The business analysis of a certified inference framework for safety-critical applications

Abstract

The emergence of AI technologies opens up new business opportunities but requires many players along the entire value chain. This work analyses the ecosystem of an automated, AI-based decision-making platform for safety-critical industrial areas such as automated driving, mixed-criticality applications or human-robot collaboration. First, the technical foundation is presented based on its problem formulation. Currently, deep learning methods as a discipline of sub-symbolic artificial intelligence lack transparency and explainability. Hence, an innovative concept for a certified inference framework in the context of safety-critical applications is proposed. The framework comprises an embedded hardware platform utilising recent findings in accelerating neural networks and a software ecosystem to decrease the risk of systematic errors during training. A comprehensive analysis of the entrepreneurial environment based on patent applications and investments in AI technologies concludes a promising $67 billion market for highly automated driving by 2020 and an $18 billion market opportunity by 2022 in the area of edge computing. Within these sectors, automated decision-making systems will demand a certified platform to ensure standard conformity and compliance. Another vital part of the work is the analysis of the business idea under three aspects, namely ecosystem, value proposition and value chains. The work identifies and investigates the stakeholder's interdependencies within the AI ecosystem using a specific framework. Moreover, the innovation potential of the proposed solution among a defined set of product criteria is developed, and the relation to existing value chains is discussed. The analysis concludes that neglecting the ecosystem of the core product most likely limits the full potential of this business idea in the field of AI.