A research company implemented a chatbot by using a foundation model (FM) from Amazon Bedrock. The chatbot searches for answers to questions from a large database of research papers. After multiple prompt engineering attempts, the company notices that the FM is performing poorly because of the complex scientific terms in the research papers. How can the company improve the performance of the chatbot?
A. Use few-shot prompting to define how the FM can answer the questions. B. Use domain adaptation fine-tuning to adapt the FM to complex scientific terms. C. Change the FM inference parameters. D. Clean the research paper data to remove complex scientific terms.
Answer: B Explanation:Domain adaptation fine-tuning involves training a foundation model (FM) further using aspecific dataset that includes domain-specific terminology and content, such as scientificterms in research papers. This process allows the model to better understand and handlecomplex terminology, improving its performance on specialized tasks.Option B (Correct): "Use domain adaptation fine-tuning to adapt the FM to complexscientific terms":This is the correct answer because fine-tuning the model ondomain-specific data helps it learn and adapt to the specific language and termsused in the research papers, resulting in better performance.Option A:"Use few-shot prompting to define how the FM can answer thequestions" is incorrect because while few-shot prompting can help in certainscenarios, it is less effective than fine-tuning for handling complex domain-specificterms.Option C:"Change the FM inference parameters" is incorrect because adjustinginference parameters will not resolve the issue of the model's lack ofunderstanding of complex scientific terminology.Option D:"Clean the research paper data to remove complex scientific terms" isincorrect because removing the complex terms would result in the loss ofimportant information and context, which is not a viable solution.AWS AI Practitioner References:Domain Adaptation in Amazon Bedrock:AWS recommends fine-tuning models withdomain-specific data to improve their performance on specialized tasks involvingunique terminology.
Question # 22
An AI practitioner is using a large language model (LLM) to create content for marketing campaigns. The generated content sounds plausible and factual but is incorrect. Which problem is the LLM having?
A. Data leakage B. Hallucination C. Overfitting D. Underfitting
Answer: B Explanation:In the context of AI, "hallucination" refers to the phenomenon where a model generatesoutputs that are plausible-sounding but are not grounded in reality or the training data. Thisproblem often occurs with large language models (LLMs) when they create information thatsounds correct but is actually incorrect or fabricated.Option B (Correct): "Hallucination":This is the correct answer because the problemdescribed involves generating content that sounds factual but is incorrect, which ischaracteristic of hallucination in generative AI models.Option A:"Data leakage" is incorrect as it involves the model accidentally learningfrom data it shouldn't have access to, which does not match the problem ofgenerating incorrect content.Option C:"Overfitting" is incorrect because overfitting refers to a model that haslearned the training data too well, including noise, and performs poorly on newdata.Option D:"Underfitting" is incorrect because underfitting occurs when a model istoo simple to capture the underlying patterns in the data, which is not the issuehere.AWS AI Practitioner References:Large Language Models on AWS:AWS discusses the challenge of hallucination inlarge language models and emphasizes techniques to mitigate it, such as usingguardrails and fine-tuning.
Question # 23
A medical company deployed a disease detection model on Amazon Bedrock. To comply with privacy policies, the company wants to prevent the model from including personal patient information in its responses. The company also wants to receive notification when policy violations occur. Which solution meets these requirements?
A. Use Amazon Macie to scan the model's output for sensitive data and set up alerts forpotential violations. B. Configure AWS CloudTrail to monitor the model's responses and create alerts for anydetected personal information. C. Use Guardrails for Amazon Bedrock to filter content. Set up Amazon CloudWatchalarms for notification of policy violations. D. Implement Amazon SageMaker Model Monitor to detect data drift and receive alertswhen model quality degrades.
Answer: C Explanation:Guardrails for Amazon Bedrock provide mechanisms to filter and control the contentgenerated by models to comply with privacy and policy requirements. Using guardrailsensures that sensitive or personal information is not included in the model's responses.Additionally, integrating Amazon CloudWatch alarms allows for real-time notification whena policy violation occurs.Option C (Correct): "Use Guardrails for Amazon Bedrock to filter content. Set upAmazon CloudWatch alarms for notification of policy violations":This is the correctanswer because it directly addresses both the prevention of policy violations andthe requirement to receive notifications when such violations occur.Option A:"Use Amazon Macie to scan the model's output for sensitive data" isincorrect because Amazon Macie is designed to monitor data in S3, not to filterreal-time model outputs.Option B:"Configure AWS CloudTrail to monitor the model's responses" isincorrect because CloudTrail tracks API activity and is not suited for contentmoderation.Option D:"Implement Amazon SageMaker Model Monitor to detect data drift" isincorrect because data drift detection does not address content moderation orprivacy compliance.AWS AI Practitioner References:Guardrails in Amazon Bedrock:AWS provides guardrails to ensure AI modelscomply with content policies, and using CloudWatch for alerting integratesmonitoring capabilities.
Question # 24
A company is implementing the Amazon Titan foundation model (FM) by using Amazon Bedrock. The company needs to supplement the model by using relevant data from the company's private data sources. Which solution will meet this requirement?
A. Use a different FM B. Choose a lower temperature value C. Create an Amazon Bedrock knowledge base D. Enable model invocation logging
Answer: C Explanation:Creating an Amazon Bedrock knowledge base allows the integration of external or privatedata sources with a foundation model (FM) like Amazon Titan. This integration helpssupplement the model with relevant data from the company's private data sources toenhance its responses.Option C (Correct): "Create an Amazon Bedrock knowledge base":This is thecorrect answer as it enables the company to incorporate private data into the FMto improve its effectiveness.Option A:"Use a different FM" is incorrect because it does not address the need tosupplement the current model with private data.Option B:"Choose a lower temperature value" is incorrect as it affects outputrandomness, not the integration of private data.Option D:"Enable model invocation logging" is incorrect because logging does nothelp in supplementing the model with additional data.AWS AI Practitioner References:Amazon Bedrock and Knowledge Integration:AWS explains how creating aknowledge base allows Amazon Bedrock to use external data sources to improvethe FM’s relevance and accuracy.
Question # 25
A company is using the Generative AI Security Scoping Matrix to assess security responsibilities for its solutions. The company has identified four different solution scopes based on the matrix. Which solution scope gives the company the MOST ownership of security responsibilities?
A. Using a third-party enterprise application that has embedded generative AI features. B. Building an application by using an existing third-party generative AI foundation model (FM). C. Refining an existing third-party generative AI foundation model (FM) by fine-tuning themodel by using data specific to the business. D. Building and training a generative AI model from scratch by using specific data that acustomer owns.
Answer: D Explanation:Building and training a generative AI model from scratch provides the company with themost ownership and control over security responsibilities. In this scenario, the company isresponsible for all aspects of the security of the data, the model, and the infrastructure.Option D (Correct): "Building and training a generative AI model from scratch byusing specific data that a customer owns":This is the correct answer because itinvolves complete ownership of the model, data, and infrastructure, giving thecompany the highest level of responsibility for security.Option A:"Using a third-party enterprise application that has embedded generativeAI features" is incorrect as the company has minimal control over the security ofthe AI features embedded within a third-party application.Option B:"Building an application using an existing third-party generative AIfoundation model (FM)" is incorrect because security responsibilities are sharedwith the third-party model provider.Option C:"Refining an existing third-party generative AI FM by fine-tuning themodel with business-specific data" is incorrect as the foundation model and part of the security responsibilities are still managed by the third party.AWS AI Practitioner References:Generative AI Security Scoping Matrix on AWS:AWS provides a securityresponsibility matrix that outlines varying levels of control and responsibilitydepending on the approach to developing and using AI models.
Question # 26
A company has terabytes of data in a database that the company can use for business analysis. The company wants to build an AI-based application that can build a SQL query from input text that employees provide. The employees have minimal experience with technology. Which solution meets these requirements?
A. Generative pre-trained transformers (GPT) B. Residual neural network C. Support vector machine D. WaveNet
Answer: A Explanation:Generative Pre-trained Transformers (GPT) are suitable for building an AI-basedapplication that can generate SQL queries from natural language input provided byemployees.GPT for Natural Language Processing:Why Option A is Correct:Why Other Options are Incorrect:
Question # 27
A company wants to develop an educational game where users answer questions such as the following: "A jar contains six red, four green, and three yellow marbles. What is the probability of choosing a green marble from the jar?" Which solution meets these requirements with the LEAST operational overhead?
A. Use supervised learning to create a regression model that will predict probability. B. Use reinforcement learning to train a model to return the probability. C. Use code that will calculate probability by using simple rules and computations. D. Use unsupervised learning to create a model that will estimate probability density.
Answer: C Explanation:The problem involves a simple probability calculation that can be handled efficiently bystraightforward mathematical rules and computations. Using machine learning techniqueswould introduce unnecessary complexity and operational overhead.Option C (Correct): "Use code that will calculate probability by using simple rulesand computations":This is the correct answer because it directly solves theproblem with minimal overhead, using basic probability rules.Option A:"Use supervised learning to create a regression model" is incorrect as itovercomplicates the solution for a simple probability problem.Option B:"Use reinforcement learning to train a model" is incorrect becausereinforcement learning is not needed for a simple probability calculation.Option D:"Use unsupervised learning to create a model" is incorrect asunsupervised learning is not applicable to this task.AWS AI Practitioner References:Choosing the Right Solution for AI Tasks:AWS recommends using the simplestand most efficient approach to solve a given problem, avoiding unnecessarymachine learning techniques for straightforward tasks.
Question # 28
A company is building a customer service chatbot. The company wants the chatbot to improve its responses by learning from past interactions and online resources. Which AI learning strategy provides this self-improvement capability?
A. Supervised learning with a manually curated dataset of good responses and badresponses B. Reinforcement learning with rewards for positive customer feedback C. Unsupervised learning to find clusters of similar customer inquiries D. Supervised learning with a continuously updated FAQ database
Answer: B Explanation:Reinforcement learning allows a model to learn and improve over time based on feedbackfrom its environment. In this case, the chatbot can improve its responses by beingrewarded for positive customer feedback, which aligns well with the goal of selfimprovementbased on past interactions and new information.Option B (Correct): "Reinforcement learning with rewards for positive customerfeedback":This is the correct answer as reinforcement learning enables the chatbotto learn from feedback and adapt its behavior accordingly, providing selfimprovementcapabilities.Option A:"Supervised learning with a manually curated dataset" is incorrectbecause it does not support continuous learning from new interactions.Option C:"Unsupervised learning to find clusters of similar customer inquiries" isincorrect because unsupervised learning does not provide a mechanism forimproving responses based on feedback.Option D:"Supervised learning with a continuously updated FAQ database" isincorrect because it still relies on manually curated data rather than selfimprovementfrom feedback.AWS AI Practitioner References:Reinforcement Learning on AWS:AWS provides reinforcement learningframeworks that can be used to train models to improve their performance basedon feedback.
Question # 29
A company is building a customer service chatbot. The company wants the chatbot to improve its responses by learning from past interactions and online resources. Which AI learning strategy provides this self-improvement capability?
A. Supervised learning with a manually curated dataset of good responses and badresponses B. Reinforcement learning with rewards for positive customer feedback C. Unsupervised learning to find clusters of similar customer inquiries D. Supervised learning with a continuously updated FAQ database
Answer: B Explanation:Reinforcement learning allows a model to learn and improve over time based on feedbackfrom its environment. In this case, the chatbot can improve its responses by beingrewarded for positive customer feedback, which aligns well with the goal of selfimprovementbased on past interactions and new information.Option B (Correct): "Reinforcement learning with rewards for positive customerfeedback":This is the correct answer as reinforcement learning enables the chatbotto learn from feedback and adapt its behavior accordingly, providing selfimprovementcapabilities.Option A:"Supervised learning with a manually curated dataset" is incorrectbecause it does not support continuous learning from new interactions.Option C:"Unsupervised learning to find clusters of similar customer inquiries" isincorrect because unsupervised learning does not provide a mechanism forimproving responses based on feedback.Option D:"Supervised learning with a continuously updated FAQ database" isincorrect because it still relies on manually curated data rather than selfimprovementfrom feedback.AWS AI Practitioner References:Reinforcement Learning on AWS:AWS provides reinforcement learningframeworks that can be used to train models to improve their performance basedon feedback.
Question # 30
Which functionality does Amazon SageMaker Clarify provide?
A. Integrates a Retrieval Augmented Generation (RAG) workflow B. Monitors the quality of ML models in production C. Documents critical details about ML models D. Identifies potential bias during data preparation
Answer: D Explanation: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 thetasks described (correlation matrix, calculating statistics, visualizing data) are allpart of the EDA process.Option A:"Data pre-processing" is incorrect because it involves cleaning andtransforming data, not initial analysis.Option B:"Feature engineering" is incorrect because it involves creating newfeatures from raw data, not analyzing the data's existing structure.Option D:"Hyperparameter tuning" is incorrect because it refers to optimizingmodel parameters, not analyzing the data.AWS AI Practitioner References:Stages of the Machine Learning Pipeline:AWS outlines EDA as the initial phase ofunderstanding and exploring data before moving to more specific preprocessing,feature engineering, and model training stages.