Executive summary: This exploratory research agenda proposes mapping a “phylogeny of agents”—tracing how different disciplines have evolved distinct frameworks for understanding agency—to improve AI alignment, arguing that systems can be simultaneously aligned and misaligned depending on the analytical lens, and that understanding these lenses’ evolutionary relationships can prevent dangerous blind spots.
Key points:
An ant colony can be described as a market system, a superorganism, or a distributed cognitive process, with each framing offering accurate yet distinct predictions about its behavior.
These divergent perspectives are not arbitrary but evolved analytical tools shaped by the specific prediction challenges each field faced, echoing Dennett’s “intentional stance.”
A “phylogeny of agents” would map the historical and mathematical relationships between these frameworks, revealing when they can be productively combined and when they might conflict.
In AI alignment, especially for multi-agent or hybrid human–AI systems, different stances can yield conflicting assessments of alignment—making a phylogenetic approach vital for avoiding mismatched solutions.
Such a map could enable transferring alignment strategies between fields, predicting the compatibility of different approaches, and identifying blind spots before they cause systemic risks.
The author’s team is launching a workshop and research program in autumn 2025 at Equilibria Network to develop this conceptual infrastructure.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.
Executive summary: This exploratory research agenda proposes mapping a “phylogeny of agents”—tracing how different disciplines have evolved distinct frameworks for understanding agency—to improve AI alignment, arguing that systems can be simultaneously aligned and misaligned depending on the analytical lens, and that understanding these lenses’ evolutionary relationships can prevent dangerous blind spots.
Key points:
An ant colony can be described as a market system, a superorganism, or a distributed cognitive process, with each framing offering accurate yet distinct predictions about its behavior.
These divergent perspectives are not arbitrary but evolved analytical tools shaped by the specific prediction challenges each field faced, echoing Dennett’s “intentional stance.”
A “phylogeny of agents” would map the historical and mathematical relationships between these frameworks, revealing when they can be productively combined and when they might conflict.
In AI alignment, especially for multi-agent or hybrid human–AI systems, different stances can yield conflicting assessments of alignment—making a phylogenetic approach vital for avoiding mismatched solutions.
Such a map could enable transferring alignment strategies between fields, predicting the compatibility of different approaches, and identifying blind spots before they cause systemic risks.
The author’s team is launching a workshop and research program in autumn 2025 at Equilibria Network to develop this conceptual infrastructure.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.