Guidelines for Responsible AI
What you will learn
In this domain, you learn the guidelines for developing and operating AI systems in a responsible manner toward society and users. The goal is to understand the eight core dimensions of "responsible AI" that AWS advocates (fairness, explainability, privacy and security, safety, controllability, veracity and robustness, governance, transparency) and the implementation tools such as SageMaker Clarify and SageMaker Model Monitor.
Key points
- The eight core dimensions of responsible AI - fairness, explainability, privacy and security, safety, controllability, veracity and robustness, governance, transparency
- Bias - the problem of a model making unfair predictions for particular attributes (gender, race, etc.) due to skew in the training data or design
- Explainability - the ability to show the basis for a model's predictions in a form humans can understand. It relates to the persuasiveness of decisions
- Human-centered design - shows users the basis for AI decisions and uses mechanisms for receiving feedback to drive improvement
- SageMaker Clarify - a tool that detects bias in data and models and provides explainability through feature importance
- SageMaker Model Monitor - monitors whether a model in production exhibits behavior different from training time (data drift)
- Tools for transparency - SageMaker Model Cards (documentation), SageMaker Clarify (bias and explanations), Bedrock model evaluation, open-source models, and disclosure of data and licenses
- Privacy - it is important to design so that personal information is neither mixed into the training data nor output at inference time
- AWS AI Service Cards - cards that document the intended uses and limitations of each AWS AI service
Terms and concepts
The eight core dimensions of responsible AI
AWS organizes responsible AI into eight core dimensions: fairness (eliminating discrimination), explainability (explaining the basis for decisions), privacy and security (protecting personal information and preventing misuse), safety (curbing harmful outputs), controllability (mechanisms to monitor and steer AI behavior), veracity and robustness (accurate outputs and resilience to abnormal inputs), governance (organizational management), and transparency (disclosing capabilities and limitations). These are guidelines to keep in mind across all AI development.
Bias and fairness
When there is skew in the training data, "bias" occurs in which the model makes unfair predictions for particular attributes (gender, race, age, etc.). For example, if past hiring data has a high proportion of men, the model may evaluate men favorably. It is prevented by rebalancing the data, evaluating with fairness metrics, and reviewing the training data.
Explainability
The ability to show, in a form humans can understand, why a model made a given prediction. In loan screening or medical diagnosis, decisions cannot be accepted without being able to explain "why it was rejected / why that diagnosis." It is often explained with SHAP values or feature importance, and SageMaker Clarify is the primary AWS tool.
Human-centered design
Human-centered design for explainable AI emphasizes that the people who use AI can understand its results and notice and correct its mistakes. Show why the AI made a decision together with its basis and confidence, and make it clear to users that a result was produced by AI. Provide mechanisms for receiving user feedback, such as rating buttons on answers and input fields for corrections, and use the collected feedback to improve models and prompts.
SageMaker Clarify and Model Monitor
Clarify provides pre-training data bias detection, post-training model bias detection, and explainability through feature importance. Model Monitor continuously monitors whether a model in production exhibits behavior different from training time (data drift or prediction drift). Both are core tools for implementing responsible AI. Note that, as of October 2026, SageMaker Clarify no longer accepts new users, and only existing users can use it.
Tools for transparency and explainability
The transparency of a model is checked by what has been documented and disclosed about it. SageMaker Model Cards are documents that record a model's intended uses, risk rating, training details, and evaluation results in one place. SageMaker Clarify is a tool that produces bias metrics and reports showing which features influenced predictions, and Amazon Bedrock model evaluation is a tool that keeps the results of measuring the quality of a foundation model's answers. Open-source models are easier to inspect to the extent that their weights, training data, and license terms are published, and the license also determines the scope in which you may use them.
→ Amazon SageMaker Glossary / → ML Development with SageMaker
AWS AI Service Cards
AWS publishes an "AI Service Card" for each AI service, describing its intended uses, limitations, and guidelines for responsible use. For example, Rekognition's face comparison feature explicitly states limitations such as "not recommended for law enforcement use." These are important documents to review before adopting a service.