Recap of my research topic and its theoretical framework:
My dream project called Symbiogenetic Evolution as a Posthuman Feminist Framework for AI Ethics: Beyond Anthropocentric and Patriarchal Paradigms in Algorithmic Design.This research proposes a posthuman feminist theoretical intervention in artificial intelligence (AI) ethics through the lens of Lynn Margulis’s symbiogenetic evolutionary theory. The project examines how Darwinian paradigms embedded in genetic algorithms (GA) perpetuate both patriarchal and anthropocentric power structures in AI development. It explores how Margulis’s theory of endosymbiosis could inspire more inclusive and ethically sound algorithmic frameworks. At its core, this research seeks to challenge the current AI development assumptions and propose alternative paradigms that better serve marginalised voices, including both human and non-human actors.
Contemporary AI development, particularly in GA, remains deeply influenced by Darwinian evolutionary theorythat emphasise conquest and competition, individual fitness, and human supremacy. This phenomenon reflects a historical pattern where scientific theories have been closely connected with social power structures and human exceptionalism. However, this “patriarchal-anthropocentric recursion” in AI development may promote patriarchal power structures, exacerbate social and ecological inequalities, which poses challenges for disadvantaged voices of marginalise humans and non-humans.
Margulisās theory, which I propose to utlise as the foundation of my reconceptualization of AI ethics, shows how symbiotic connections outside species borders and hierarchical power systems drove biological evolution. Her hypothesis of endosymbiosis contradicts Darwinian narratives of competition and human exceptionalism by showing how human existence is based on ancient symbiotic ties with microorganisms. Thus, I argue that posthuman feminist approaches can interrupt the limited loop of AI development. The truly innovative AI systems may emerge from algorithms that emphasise multi-species cooperation and mutual enhancement.
The methods I have explored so far:
The research will proceed through several integrated phases, beginning with a thorough analysis of how Darwinian concepts currently shape AI systems, considering both patriarchal and anthropocentric aspects of current systems. By quantifying and mapping implicit patriarchal and anthropocentric biases through techniques such as topic modelling, sentiment analysis, and network analysis of AI training corpora, the research can then build a rigorous evidence base to provide scientific support for subsequent interventions or algorithmic improvements.
In theĀ first phase, this analysis will focus on examining how Darwinian paradigms influence current AI systems, particularly in genetic algorithms (GA). To achieve this, I will employĀ multi-case studiesĀ andĀ content analysisĀ to investigate publicly available documentation, research papers, and software development practices that reflect the underlying assumptions of AI algorithms. The collected data will then be subjected to a combination ofĀ deductive and inductive thematic analysisĀ to identify recurring patterns and themes related to patriarchal and anthropocentric biases. Tools such asĀ topic modelling,Ā sentiment analysis, andĀ network analysisĀ will be utilised to quantify and map these biases in AI training corpora. This phase aims to develop an evidence base that demonstrates the specific mechanisms through which Darwinian concepts shape AI systems.
TheĀ second phase, informed by the results and analysis from the first phase, will involve developing theoretical frameworks for translating Margulisās symbiogenetic principles into algorithmic terms, emphasising multi-species cooperation and non-hierarchical relationships. This will require conductingĀ secondary data analysisĀ of existing AI systems that have incorporated cooperative or symbiotic elements. Additionally, I will gather insights fromĀ expert interviewsĀ with AI developers and ethicists to refine the theoretical foundations. UsingĀ graph-based modellingĀ andĀ multi-agent systems, I aim to simulate cooperative, reciprocal, and resource-sharing dynamics. Evolutionary techniques, such asĀ co-evolutionaryĀ andĀ multi-objective algorithms, will be employed to embed symbiotic interactions. Furthermore, interpretability and fairness metrics will be integrated to ensure the resulting frameworks align with ethical considerations.
TheĀ final phaseĀ will focus on validating and refining the proposed frameworks through practical implementation. This phase will involveĀ small-scale experimental simulationsĀ of the proposed algorithms in controlled environments to evaluate their ethical and functional efficacy. Iterative testing will combineĀ qualitative feedbackĀ from experts withĀ quantitative metricsĀ of algorithm performance. The goal is to develop a set of practical implementation guidelines for integrating these algorithms into real-world AI systems. This will culminate in the creation of what I termĀ ‘endosymbiotic evolutionary algorithms’, which emphasise multi-species cooperation and mutual enhancement.
The potential impact of this research extends beyond academic circles. By providing both theoretical frameworks and practical guidelines for creating more inclusive and collaborative AI systems, this work could help reshape how we approach AI development more broadly. The truly innovative AI systems may emerge from algorithms that amplify marginalised voices, fostering equitable technological development and challenging the limitations of patriarchal and anthropocentric paradigms.
