Daniel Burgwinkel & Asmaa Ali Abdelwahab

Contact Info
Daniel Burgwinkel & Asmaa Ali Abdelwahab
Edelweiss Connect

OpenTox Summer School 2024

Application of Knowledge Graphs and AI in Risk Assessment

Daniel Burgwinkel, Asmaa Ali Abdelwahab

Edelweiss Connect

Abstract

This session explores the application of Knowledge Graphs (KGs) and Artificial Intelligence (AI) in the domain of safety risk assessment. We identify key use cases where AI and KGs can enhance risk assessment processes, providing valuable insights and improving decision-making. The talk delves into the benefits and limitations of integrating AI and KGs into risk assessment frameworks, emphasizing the potential for improved efficiency, accuracy, and comprehensiveness. Given the increasing reliance on AI in high-stakes decision-making, we examine the regulatory landscape, focusing on the EU AI Act as a representative example. We highlight the specific challenges and requirements for deploying AI, including Generative AI and Large Language Models (LLMs), and KGs in risk assessment contexts. Addressing security concerns is paramount, and we discuss potential vulnerabilities and mitigation strategies when using AI in risk assessment systems. To ensure the trustworthiness and reliability of AI and KG-driven risk assessments, we propose a quality and compliance framework. This framework outlines essential components for evaluating the performance, robustness, and ethical implications of AI and KG systems in the context of safety risk assessments. This session contributes to the ongoing discourse on responsible AI development and deployment by providing practical guidance for organizations seeking to leverage these technologies for enhanced risk management.

Topics:

● Use Cases for applying AI and KG in Risk Assessments

  • Use Case AI: Prediction with AI of risks (e.g. Smart Safety)
  • Use Case KG: search for available information (current state of the literature)
  • Use Case KG: Identify chemical compounds in the product and in similar products

● Benefits and constraints of AI and KG

● Regulatory requirements for using AI (incl. GenAI and LLMs) and KG in decision making with High-Risk AI Systems (e.g. EU AI Act)

● Security issues and risks of using AI

● Towards a quality and compliance framework for AI and KG in safety risk assessment