Research on Artificial Intelligence continues to advance at a rapid pace, with developments chased by a host of governments, regulatory agencies, and oversight bodies. No one is sure what the future will bring, either for the technology itself or the laws and policies that govern it. What is clear is the transformative power of AI for human experience.

 

Part of the trouble with prediction, regulation, and ethics in this area is that the term ‘Artificial Intelligence’ refers to a sprawling and diverse field of technology. It is not the case that every AI system is like every other. Task-respondent AI systems function differently than AI systems designed to simulate agents within teams or dynamic decision-making. Different goals yield different methods, functions, applications, and outcomes, and moral relevance and outcomes change when we consider various aspects of human-AI activity. How to design an AI system with goals such as trustworthiness in mind is one question; how doctors should make use of recommendations from AI is another. These questions differ and cannot be answered by the same moral concepts, methods, or practices.

 

To narrow the focus and frame our conversations, the topic for Policy Talks 2024 is “When Human Narratives Meet Artificial Intelligence: Responsible Design and User Protection.” Narratives encode and convey information among humans with an efficiency, memorability, and impact that mere propositions cannot. Narratives are powerful tools for communication and understanding, and AI systems that make use of narrative features and dynamic functions affect user psychology in powerful ways.

 

Our focus on narrative is not about students cheating on their creative writing papers using AI systems. It is about how some AI systems utilize narrative features and how they impact the psychology of human agents. We are familiar with algorithms on social media that can maintain and transmit narratives which strongly reinforce beliefs that desperately need to be questioned. Similarly, AI systems used in police training can support a narrative—operating below the level of a user’s conscious awareness—that has devastating consequences.

 

How are we to understand narratives and AI? What moral obligations do users and developers have to ensure its safe and ethical use? What guardrails ought to be in place for design and application? These are open questions that need to be answered. Policy Talks 2024 is our contribution to mapping out the moral landscape for AI, ethics, and narratives. We will examine existing policy proposals, analysis and ethical concepts from philosophy, methods and technologies from data and computer science, concepts of narrative and their features from narratology, and research from psychologists on the impact on human agents to explore answers. We will examine this underexplored area of human-AI interaction, ask hard questions, and work toward practical solutions to real-world problems.