Skip to content
Sprint projectJun 22, 2026Bogotá

Contestabilidad algorítmica en el Estado colombiano: un canal de objeción asistido por IA

Emely Condor, Federico Perez · Team FEVS

Submitted to Global South AI Safety Hackathon. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Contestabilidad algorítmica en el Estado colombiano: un canal de objeción asistido por IA

More on docs.google.com (opens in new tab)
Share

Los sistemas algorítmicos del Estado colombiano ya toman o asisten decisiones que afectan derechos, pero los canales para objetar esas decisiones casi no existen. Construimos un Índice de Contestabilidad Ciudadana que operacionaliza, artículo por artículo, la Directiva Conjunta 007 de 2025 de la Procuraduría y la Defensoría, y lo aplicamos a 409 sistemas de decisión automatizada del repositorio de la Universidad de los Andes; sistemas de alto impacto se verificaron contra las webs oficiales de cada entidad. El resultado central es una brecha medible: la transparencia informacional es alta (las entidades publican qué es el sistema, su objetivo y sus datos) mientras la contestabilidad es casi nula. Solo 6 de 409 sistemas ofrecen un canal de objeción completo, 391 no ofrecen ninguno, y ninguno de los 154 sistemas de alto riesgo publica un análisis de impacto algorítmico. A partir de esta evidencia proponemos: el diseño de un protocolo de canal de objeción asistido por IA que recibe el reclamo de un ciudadano, lo clasifica por tipo de sistema y lo enruta al responsable según la Directiva 007. La conclusión para la seguridad de la IA es que la contestabilidad debe diseñarse antes de que los agentes IA lleguen, no después.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Your private critique is not published; only the Apart team reads it. If you agree below, we share your review with grantmaking.ai (opens in new tab) and the Transformative AI Fund so strong projects can be funded.

Not shown on this page.

Shown on this page, without your name.

Only the Apart team reads this, and funders if you agree below.

Share my name publicly on grantmaking.ai *
Share my private critique with funders *

How much would this matter for AI safety if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical AI safety problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.

  1. This is a clear and policy-relevant project. Its most valuable contribution is the Contestability Index applied to 409 automated decision systems in the Colombian public sector. The finding is concrete and actionable. Public entities often disclose what systems exist and what they do, but almost never provide a meaningful channel for affected citizens to object. The contrast between informational transparency and practical contestability is the paper’s strongest insight, and the figures and tables communicate it effectively.

    The proposed AI-assisted objection channel is useful as a governance design pattern. The routing architecture (receiving a citizen complaint, identifying the relevant system, classifying it under risk categories, routing it to the correct institutional contact, and preserving traceability) is practical and well aligned with the paper’s diagnosis. The Appendix A proposal that contestability should “follow the function” when an AI agent replaces a human official is promising and could be developed into a model contractual clause for public procurement.

    The first limitation is methodological. For example, the core empirical claim depends heavily on whether missing published information is correctly interpreted as lack of contestability. It should be separated from the stronger claim that no objection channel exists in practice.

    The second limitation is the connection to AI safety. The paper’s immediate contribution is best understood as algorithmic accountability and public-sector redress. Its AI safety relevance comes from the future scenario in which autonomous agents replace human officials and make contestability structurally harder. That argument is plausible, but underdeveloped. The paper would be stronger if it specified what changes when the decision-maker is an AI agent rather than a conventional automated system. Right now, this is a compelling policy intuition rather than a fully developed safety argument.

    For future work, the most important next step is to build and test the objection-channel prototype with realistic or simulated citizen complaints, measuring relevant outcomes.

    Read full reviewShow less
  2. the paper addresses a critical safety problem, bravo. Methodology is also impressively thorough for a hackathon project, and it's easy to follow. The gap is that this pilot hasn't been used by "real users" limiting it's applicability (and very understandable given the timeframe) but it does slightly limit it's impact, still amazing work.

  3. Este es un proyecto muy sólido, con una contribución clara y directamente relevante para la gobernanza y seguridad de la IA en el sector público: medir la brecha entre transparencia algorítmica y contestabilidad efectiva. La construcción del Índice de Contestabilidad Ciudadana y su aplicación a 409 sistemas del Estado colombiano muestran una ejecución fuerte para un hackathon, especialmente porque el análisis combina codificación sistemática, verificación de casos de alto impacto y una propuesta práctica de canal de objeción asistido por IA. La conexión con agentes de IA es pertinente y valiosa, aunque en algunos momentos podría desarrollarse con mayor precisión técnica para distinguir mejor entre sistemas automatizados actuales y futuros agentes autónomos. La presentación es clara y convincente, con resultados fáciles de interpretar; para fortalecerlo aún más, sería útil profundizar en cómo se implementaría y evaluaría el prototipo del canal de objeción en un piloto real con una entidad pública.

    Read full reviewShow less

Cite this project

@misc{condor2026contestabilidad,
  title = {{Contestabilidad algorítmica en el Estado colombiano: un canal de objeción asistido por IA}},
  author = {Emely Condor and Federico Perez},
  year = {2026},
  month = jun,
  note = {Submitted to Global South AI Safety Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/contestabilidad-algortmica-en-el-estado-colombiano-un-canal-de-objecin-asistido-por-ia-lbsm}},
  url = {https://apartresearch.com/sprints/projects/contestabilidad-algortmica-en-el-estado-colombiano-un-canal-de-objecin-asistido-por-ia-lbsm}
}

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026