Free AIF-C01 Exam Dumps Preview

BLEU evaluates surface-level n-gram overlap between the LLM's outputs and reference examples--in this case, targets written in creative spelling and shorthand, making it well suited to measure how closely the model's style matches the teenager-oriented language.

Bilingual Evaluation Understudy measures how closely a generated translation matches one or more high-quality reference translations, making it well suited for evaluating the quality of multilingual summary translations.

Prompt management in Amazon Bedrock lets you create, save, version, and reuse prompts, including system instructions and task context, across repeated interactions.

Explainable AI techniques provide transparency by identifying and displaying the specific factors that influenced the generative AI model's credit limit decisions, making the decision-making process understandable to customers.

Amazon Q Business is designed to build generative AI assistants for querying and interacting with internal organizational data, making it the ideal service for creating an AI assistant for employees to access company information.

Moderation APIs, such as Amazon Rekognition's Content Moderation API, can help filter and block inappropriate or unwanted images from being returned by a chatbot. These APIs are specifically designed to detect and manage undesirable content in images.
Option A (Correct): "Implement moderation APIs":This is the correct answer because moderation APIs are designed to identify and filter inappropriate content, ensuring the chatbot does not return unwanted images.
Option B:"Retrain the model with a general public dataset" is incorrect because retraining does not directly prevent inappropriate content from being returned.
Option C:"Perform model validation" is incorrect as it ensures model correctness, not content moderation.
Option D:"Automate user feedback integration" is incorrect because user feedback does not prevent inappropriate images in real-time.

References:
AWS Content Moderation Services:AWS provides moderation APIs for filtering unwanted content from applications.

Assigning categories such as "personal" or "business" to financial transaction records is an example of data labeling, where each data point is tagged with a specific label for use in machine learning or data analysis tasks.

Amazon Kendra is an intelligent enterprise search service that enables organizations to search large volumes of unstructured data, such as educational materials, making it ideal for enterprise search needs.

Few-shot prompting involves providing the generative AI model with several examples (drawn from public bank documents) to guide it in generating more accurate and relevant responses. This technique helps the chatbot better align its answers with the desired content and format.

Provisioned Throughput is priced lower per request when you have a predictable, steady volume of calls.
By committing to a fixed throughput level, you secure the necessary capacity at a reduced unit cost compared to on-demand, making it the most cost-effective choice for steady-minute usage.

The "temperature" parameter controls the randomness and diversity of outputs generated by a large language model. Increasing the temperature produces more diverse and creative responses, while lowering it makes the outputs more focused and deterministic.

"Classify the following text as either sports, politics, or entertainment: [input text]." Correct Answer: Zero-shot learning Zero-shot learning involves providing the model with a task and no prior examples, relying entirely on the model's pre-trained knowledge to perform the classification. "A [image 1], [image 2], and [image 3] are examples of [target class]. Classify the following image as [target class]."
Few-shot learning provides the model with a few examples (image 1, image 2, and image 3) before asking it to classify a new instance, helping it generalize to the task with minimal examples.

"[Question.] [Instructions to follow.] Think step by step and walk me through your thinking." Correct Answer: Chain-of-thought reasoning Chain-of-thought reasoning encourages the model to break down its reasoning process step by step, enhancing its ability to solve complex tasks logically.

Amazon SageMaker Unified Studio provides a collaborative, centralized environment for data and AI development where multiple teams can securely build, share, and manage models and generative AI applications.

Clustering can group similar pixel patterns or image regions without labeled data, allowing the model to identify areas that share visual characteristics associated with culture growth.

Amazon OpenSearch Service (formerly Amazon Elasticsearch Service) has introduced capabilities to support vector search, which allows companies to build vector database applications. This is particularly useful in machine learning, where vector representations (embeddings) of data are often used to capture semantic meaning. Scalable index management and nearest neighbor search capabilityare the core features enabling vector database functionalities in OpenSearch. The service allows users to index high-dimensional vectors and perform efficient nearest neighbor searches, which are crucial for tasks such as recommendation systems, anomaly detection, and semantic search.
Here is why option C is the correct answer: Scalable Index Management:OpenSearch Service supports scalable indexing of vector data. This means you can index a large volume of high-dimensional vectors and manage these indexes in a cost-effective and performance-optimized way. The service leverages underlying AWS infrastructure to ensure that indexing scales seamlessly with data size.
Nearest Neighbor Search Capability:OpenSearch Service's nearest neighbor search capability allows for fast and efficient searches over vector data. This is essential for applicationslike product recommendation engines, where the system needs to quickly find the most similar items based on a user's query or behavior.

References:
The other options do not directly relate to building vector database applications: Option
A. Integration with Amazon S3 for object storageis about storing data objects, not vector-based searching or indexing. Option
B. Support for geospatial indexing and queriesis related to location-based data, not vectors used in machine learning. Option
D. Ability to perform real-time analysis on streaming datarelates to analyzing incoming data streams, which is different from the vector search capabilities.

Amazon Lex and Amazon Polly are the correct answers because Lex can build conversational interfaces and Polly can convert the chatbot's text responses into speech. Together they support a voice-enabled chatbot experience. Option A (Correct): "Amazon Lex": This is correct because it supports conversational interfaces and can process user utterances.
Option B (Correct): "Amazon Polly": This is correct because it converts generated text responses into speech.
Option C: "Amazon Macie" is incorrect because it is used for sensitive data discovery and protection.
Option D: "AWS Artifact" is incorrect because it provides compliance documentation.
Option E: "Amazon S3 Glacier" is incorrect because it is archival storage and not part of voice interaction.

AWS CloudTrail records and logs all API calls and user actions in an AWS account, providing a complete audit trail of activities performed by each AI practitioner for compliance and audit purposes.

Automatically analyzing and classifying free-text feedback into thematic categories is a core NLP task (text classification).

Human evaluation with custom prompts is the correct answer because usefulness is a qualitative requirement that depends on the preferences and needs of the company's users. Human reviewers are best suited to assess that kind of output quality. Option A (Correct): "Human evaluation with custom prompts": This is correct because people can directly assess usefulness and quality.
Option B: "Only measure invocation latency" is incorrect because speed does not measure summary usefulness.
Option C: "Only compare the number of model parameters" is incorrect because size alone does not determine output quality.
Option D: "Only increase the context window" is incorrect because that is not an evaluation method.

Amazon SageMaker Model Cards provide a centralized and standardized repository for documenting machine learning models. They capture key details such as the model's intended use, training and evaluation datasets, performance metrics, ethical considerations, and other relevant information. This documentation facilitates governance and reporting by ensuring that all stakeholders have access to consistent and comprehensive information about each model. While Amazon SageMaker Debugger is used for real-time debugging and monitoring during training, and Amazon SageMaker Model Monitor tracks deployed models for data and prediction quality, neither offers the comprehensive documentation capabilities of Model Cards. Amazon SageMaker JumpStart provides pre-built models and solutions but does not focus on governance documentation.

References:
Amazon SageMaker Model Cards

The data collection phase of the ML lifecycle is the most relevant for determining compliance and regulatory requirements. During this phase, organizations must ensure that the data being collected and used for training the model complies with legal and regulatory standards, such as data privacy laws (e.g., GDPR, HIPAA), industry-specific regulations, and ethical considerations. The organization must also verify that they have the proper consent to use the data and that the data does not contain any biases or violate any regulations.

Amazon Augmented AI (Amazon A2I) provides built-in workflows for adding human review to AI processes, including generative AI content. When a model generates product descriptions, A2I can route outputs for human approval before publishing. This ensures quality control and brand compliance and fits scenarios where human validation is required in a content generation pipeline.

Increasing the temperature is the correct answer because higher temperature values generally make model outputs more diverse and less deterministic. This is useful when a business wants multiple creative variations. Option B (Correct): "Increase the temperature": This is correct because a higher temperature
encourages more varied generations.
Option A: "Decrease the temperature" is incorrect because it usually makes outputs more predictable and repetitive.
Option C: "Decrease the number of input tokens" is incorrect because that does not directly control creativity.
Option D: "Use batch inference" is incorrect because batch inference relates to processing mode, not output diversity.

Amazon SageMaker Model Monitor continuously tracks deployed ML models in production to detect data drift, model drift, and quality degradation over time. This is essential for ensuring that the recommendation model remains accurate as viewer behavior and demographics change. Model Monitor helps detect anomalies and provides alerts when model performance deviates from expected trends, allowing the company to take corrective action.

Providing direct links to the exact sections of the product manual that support each answer lets customers verify and trust the information, boosting confidence in the AI agent's responses.

Nova Lite is a multimodal model in Amazon Bedrock that supports multiple languages and is designed to be the most cost-effective option among the Nova models, making it suitable for organizations seeking efficient, scalable, and economical AI solutions.

Training on bias awareness and responsible AI equips the development team with the knowledge and practices necessary to ensure that AI systems are fair, unbiased, and explainable, which aligns with the company's requirements.

AWS Artifact provides on-demand access to AWS's compliance reports and certifications (for example, ISO, SOC, GDPR), enabling the company to demonstrate its generative AI workloads and data handling practices adhere to international regulatory requirements.

An AWS Lambda function can efficiently perform numerical image transformations such as transposing and rotating images at scale without the need to build or train a neural network, making it the most operationally efficient solution.

For real-time applications, the model's inference speed - ability to generate responses with low latency - is the critical characteristic to ensure users receive answers promptly.

AWS PrivateLink enables private, internal-only connectivity to Bedrock APIs, ensuring no traffic traverses the public internet. This is essential for secure enterprise deployments.

Different large language models have varying maximum token limits, which affects how much input and output text a prompt can include, so prompts must account for these differences to work consistently across models.

AWS PrivateLink is the correct answer because it enables private connectivity from a VPC to supported AWS services without exposing traffic to the public internet.
Option B (Correct): "AWS PrivateLink": This is correct because it provides private network access to supported services.
Option A: "Internet gateway" is incorrect because it enables internet connectivity.
Option C: "Amazon CloudFront" is incorrect because it is a content delivery service.
Option D: "Amazon Route 53" is incorrect because it is a DNS service.

Amazon Bedrock operates regionally and, combined with PrivateLink, ensures all data stays within the customer's VPC and Region. This satisfies data residency and compliance requirements.

To generate concise, feature-specific descriptions for each product, the company should create prompts tailored to specific product categories. By highlighting the key features of each product category in the prompt, the model can focus on generating descriptions that are relevant and aligned with the unique attributes of each product. Additionally, specifying the desired output format and length ensures that the responses meet the company's requirements for conciseness and clarity.

Tailored prompts help ensure the model generates relevant and accurate descriptions by focusing on the most important features of each product category.

Desired output format and length ensure the responses are consistent and concise, as required.

Neural networks typically provide higher predictive performance but are less transparent than decision trees, resulting in reduced interpretability of how decisions are made.

To manage the flow of data from Amazon S3 to SageMaker Studio notebooks securely, using a VPC with an S3 endpoint is the best solution.

Using a prebuilt AI assistant allows the company to deploy generative AI quickly without the time and cost of training custom models. This accelerates implementation, enabling measurable business impact such as improved customer engagement and increased revenue within the short 6-month timeframe.

Amazon SageMaker Model Cards provide a standardized way to document important details about an AI model, such as its purpose, performance, intended usage, and known limitations. This enables transparency and compliance while fostering better communication between stakeholders. It does not store models physically or optimize computational requirements.

References:
AWS SageMaker Model Cards Documentation.

Binary classification is designed to separate data into two classes -- in this case, urgent vs. non-urgent. It uses labeled data to learn patterns distinguishing the two categories, aligning perfectly with the company's needs.

Amazon Augmented AI (Amazon A2I) enables you to build workflows for human review of machine learning predictions, allowing you to define and adjust confidence thresholds for when human intervention is required.

Use Amazon Transcribe and Amazon Translate to generate subtitles in other languages: Amazon Transcribe converts spoken dialogue to text (subtitles), and Amazon Translate can then translate these subtitles into multiple languages.

Use Amazon Polly to generate voice-overs in other languages: Amazon Polly converts translated text into lifelike speech, enabling the creation of multilingual voice-overs.

Sampling bias is the correct type of bias affecting the model output when it disproportionately flags people from a specific ethnic group.

Presenting the model Shapley values provides transparency and explainability by showing how each feature contributes to individual predictions, helping stakeholders understand the reasoning behind the model's outputs.

Few-shot prompting provides examples that teach the model how to respond in a specific domain or style.
This improves accuracy and relevance without needing fine-tuning.

Amazon SageMaker AI provides full training and fine-tuning capabilities for foundation models using custom data, and the resulting trained model can then be imported and deployed in Amazon Bedrock for internal use.

The Amazon Nova Reel model on Amazon Bedrock is specifically designed for generative video creation, enabling the company to efficiently generate new creative content and significantly reduce video creation time in an operationally efficient manner.

Embedding specific input-output examples directly into the agent's advanced prompt (few-shot prompting) guides the model toward more accurate behavior without retraining or additional tooling.

Batch inference is designed for high-volume, scheduled predictions: you can point it at the day's gigabytes of data, run a job once daily, and generate all forecasts in one go without needing a persistent endpoint.

A RAG architecture is the least-effort solution because it uses the existing product catalog as the knowledge source at runtime, so the foundation model can answer questions with current product information without requiring extensive labeled data or custom model training.

Tokens in generative AI models are the smallest units that the model processes, typically representing words, subwords, or characters. They are essential for the model to understand and generate language, breaking down text into manageable parts for processing.
Option A (Correct): "Tokens are the basic units of input and output that a generative AI model operates on, representing words, subwords, or other linguistic units":This is the correct definition of tokens in the context of generative AI models.
Option B:"Mathematical representations of words" describes embeddings, not tokens.
Option C:"Pre-trained weights of a model" refers to the parameters of a model, not tokens.
Option D:"Prompts or instructions given to a model" refers to the queries or commands provided to a model, not tokens.

References:
Understanding Tokens in NLP:AWS provides detailed explanations of how tokens are used in natural language processing tasks by AI models, such as in Amazon Comprehend and other AWS AI services.

Reinforcement learning from human feedback allows the model to continuously learn from real-time input provided by skilled moderators, aligning it with company-specific values and ethics while adapting to emerging trends in content.

The evaluation stage is when you run tests and benchmarks, such as accuracy, precision, and other performance metrics, to measure how well the generative AI model performs on hold-out or validation data before moving on to deployment.

Unsupervised learning is the correct answer because the data is unlabeled and the goal is to discover natural groupings in the comments. Clustering is a common unsupervised technique for this type of task.
Option B (Correct):
"Unsupervised learning": This is correct because it can identify patterns and group similar data without predefined labels.
Option A: "Supervised learning" is incorrect because supervised learning requires labeled examples.
Option C: "Reinforcement learning" is incorrect because it focuses on learning from rewards and actions over time.
Option D: "Transfer learning" is incorrect because it is a model adaptation strategy, not the core technique for grouping unlabeled comments.

RAG enhances a language model by fetching up-to-date, domain-specific data (e.g., current inventory) at inference time and conditioning the generation on those facts, ensuring the output reflects the latest information and improving accuracy without retraining the core model.

The model is being trained from randomly initialized weights on a large corpus using a general language modeling objective, which corresponds to the pre-training phase where foundational knowledge is learned.

Apply human feedback across the ML lifecycle to improve the accuracy and relevancy of models. ->
Amazon SageMaker Ground Truth
Implement safeguards that align with responsible AI policies. -> Amazon Bedrock Guardrails Detect potential bias during data preparation and model training. -> Amazon SageMaker Clarify

Amazon SageMaker Ground Truth is used for human-in-the-loop data labeling and model improvement.

Amazon Bedrock Guardrails are designed to enforce responsible AI safeguards. Amazon SageMaker Clarify detects and helps mitigate bias during data preparation and model training.

Balancing class representation ensures that all groups in the dataset are adequately represented, which helps the model learn patterns across diverse users and reduces bias.

System prompts are used to establish the role, tone, rules, and behavioral constraints of the AI, ensuring it responds in a controlled and consistent manner aligned with the application's intended purpose.

AWS CloudTrail is the AWS service designed for logging and monitoring API calls made to AWS services, including Amazon Bedrock. CloudTrail records detailed information about the API requests, including the identity of the requester, the time of the request, and the source IP address. This service is ideal for logging all requests made to the Amazon Bedrock API and meets the logging requirement.

Amazon S3 Intelligent-Tiering is a storage class designed for storing data that has unpredictable access patterns. It automatically moves data between two access tiers (frequent and infrequent) based on usage, which helps reduce costs while ensuring data is still available when needed. For retaining logs securely over 5 years at the lowest possible cost, this storage class provides an efficient way to handle long-term storage requirements without incurring unnecessary costs.

Amazon SageMaker provides a fully managed environment for building, training, and deploying machine learning models, including automatic model tuning. Specifically, SageMaker includes a feature called Automatic Model Tuning (or Hyperparameter Optimization), which automates the process of finding the best hyperparameters for your machine learning model. This is essential when you want to optimize the model's performance without manual intervention.

Amazon SageMaker allows you to automate the training process and hyperparameter tuning, which aligns perfectly with the company's need for fully automated model tuning.

Create high-quality images that are influenced by the generated slogans and product -> Diffusion model
Create contextually relevant slogans based on the advertisement product -> Transformer-based model
Ensure that company brand elements are properly placed in the images -> Object detection model

Diffusion models are state-of-the-art for high-quality image generation.

Transformer-based models excel at natural language processing, making them suitable for generating contextually relevant slogans.

Object detection models are designed to identify and ensure correct placement of specific objects (e.g.,
brand elements) within images.

A RAG workflow with Amazon Bedrock Knowledge Bases allows the chatbot to access the most recent company policy documents dynamically, ensuring responses reflect policy updates in near real time without the need to retrain or fine-tune the LLM each time content changes.

A large multi-modal language model can natively ingest both text and images as inputs and generate text outputs, making it ideal for a system that accepts typed questions or photos of questions and returns written answers with explanations.

In a RAG solution, documents are split into chunks and converted into embeddings so they can be indexed and retrieved based on semantic similarity to user queries.

Encrypt the application data, and isolate the application on a private network -> Privacy and security
Evaluate how different population groups will be impacted -> Fairness Test the application with unexpected data to ensure the application will work in unique situations -> Robustness

Protecting data and networks is part of privacy and security.

Assessing impacts on different groups ensures fairness.

Testing with unexpected data checks the application's robustness.

The company is actively seeking to ensure that the image classification model is trained on a diverse dataset that includes insect bite photos from various genders, ethnicities, and geographic locations. This reflects the fairness principle of responsible AI, which emphasizes creating models that make unbiased decisions across all demographic groups. By including a diverse range of data, the company is aiming to prevent biases that could lead to inaccurate diagnoses or treatments for certain groups of people.

Fairness ensures that AI systems do not discriminate based on race, gender, geography, or other characteristics.

ROUGE is a set of metrics that evaluates the quality of summaries by comparing the overlap of n-grams, word sequences, and word pairs between the model output and reference examples. Since the company is working on a language model that improves the coherence of text and wants the output to resemble the provided examples (which are more readable versions of the original text), ROUGE is the most appropriate metric. It assesses how closely the generated text matches the reference text in terms of content and readability.

ROUGE score is commonly used to evaluate the performance of models in tasks like summarization, where the goal is to ensure the generated text aligns closely with human-provided examples.

Supervised learning is used when the dataset includes both input features (like age, cholesterol, blood pressure, etc.) and a target value indicating the presence of heart disease. The model learns to predict the target value from labeled examples.

Exploratory data analysis (EDA) involves understanding the data by visualizing it, calculating statistics, and creating correlation matrices. This stage helps identify patterns, relationships, and anomalies in the data, which can guide further steps in the ML pipeline.
Option C (Correct): "Exploratory data analysis":This is the correct answer as the tasks described (correlation matrix, calculating statistics, visualizing data) are all part of the EDA process.
Option A:"Data pre-processing" is incorrect because it involves cleaning and transforming data, not initial analysis.
Option B:"Feature engineering" is incorrect because it involves creating new features from raw data, not analyzing the data's existing structure.
Option D:"Hyperparameter tuning" is incorrect because it refers to optimizing model parameters, not analyzing the data.

References:
Stages of the Machine Learning Pipeline:AWS outlines EDA as the initial phase of understanding and exploring data before moving to more specific preprocessing, feature engineering, and model training stages.