Machine Learning
Read interview
What Banks Actually Find When They Deploy Explainable ML
An honest look at what explainable machine learning delivers in banking operations, where it genuinely helps, and where the gaps still exist.
AI Compliance
Read interview
Loan Decisions, Black Boxes, and the Compliance Problem Banks Cannot Ignore
A frank examination of how explainable AI affects loan decisioning workflows and what compliance teams are actually asking for.
Model Governance
Read interview
How Model Risk Teams Are Evaluating Explainability in 2024
An inside look at the criteria model risk management teams at banks are using to assess whether ML explainability meets internal governance standards.
Fraud Detection
Read interview
Fraud Detection Models and the Explainability Trade-Off Banks Face
An analytical look at why explainability requirements create a genuine tension in fraud detection, and how banks are navigating the trade-off.
SHAP Most discussed explainability method across all interviews
LIME Local surrogate approach cited for credit decision audits
SR 11-7 Regulatory guidance most referenced by practitioners
Interview Themes
What each discussion covers — at a glance
The interviews span regulatory, technical, and organizational angles. The table below maps key themes across dimensions that matter most to compliance teams and model risk officers.
| Theme | Regulatory Relevance | Technical Depth | Practitioner Focus |
|---|---|---|---|
| Model transparency requirements | High | Medium | High |
| SHAP value interpretation in credit scoring | Medium | High | High |
| Audit trail and documentation | High | Low | Medium |
| Fairness and bias detection | High | High | Medium |
| Operationalizing XAI in production | Medium | High | High |
About these interviews
How Domain selects and structures each conversation
Each interview goes through an editorial process before publication. Speakers are working professionals — risk analysts, ML engineers, compliance leads — not generalists summarizing secondary sources. Questions are prepared with input from participants to keep the exchange grounded in current practice rather than theory.
The format is structured but not scripted. Participants are encouraged to flag where common assumptions break down, where tooling falls short, and what regulators actually check during model audits.
- Speakers have direct hands-on experience with XAI in regulated environments
- Each interview references specific tools, frameworks, or regulatory documents
- Editorial review checks for accuracy before publication