AWS AI CERTIFICATION DOMAIN 4&5 WITH COMPLETE SOLUTIONS 100% VERIFIED
LATEST UPDATE!!
Fairness - (ANSWER)Ensuring AI models don't discriminate against or unfairly disadvantage certain
groups of people based on characteristics like race, gender, age, or socioeconomic status
Explainability - (ANSWER)The ability to understand and explain why an AI model made a specific decision
or prediction in terms humans can understand
Privacy & Security - (ANSWER)Protecting sensitive personal data used to train or run AI models from
unauthorized access, breaches, or misuse, and ensuring models don't inadvertently reveal private
information about individuals
Safety - (ANSWER)Ensuring AI systems don't cause physical, psychological, or societal harm when
deployed in the real world
Controllability - (ANSWER)The ability for humans to oversee, intervene in, or override AI decisions when
needed, maintaining human agency and preventing autonomous systems from taking actions outside
intended boundaries
Veracity & Robustness - (ANSWER)Ensuring AI models produce accurate, reliable results consistently
across different conditions and don't break or give wildly incorrect answers when faced with unexpected
inputs or adversarial attacks
Governance - (ANSWER)The policies, processes, and organizational structures for managing AI
development and deployment responsibly, including who's accountable, how decisions are made, and
how compliance is ensured
Transparency - (ANSWER)Making information about how AI systems work, what data they use, their
limitations, and their decision-making processes accessible and understandable to users and
stakeholders
Guardrails - (ANSWER)Technical controls and safety mechanisms built into AI systems to prevent
harmful, inappropriate, or out-of-scope outputs and behaviors
, AWS AI CERTIFICATION DOMAIN 4&5 WITH COMPLETE SOLUTIONS 100% VERIFIED
LATEST UPDATE!!
Types of AI Safeguards - (ANSWER)Content Filters: Protect and block harmful content and predefined
categories
Denied Topics: Block specific topics for app not to talk about
Work Filter: Block exact words, certain phrases, competitor names
Sensitive Information Blocks: Mask PII
Grounding Checks: Filters out hallucinations
Reasoning Checks: Making sure answers are logical, detect hallucinations
Underfitting - (ANSWER)Model performs poorly on training data, unable to capture the relationship
between input examples and target examples
Overfitting - (ANSWER)Model performs well on training data but not on the evaluation data because the
model memorizes the data it has seen and cannot generalize to unseen examples (training and test data
doesn't sync up)
AI Governance: Inception - (ANSWER)Use AI to solve a problem that has been identified, not to identify
the problem. The latter is costly.
AI Governance: Design & Development - (ANSWER)Define system architectures, data flows, and training
models
AI Governance: Verification & Validation - (ANSWER)Once you have a model, pull in production data
after the model has been trained. Verify that it's working the right way.
AI Governance: Deployment - (ANSWER)Put the model out into the world and enable people to use it.
AI Governance: Operation & Monitoring - (ANSWER)Run the system, log activity, monitor performance
and outcomes
LATEST UPDATE!!
Fairness - (ANSWER)Ensuring AI models don't discriminate against or unfairly disadvantage certain
groups of people based on characteristics like race, gender, age, or socioeconomic status
Explainability - (ANSWER)The ability to understand and explain why an AI model made a specific decision
or prediction in terms humans can understand
Privacy & Security - (ANSWER)Protecting sensitive personal data used to train or run AI models from
unauthorized access, breaches, or misuse, and ensuring models don't inadvertently reveal private
information about individuals
Safety - (ANSWER)Ensuring AI systems don't cause physical, psychological, or societal harm when
deployed in the real world
Controllability - (ANSWER)The ability for humans to oversee, intervene in, or override AI decisions when
needed, maintaining human agency and preventing autonomous systems from taking actions outside
intended boundaries
Veracity & Robustness - (ANSWER)Ensuring AI models produce accurate, reliable results consistently
across different conditions and don't break or give wildly incorrect answers when faced with unexpected
inputs or adversarial attacks
Governance - (ANSWER)The policies, processes, and organizational structures for managing AI
development and deployment responsibly, including who's accountable, how decisions are made, and
how compliance is ensured
Transparency - (ANSWER)Making information about how AI systems work, what data they use, their
limitations, and their decision-making processes accessible and understandable to users and
stakeholders
Guardrails - (ANSWER)Technical controls and safety mechanisms built into AI systems to prevent
harmful, inappropriate, or out-of-scope outputs and behaviors
, AWS AI CERTIFICATION DOMAIN 4&5 WITH COMPLETE SOLUTIONS 100% VERIFIED
LATEST UPDATE!!
Types of AI Safeguards - (ANSWER)Content Filters: Protect and block harmful content and predefined
categories
Denied Topics: Block specific topics for app not to talk about
Work Filter: Block exact words, certain phrases, competitor names
Sensitive Information Blocks: Mask PII
Grounding Checks: Filters out hallucinations
Reasoning Checks: Making sure answers are logical, detect hallucinations
Underfitting - (ANSWER)Model performs poorly on training data, unable to capture the relationship
between input examples and target examples
Overfitting - (ANSWER)Model performs well on training data but not on the evaluation data because the
model memorizes the data it has seen and cannot generalize to unseen examples (training and test data
doesn't sync up)
AI Governance: Inception - (ANSWER)Use AI to solve a problem that has been identified, not to identify
the problem. The latter is costly.
AI Governance: Design & Development - (ANSWER)Define system architectures, data flows, and training
models
AI Governance: Verification & Validation - (ANSWER)Once you have a model, pull in production data
after the model has been trained. Verify that it's working the right way.
AI Governance: Deployment - (ANSWER)Put the model out into the world and enable people to use it.
AI Governance: Operation & Monitoring - (ANSWER)Run the system, log activity, monitor performance
and outcomes
