AI Credit Decisioning and Lending Model Risk Course – New York, 22–26 Mar 2027

AI Credit Decisioning and Lending Model Risk is a 1-week training course for bank managers, credit officers, risk professionals and lending operations leads who approve, monitor or challenge machine learning models in lending, and who leave with a Lending Model Risk Dossier. Banks now score applicants, set limits and flag suspicious behaviour with models that few decision makers can explain. Jupiter Prime Training runs this course on AI credit decisioning and lending model risk for the people who answer for those decisions.

Why AI Credit Decisioning and Lending Model Risk Matter Now

Three groups shape AI credit decisioning today: borrowers who expect instant decisions and fair treatment, supervisors and auditors who expect banks to explain and validate every model, and fintech competitors who approve loans in minutes. Lenders that manage model risk well grow faster with fewer defaults and complaints; those that do not face biased outcomes, unexplained declines and costly remediation. The same models also feed fraud screening and anti-money laundering alerts. The course draws on ISO/IEC 42001, the international standard for an AI management system, and the NIST AI Risk Management Framework (NIST AI RMF) to structure validation, monitoring and accountability.

Who Gains Most from AI Credit Decisioning and Lending Model Risk

• Credit officers and underwriting heads who approve scorecards and lending policy rules
• Model risk and validation staff who test, document and challenge models
• Retail and SME lending managers who own approval rates and portfolio quality
• Financial crime leads who tune fraud and anti-money laundering alert models
• Lending operations managers who run onboarding, servicing and collections workflows

A lending trio made up of a credit policy owner, a model validator and an operations lead gains most, because each model decision touches all three.

AI Credit Decisioning and Lending Model Risk Capabilities

Participants leave able to judge whether a lending model is sound, fair and explainable before it touches a customer.

• Assess machine learning credit scorecards against traditional logistic models — credit committees approving changes
• Test models for disparate impact and proxy variables — fair lending and conduct teams
• Apply explanation methods to produce adverse action reasons — declined applicants and complaint handlers
• Specify validation, monitoring and challenger model routines — model risk committees
• Calculate the business effect of cut-off changes on approvals and losses — lending heads
• Design human review points for fraud, anti-money laundering and service automation — operations and compliance leads

Key AI Credit Decisioning and Lending Model Risk Terms

• Credit scorecard — a model that converts applicant data into a score used to approve, price or decline credit.
• Probability of default — the estimated likelihood that a borrower will fail to repay within a set horizon.
• Model risk — the chance of loss or harm from a model that is wrong, misused or poorly understood.
• Disparate impact — an outcome that disadvantages a protected group even without intent, often through proxy variables.
• Adverse action reason — a plain explanation given to an applicant about why credit was declined or limited.
• Challenger model — an alternative model run alongside the live one to test whether it performs better.

AI Credit Decisioning and Lending Model Risk Day by Day

Day 1: AI Across the Lending Lifecycle

• Lending Value Chain Map — Onboarding, Scoring, Pricing, Servicing and Collections
• Traditional Versus Machine Learning Scorecards — Accuracy, Stability and Transparency
• Alternative Data Sources — Transaction, Behavioural and Open Data Signals
• Retail Mortgage and SME Lending Examples — Contrasting Model Needs
• Model Inventory and Tiering — Ranking Models by Materiality

Day 2: Building and Validating Credit Models

• Probability of Default and Loss Given Default Models — Inputs to Pricing
• Gini, KS and Calibration Measures — Judging Model Discrimination
• Out-of-Time and Out-of-Sample Testing — Detecting Overfitting Early
• Reject Inference Techniques — Correcting Bias From Declined Applicants
• Validation Report Structure — Evidence a Model Committee Expects

Day 3: Fairness and Explainability in Lending

• Disparate Impact Testing — Approval and Pricing Gaps Across Groups
• Proxy Variable Detection — Postcode, Device and Spending Pattern Risks
• SHAP and Reason Codes — Producing Adverse Action Explanations
• Cut-Off and Override Policy — Balancing Growth, Losses and Fairness
• Complaint and Appeal Handling — Human Review of Automated Declines

Day 4: Monitoring, Financial Crime and Operations

• Population Stability and Drift Monitoring — Triggers for Recalibration
• Fraud Detection and Anti-Money Laundering Alert Models — Tuning False Positives
• Customer Service Chatbots in Lending — Escalation and Disclosure Rules
• Back-Office Automation — Document Reading, Income Checks and Exceptions
• ISO/IEC 42001 and NIST AI RMF Controls — Model Ownership and Escalation

Day 5: Practising the Lending Model Risk Dossier

• Scorecard Challenge Session — Applying Day 2 Validation Measures to a Case Model
• Fairness Audit Exercise — Using Day 3 Disparate Impact Tests
• Monitoring Plan Drafting — Applying Day 4 Drift Triggers
• Credit Committee Simulation — Defending a Model Change With Evidence
• Lending Model Risk Dossier — Inventory, Validation, Fairness, Monitoring and Owners

Applying AI Credit Decisioning and Lending Model Risk After the Course

• Run a disparate impact test on one live scorecard and present the findings to the credit committee.
• Set population stability triggers for the highest-tier lending model and assign a named owner for each alert.

AI Credit Decisioning and Lending Model Risk Questions (FAQ)

What is AI credit decisioning and lending model risk and why does it matter now?

AI credit decisioning uses machine learning to approve, price and monitor loans, and lending model risk is the harm such models can cause. It matters now because faster automated lending raises expectations for fairness, explanation and validation.

Which roles gain most from AI credit decisioning and lending model risk training?

Credit officers, model validators, lending managers and financial crime leads gain most. If you approve scorecards or answer for declines, the course gives you tools to test and defend them.

How does AI credit decisioning and lending model risk differ from a general banking AI or fraud course?

AI credit decisioning and lending model risk centres on lending models, fairness and validation. A general banking AI course surveys many uses, and a fraud course targets detection; this one goes deep on credit decisions and their controls.

What do you bring back from AI credit decisioning and lending model risk training?

You bring back a model inventory approach, validation measures, fairness tests, explanation methods and a monitoring plan ready for your credit or model risk committee.

AI Credit Decisioning and Lending Model Risk: The Way Forward

The Lending Model Risk Dossier first informs whether the next scorecard change goes to production. Credit officers, validators and lending heads use it to agree evidence, fairness thresholds and owners. After the first monitoring cycle, the drift triggers deserve another look, since real portfolio shifts often reveal which variables move faster than expected.