Designing a Virtuous Automated Decision system using multiple agents with Personas and human-in-the-loop
Bibliographic record
Abstract
This conference contribution was presented by Asma Tajuddin, University of Windsor, Canada. (Qatar International Conference on Debate and Dialogue. An Initiative of QatarDebate 2025).This paper is exploratory and aims to achieve virtuous decision-making in automated systems. Virtuous decision-making in automated systems is attainable by either programming agents with ethical frameworks or principles or by having a human in the loop. Programming AI agents with virtues which make them decide like humans is complex and challenging. The proposed approach is unique as it involves a human in the loop along with multiple agents which have different personas. The model has four agents (4 LLMs) with unique personas and a user (human) interacting with each other to reach a decision. The first input is provided by the user, and the agents simultaneously respond to the user's arguments accordingly to reach a decision. The agents engage in a form of deliberation or debate, where they respond to the user's arguments and provide counter arguments or supporting arguments. As an example, the deliberation model discussed in Scott Aikin and Caleb Clanton's paper on "Developing Group-Deliberative Virtues" is implemented with the user (human) introducing the "lost at sea" scenario. The agents are modelled as virtues like WittyOne, FriendlyOne, Temperate and Courageous, helping the user decide on the things to be chosen for survival at sea. This proposed model can be used to simulate different scenarios and experiments by programming the agents with different personas for virtuous automated decision-making.Other InformationConference information: The 2nd International Conference on Debate & Dialogue : An Initiative of QatarDebate (19 - 20 May 2025, Qatar National Conventions Center - QNCC, Doha - Qatar) License: https://creativecommons.org/licenses/by/4.0/See the conference information on the organizer's website: https://qatardebate.org/programs/academic-programs/2nd-icdd/
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".