Projects

Open Credit Scoring is developing open research, technical frameworks, and governance models for trustworthy AI-driven credit and underwriting systems.

Our projects focus on advancing causal AI, transparency, fairness, and institutional trust in high-stakes financial decision-making.

Open Credit Scoring Initiative

The Open Credit Scoring Initiative explores how open standards, causal AI, and transparent governance can help modernize credit scoring and underwriting systems.

The initiative aims to create an open technical foundation for next-generation financial AI systems.

Key focus areas include:

Transparent
Explainable
Accountable
Interoperable
Safe by design
Open governance frameworks
Causal credit scoring architectures
Fairness and antidiscrimination analysis
Alternative data evaluation
Trustworthy AI infrastructure
Long-term institutional accountability

Causal AI for Credit Underwriting

This project explores how causal inference and causal Bayesian networks can augment traditional machine learning systems used in credit underwriting.

Rather than relying solely on statistical correlations, causal models attempt to explicitly represent relationships between key financial and decision variables.

This work investigates how causal AI may help financial institutions better distinguish between legitimate predictive signals and potentially discriminatory effects.

Creditworthiness
Financial behavior
Alternative data
Protected attributes
Lending decisions
Loan outcomes
Explainability
Fairness analysis
Regulatory alignment
Policy control
System stability

Causal Fairness and Antidiscrimination Research

This research project focuses on applying causal models to questions of fairness, discrimination, and compliance in financial AI systems.

The project also explores how causal models may provide a more operational framework for translating antidiscrimination principles into system design constraints.

Disparate treatment
Disparate impact
Proxy discrimination
Mediation analysis
Counterfactual reasoning
Causal debiasing
Fair lending compliance

Alternative Data and Causal Debiasing

Alternative data has the potential to improve financial inclusion, but it also introduces new risks related to proxy discrimination, digital redlining, and opaque decision-making.

This project explores how causal inference techniques may help evaluate alternative data sources more rigorously.

The goal is to develop more trustworthy approaches for evaluating whether alternative data has a legitimate causal relationship to creditworthiness.

Education data
Employment data
Behavioral data
Digital footprint data
Natural language signals
Psychometric indicators

Systems Thinking and Financial AI

Most AI systems are optimized primarily for short-term prediction.

This project explores how systems thinking and system dynamics can help model long-term feedback effects within financial systems.

The objective is to move beyond static AI models toward systems capable of simulating long-term economic and social outcomes.

Feedback loops
Institutional incentives
Dynamic discrimination effects
Financial inclusion dynamics
Consumer behavior adaptation
Policy intervention modeling

Open Standards for Trustworthy AI

Open Credit Scoring supports the development of open technical standards for high-stakes AI systems.

We believe open standards are essential for creating trusted and interoperable AI ecosystems in finance.

AI governance frameworks
Explainability standards
Causal model interoperability
Fairness testing methodologies
Regulatory alignment
Transparent AI infrastructure

Future Directions

Open Credit Scoring is exploring future research directions including:

Open-source reference implementations
Benchmark datasets and evaluation frameworks
Public-interest AI governance models
AI safety for financial systems
Human-centered underwriting systems
Simulation environments for financial AI
Causal decision models for socioeconomic systems

Financial AI needs more than better prediction

We believe the future of financial AI depends not only on better prediction, but on better governance, transparency, and institutional trust.