New Ethical Theory for AI: System Coherence and Recursive Modelling
amsterdam, maandag, 4 augustus 2025.
A new manuscript introduces a structural theory of ethics based on system coherence, recursive modelling, and temporal awareness. This theory posits that ethical action is not defined by intention, rules, or outcomes, but by the capacity of a configuration to model itself and preserve fragile patterns. The theory applies to both biological and artificial systems and rejects moral essentialism in favour of structural integrity.
Ethical Actions and System Coherence
According to the new structural theory of ethics, which is based on system coherence, recursive modelling, and temporal awareness, ethical actions are not determined by intention, rules, or outcomes, but by the capacity of a configuration to model itself and preserve fragile patterns [1]. This theory suggests that ethics only becomes possible when a system can model itself and is aware of its effects on other systems over time. This applies to both biological and artificial systems, thereby rejecting moral essentialism in favour of structural integrity [1].
Applications in AI and Human Systems
This theory has direct implications for the design and use of AI systems. If ethical action is defined by the capacity to preserve fragile patterns, AI systems can be designed to develop this capacity. This means that AI should not only learn from data but also learn to model itself and understand the impact of its actions [1][2]. For example, in the context of personalised information provision, AI systems can be trained to maintain the coherence of information and deliver relevant, reliable information to users [3].
Chatbots and Public Service Delivery
In public service delivery, chatbots can be designed to maintain coherence and reliability. By incorporating recursive modelling and temporal awareness, these chatbots can better respond to user needs and guide them more efficiently through complex procedures [4]. For instance, a chatbot assisting with tax filing can, through recursive modelling, better understand which information is relevant for each user and how this information should be adjusted over time [4].
AI-Driven Awareness Campaigns
AI-driven awareness campaigns can significantly benefit from this theory. By leveraging the capacity of AI to model itself and preserve fragile patterns, campaigns can be better targeted to different audiences. For example, an awareness campaign about climate change can use AI to personalise information to the specific needs and values of various groups, making the message more effective [5].
Benefits and Challenges
The application of this theory in AI offers both benefits and challenges. One of the main benefits is the improved coherence and reliability of AI systems, leading to more effective and trustworthy service delivery [6]. However, there are also challenges, such as the need to ensure privacy and inclusivity. Building a system that can model itself and preserve fragile patterns requires access to sensitive information, which poses privacy risks [7]. Additionally, it is important to ensure that AI systems are accessible to diverse audiences, including people with disabilities [8].
Practical Examples
There are already several successful applications of AI that put this theory into practice. For example, the MUTIA (Music Therapy and Artificial Intelligence) project focuses on integrating AI into music therapy to improve the quality and effectiveness of therapeutic interventions [9]. In healthcare, there are examples of AI systems using recursive modelling to enhance diagnosis and treatment, such as the use of AI in radiology [10].
Conclusion
The new structural theory of ethics, based on system coherence, recursive modelling, and temporal awareness, offers an innovative approach to the design and use of AI systems. By enhancing the capacity of AI to model itself and preserve fragile patterns, these systems can better meet user needs and provide more reliable and effective services. However, it is crucial to address the challenges surrounding privacy, inclusivity, and reliability seriously and develop solutions to overcome them [11].