of host publicationHCII 2020
    EditorsDon Harris, Wen-Chin Li
    Place of PublicationCham
    PublisherSpringer
    Pages147-160
    ISBN (Electronic)978-3-030-49183-3
    ISBN (Print)978-3-030-49182-6
    DOIs
    Publication statusPublished - 2020
    MoE publication typeA4 Article in a conference publication
    Event22nd International Conference on Human-Computer Interaction, HCII 2020 - Copenhagen, Denmark
    Duration: 19 Jul 202024 Jul 2020

    Publication series

    SeriesLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume12187
    ISSN0302-9743

    Conference

    Conference22nd International Conference on Human-Computer Interaction, HCII 2020
    Country/TerritoryDenmark
    CityCopenhagen
    Period19/07/2024/07/20

    Keywords

    • Artificial intelligence
    • Autonomous systems
    • Human factors
    • Safety
    • Systems engineering

    Fingerprint

    Dive into the research topics of 'Safety challenges of ai in autonomous systems design – solutions from human factors perspective emphasizing ai awareness'. Together they form a unique fingerprint.

    Cite this