Artificial Intelligence Governance in Public Services to Accelerate Poverty Alleviation: Accountability Model and Oversight Mechanism for Indonesia
Bibliographic record
Abstract
Background: Indonesian public administration is accelerating the use of artificial intelligence (AI) to improve social protection targeting. The legal and digital governance foundations are in place through SPBE and Satu Data regulations, complemented by the Data Protection Law, Public Information Law, and the PSTE regulation. Recent policy integrates Regsosek, DTKS and P3KE into a single national data basis (DTSEN) for beneficiary determination. Objective: This paper proposes an operational model of algorithmic accountability and multi?layer oversight for AI used in public services to accelerate poverty alleviation while protecting fundamental rights. Methods: A normative?doctrinal approach augmented by a targeted socio?legal case mapping is used, combining legal gap analysis and design?science techniques. The model is benchmarked against global practices (EU AI Act and FRIA, Canada’s AIA/DADM, the Netherlands’ Algorithm Register, the UK ATRS, and NIST AI RMF). Results: We outline a three?stage governance architecture—ex ante (mandatory AIA+FRIA for high?risk systems; legality, data quality and bias testing), in?process (human?in?the?loop at decision thresholds; logging/versioning; explainability), and ex post (reason?giving and appeal; algorithm register; periodic audits)—tightly linked to SPBE/Satu Data controls and the DTSEN pipeline. During the transition to a full PDP authority, external oversight is bridged by the Ombudsman and the Information Commission. Conclusion: The proposed model operationalizes administrative due process for AI?assisted decisions, strengthens transparency and accountability, and is expected to reduce inclusion/exclusion errors and improve exit and persistence rates above the poverty line. A 12?month implementation roadmap and measurable indicators are provided.
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 | 0.007 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.005 |
| Scholarly communication | 0.007 | 0.005 |
| Open science | 0.001 | 0.004 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.005 | 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".