IBM named a Leader

See why in The Forrester Wave™: Multimodal Predictive Analytics and Machine Learning, Q3 2020.

Overview

What is AI model risk management?

The Federal Reserve and Office of the Comptroller of the Currency guidance SR Letter 11-7 (link resides outside IBM) defines a model as "…a quantitative method, system, or approach that applies statistical, economic, financial, or mathematical theories, techniques, and assumptions to process input data into quantitative estimates."

Model risk can occur when a model is used to predict and measure quantitative information but the model performs inadequately. Poor model performance can lead to adverse outcomes and result in substantial operational losses. Implementing model risk management in a modern information architecture helps you:

  • Speed time to help meet regulatory compliance and other risk objectives.
  • Simplify model validation across multiple clouds.
  • Take advantage of models and data running virtually anywhere.

Learn more

Five ways to simplify model risk management

Enhance model compliance with custom tests and thresholds.

Podcast: KPMG-IBM on AI

Listen to AI experts discuss digitization of governance in the age of AI.

Product images

Risk model evaluations

Screen shot showing risk model evaluations, including fairness, quality and drift metrics

Risk model evaluations

Show fairness, quality and drift metrics. Flag models below custom thresholds. Drill down for details.

Fairness

Screen shot showing the details of the fairness metric for a credit risk model

Fairness

Configure and perform model validation. Test model metrics including model fairness.

Model comparison

Screen shot showing the comparison of two models based on fairness and quality

Model comparison

Compare model test results. Select and speed the development of more effective models.

Metrics summary

Screen shot showing metric details for a credit risk model

Metrics summary

Generate a fact sheet in PDF automatically. Summarize model details, relevant data and test results.

Get started with explainable AI

Explore model monitoring and model management in IBM Watson Studio.