CHAPTER 23 OF 29
Social Intelligence
From scattered research to causal understanding and better institutional choices.
The purpose of social intelligence is to turn scattered knowledge into better explanations, decisions, and shared conditions.
- 01Four engines distinguish measurement, evidence appraisal, causal inquiry, and institutional recommendation.
- 02Policy comparisons must preserve uncertainty, local applicability, distribution, and feasibility.
- 03A system becomes useful when its findings improve decisions in practice, not merely when it produces persuasive reports.
In this chapter · 7 sections
The research and institutional domain of Entelect
Social Intelligence develops the conception of Entelect as an AI-assisted research and policy system. It draws on scientific findings, historical experience, institutional evaluations, professional knowledge, and relevant local accounts. Its ambition is broad, improving coverage; claims to have exhaustively absorbed all human evidence would be unjustified.
The domain addresses an institutional problem as much as an information problem. A finding has to be understood, judged applicable, translated into a feasible choice, implemented competently, and evaluated. Failure at any stage can leave good knowledge without useful consequences.
The Eudometric Engine: what is happening?
This engine describes dimensions and distributions of human outcomes. It might examine learning, health, agency, meaningful activity, relationships, security, access, and trust. Measurement retains distinctions between outcomes and administrative activity, between short and long horizons, and between averages and affected groups.
Its task also includes noticing what is missing. A readily available proxy must not become the definition of the good simply because the database can supply it. Qualitative evidence and participants' accounts can challenge the measures or reveal a neglected dimension. Composite summaries should expose their weights and sensitivity rather than conceal conflict behind one score.
The Evidence and Epistemic Engine: what do we know?
This engine gathers and appraises findings while retaining their origins, methods, populations, dates, limitations, and relationships. It identifies overlapping datasets, contradictory results, null findings, retractions, missing populations, and uncertain transfer between settings. Broad coverage must be joined to quality assessment.
Its useful output distinguishes observation, inference, dispute, and consequential uncertainty. Historical examples can suggest mechanisms and institutional possibilities, but resemblance does not establish causal equivalence. The same rule applies to the Coactualist framework: a philosophical proposal can define what is being suggested without demonstrating its empirical effectiveness.
The Causal Intelligence Engine: what would change?
Prediction and intervention answer different questions. A system may accurately forecast an outcome while misunderstanding how to improve it. Causal inquiry asks what would happen under a specified change compared with a relevant alternative, for particular people, in particular conditions, over a defined period.
The inquiry examines competing explanations, selection, reverse causation, implementation, and effects beyond the immediate target. Experiments, natural experiments, longitudinal comparisons, and other research designs can contribute where appropriate. Simulations explore assumptions; their detail does not validate those assumptions. Sometimes the responsible output is a proposal for further investigation rather than a confident prescription.
The Aretonomic Engine: what action is justified?
This engine compares feasible responses to the evidence. It considers benefits, burdens, distribution, resources, implementation capacity, transition, reversibility, and the consequences of continuing current arrangements. A recommendation should identify the justified action, alternatives, responsible decision maker, uncertainties, and review conditions.
Where goods conflict, the value judgment remains visible. An expensive intervention with a larger individual effect may reach fewer people than a modest alternative. A promising result may warrant a pilot before broad adoption. Aretonomy prevents the comparison from ending in endless acknowledgment: sufficiently established improvements demand a consequential institutional response.
Local adaptation and the meso level
A policy enters life through organizations. Staffing, routines, resources, relationships, and incentives can change what the same formal intervention accomplishes. Meso-level inquiry studies those conditions instead of explaining every failure through the characteristics of participants.
Shared learning should preserve contextual interpretation. Common definitions can support comparison while different communities retain room to adapt. A population average cannot automatically determine an individual's needs, and a success with one person does not establish what will happen when an approach is adopted across a system.
AN ILLUSTRATIVE APPLICATION
Comparing school schedules
A district considers changing its timetable. The evidence engine reviews relevant findings; causal inquiry examines alternative explanations; Eudometrics defines learning, fatigue, attendance, family burdens, and uneven effects. The Aretonomic Engine compares workable options, including transport and staffing. A pilot's criteria are established before results arrive. Responsible public institutions decide, and subsequent experience can revise both the policy and its supporting model.
CONTINUE ALONG THE THREAD
Follow the connections
The whole Entelect architecture
How this domain cooperates with personal development and oversight.
Eudonomy
The policy design questions the intelligence is meant to serve.
The learning loop
How practical outcomes return to improve the knowledge system.
FROM READING TO CONVERSATION
What would you like to explore?
Take an idea from this chapter further. Choose a starting question, then make it your own before sending.
What information would turn an attractive policy idea into a justified, implementable, and reviewable decision?
Explore this questionAsk your own question