Artificial Intelligence (AI) offers huge improvements in the way cars are developed, especially for autonomous vehicles, innovative user interfaces and predictive maintenance. To make that a reality, companies face the question: Make or buy of AI? To be more precise: How much “Make AI” should a company do to benefit best from these opportunities? Experience from non-AI systems is that companies need to be able to specify, integrate and test acquired content in the context of the system. For AI, the make or buy decision has additional important dimensions such as the degree to which the acquired learning module is trained by the buyer, and to which own data is used instead of publicly available or supplied data to train it. Competencies, resources, cost, value creation and the access to domain knowledge that one may open to a supplier (that would be better kept for the company itself) are additional strategic aspects of the decision. Several levels of “Make-AI” are defined and evaluated, ranging from “Black Box AI” to “Complete Responsibility for Design, Coding and Data”. These levels are evaluated concerning specification, data collection and preparation, implementation, integration, SW test, system test, Safety/SOTIF/ISO TR 4804 evaluation as well as strategic criteria. The result guides practitioners and management to select the appropriate level and benefit best from the huge opportunities with AI.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Make or Buy Strategy for AI in Automotive: How Much “Make-AI” is Necessary to Succeed?


    Weitere Titelangaben:

    Proceedings


    Beteiligte:
    Bargende, Michael (Herausgeber:in) / Reuss, Hans-Christian (Herausgeber:in) / Wagner, Andreas (Herausgeber:in) / Bodenhausen, Ulrich (Autor:in)

    Erschienen in:

    Erscheinungsdatum :

    2022-03-14


    Format / Umfang :

    12 pages





    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Deutsch