HOTSPOTFor each of the following statements, select Yes if the statement is true. Otherwise, select No.Note: Each correct selection is worth one point.Hot Area:
Answer(s): A
Box 1: NoNo - A generative AI solution is well-suited to predict next-quarter sales trends.Predictive AI, not generative AI, is designed to forecast future trends and outcomes.While Generative AI (GenAI) and Predictive AI both use historical data and machine learning, their primary purposes are distinct: Generative AI is designed to create new, original content (text, images, code), whereas Predictive AI is designed to forecast future trends and outcomes.Box 2: YesYes - A generative AI solution is well-suited to summarize lengthy policy documents.Generative AI solutions are exceptionally well-suited to summarize lengthy policy documents, offering a transformative approach to document management that enhances efficiency, reduces manual labor, and improves accuracy. These systems can analyze hundreds of pages in minutes, extracting key obligations, deadlines, and critical information while removing redundancy.Box 3: YesYes - A generative AI solution can create product descriptions from product specifications.Generative AI solutions significantly streamline e-commerce by transforming raw product specifications--such as materials, dimensions, and technical features--into engaging, human-readable product descriptions. By leveraging Large Language Models (LLMs) and Natural Language Processing (NLP), these tools can automate content creation for thousands of items in minutes, ensuring consistency in brand voice and improving search engine optimization (SEO)
Which statement accurately describes the difference between a pretrained generative AI model and a fine- tuned generative AI model?
Answer(s): C
Pretrained generative AI models are trained on massive, diverse datasets to gain foundational knowledge, while fine-tuned models take these pretrained weights and further train them on smaller, specific datasets to improve accuracy for narrow tasks or industries. This process aligns the model's output to specialized styles, domains, or tasks.Key Differences and Details:Pretrained Models (Foundational): These models (e.g., GPT-4) learn general language, concepts, and patterns from massive, broad datasets like Common Crawl. They are versatile but may lack expertise in specialized fields.Fine-tuned Models: By adjusting the weights of a pretrained model on a smaller, labeled dataset, the model is tailored to specific applications, such as medical analysis, legal document review, or a particular brand voice.Performance Benefits: Fine-tuning improves precision and reduces irrelevant outputs compared to a generic model.Methodology: While pretraining is unsupervised or self-supervised, fine-tuning often uses supervised learning
Box 1: NoNo - Retrieval Augmented Generation (RAG) requires model fine-tuning.Retrieval Augmented Generation (RAG) does not require model fine-tuning; it is designed to enhance LargeLanguage Models (LLMs) with external data without modifying their internal parameters. RAG enables fast knowledge updates and reduces hallucinations by fetching relevant information. While fine-tuning adjusts weights for domain-specific behavior, RAG is for dynamic, up-to-date knowledge.Box 2: YesYes - Retrieval Augmented Generation (RAG) is helpful when you need a generative AI solution that can access current, verifiable information.Think of Retrieval Augmented Generation (RAG) as giving an AI an "open-book exam" instead of forcing it to rely solely on its internal memory.By connecting the model to external, authoritative data sources--like a company's private knowledge base or real-time news--it becomes significantly more reliable in several ways:Reduces Hallucinations: Because the AI must ground its answers in the retrieved documents, it's less likely to "make things up".Transparency: You can see the exact source used for the answer, making it easy to verify facts.Cost-Efficiency: It is often much cheaper and faster to update a RAG database than it is to retrain or fine-tune a massive model on new informationBox 3: YesYes - Retrieval Augmented Generation (RAG) enables you to get more relevant responses based on your organization's documents without retraining the base model.Retrieval-Augmented Generation (RAG) is an AI framework that improves the accuracy and relevance of Large Language Model (LLM) outputs by incorporating, in real-time, external data that was not part of the model's original training, all without the need to retrain or fine-tune the base model. This method is particularly effective for allowing AI systems to access and utilize an organization's proprietary, private, or constantly updating data to generate more contextually accurate and authoritative responses.
Box 1: YesYes - Lack of trust in the security and accuracy of AI solutions is a common barrier to AI adoption.Lack of trust in the security and accuracy of AI solutions is a primary, persistent barrier to widespread AI adoption. As of 2025-2026, while adoption of AI is high, many initiatives are stuck in "pilot mode" or fail to scale because leaders and users do not fully trust the outputs of AI models.Core Trust and Accuracy Concerns"Black Box" Nature: Many AI models, particularly large language models (LLMs), operate as "black boxes," meaning their internal decision-making processes are not transparent, making it hard to explain how they arrived at a specific output.Data Inaccuracy and Hallucinations: Roughly 45% of organizations report worries about AI accuracy, including hallucinations (confidently stating false information) and bias.Security Risks: The increased reliance on AI introduces vulnerabilities, such as adversarial attacks (manipulating inputs to cause errors) and data poisoning.Shadow AI: A significant concern is "shadow AI," where employees use unauthorized, public AI tools to process sensitive company data, creating massive data leakage risks.Box 2: YesYes - AI adoption efforts often fail when organization overlook the important of high-quality data.AI adoption efforts frequently fail because organizations prioritize flashy AI tools over the foundational necessity of high-quality data. Research shows that up to 85% of AI projects fail due to issues like poor data quality, lack of relevant data, or inadequate data management.Why Poor Data Quality Causes AI Failure"Garbage In, Garbage Out": AI models learn from historical data. If the data is inaccurate, outdated, or incomplete, the AI will produce unreliable, flawed, or biased predictions.Data Silos and Inconsistency: Inconsistent data formats across systems prevent AI from creating a cohesive, accurate view, leading to fragmented and unreliable insights.Lack of Context/Metadata: AI models require not just raw data, but also context (metadata) to understand what they are processing; without it, they struggle to interpret information correctly.Data Decay: B2B data decays at a high rate (up to 22.5% annually), meaning outdated data quickly makes models obsolete.Box 3: YesYes - AI adoption is most successful when implemented with business and compliance stakeholders.AI adoption is most successful when implemented with business and compliance stakeholders because this cross-functional approach transforms AI from a mere technology experiment into a strategic, compliant, and sustainable business tool. Organizations that involve both, such as through AI Centers of Excellence, report 1.5x higher revenue growth and 1.6x greater shareholder returns, as they align technical capabilities directly with revenue-generating or cost-saving objectives.
https://www.stack-ai.com/blog/the-biggest-ai-adoption-challenges https://www.ibm.com/think/topics/ai-data-qualityhttps://www.bcg.com/press/24october2024-ai-adoption-in-2024-74-of-companies-struggle-to-achieve-and- scale-value
Your company creates a custom Azure Machine Learning model that uses a generative AI assistant.The model initially delivers strong results. However, six months later, the model predictions become noticeably less accurate.What is a possible cause of the issue?
Answer(s): B
This phenomenon is known as data drift.It is one of the top reasons model accuracy declines after deployment. In the context of a generative AI assistant, data drift occurs when the statistical properties or distributions of the input data (the prompts or context provided by users) change significantly from what the model was originally trained on.Key Causes of Data DriftChanging User Behavior: Users may start interacting with the assistant differently, using new slang, jargon, or evolving topics of interest.Real-World Changes: External events (like a pandemic or economic shift) can suddenly make the model's training data outdated and irrelevant.Data Pipeline Issues: Changes in how data is collected, such as updated sensors or modified web forms, can alter the format or scale of inputs.Seasonality: Patterns may fluctuate based on the time of year, such as different holiday shopping behaviors.Impact on Generative AIFor a generative assistant, data drift often manifests as:Reduced Relevance: Outputs feel outdated or fail to address contemporary topics.Increased Hallucinations: When faced with unfamiliar input patterns, the model may produce factually incorrect or nonsensical responses.Loss of Quality: Outputs may become less creative, more repetitive, or exhibit biased behavior.
https://nexla.com/ai-infrastructure/data-drift
Box 1: YesYes - Generative AI can be used in the automation of complex workflows.Generative AI (GenAI) can be used to automate complex workflows by acting as an orchestration layer that interprets intent, processes unstructured data, and dynamically adjusts process logic.Key Capabilities in Complex WorkflowsContext-Aware Decision Making: Unlike traditional rule-based automation, GenAI can reason through ambiguous situations and make data-driven decisions based on learned patterns and real-time inputs.Adaptive Process Intelligence: Systems can automatically adjust routing, task priority, and branch logic without manual reprogramming.Multimodal Integration: GenAI unifies diverse data types--such as emails, voice notes, scanned documents, and images--into a single automated flow.Continuous Optimization: Workflows can evolve by learning from historical data, user feedback, and observed performance to improve efficiency over time.Box 2: YesYes - Generative AI can be used to generate new content, such as text, images, and music.Generative AI is a branch of artificial intelligence focused on creating new, original content by identifying patterns in existing data, typically using neural networks. This technology has revolutionized content creation across various media, including text, images, music, code, and video, by enabling users to produce, edit, and refine content based on simple prompts.Box 3: NoNo - Generative AI is best suited to perform mathematical calculations and solve equations.Generative AI is a powerful tool for mathematical tasks, but it is not inherently best suited for basic, high- precision numerical calculations. While it excels at interpreting complex problems, explaining concepts, and generating step-by-step solutions, it is fundamentally a pattern-matching system that can "hallucinate" (invent) incorrect answers.For reliable, precise calculations, traditional calculators or Computer Algebra Systems (CAS) like Wolfram Alpha are superior because they are deterministic.
https://gleecus.com/blogs/generative-ai-building-intelligent-workflows-automation https://www.coursera.org/articles/what-is-generative-aihttps://arxiv.org/html/2511.07420v2
Box 1: NoNo - Retrieval Augmented Generated (RAG) architectures require that a large language model (LLM) be retrained on proprietary data.Retrieval-Augmented Generation (RAG) architectures do not require retraining the large language model (LLM) on proprietary data.The primary purpose of RAG is to provide an LLM with access to external, proprietary information at query time without modifying the model's underlying weights.Box 2: YesYes - Retrieval Augmented Generated (RAG) grounds a language model to produce more factual and context- relevant responses.Retrieval augmented generation (RAG) is a transformative architecture for generative AI, combining large language models (LLMs) with external knowledge retrieval to produce more accurate, grounded, and contextually relevant responses.Box 3: YesYes - Retrieval Augmented Generated (RAG) retrieves information from external knowledge sources at runtime instead of relying solely on the knowledge of a generative AI model.Retrieval-Augmented Generation (RAG) improves Large Language Model (LLM) accuracy by fetching relevant data from external, trusted sources (documents, databases, APIs) to ground AI responses in specific, up-to- date, or private information. This reduces hallucinations, avoids expensive retraining, and enables context- aware, verifiable answers to user queries.
https://theblueowls.com/blog/training-ai-a-comprehensive-guide-to-rag-implementations
Box 1: YesYes - A text-to-image generator can be used to translate content into other languages.AI tools, including specialized AI Image Translators and OCR-enabled systems, can extract and translate text within images into multiple languages instantly. These tools, often using models like Gemini and OpenAI, are used for localizing images, interpreting menus, and translating documents. While highly efficient for rapid, multi- language content generation, they may require human review for accuracy.Box 2: NoNo - A predictive analytics model can generate new marketing content for a company's online ads.While predictive analytics models do not typically "generate" creative content themselves, they are essential for guiding and personalizing the content that is shown to users. The actual creation of new text, images, or videos for ads is the primary function of generative AI.Key Differences in MarketingPredictive Analytics (The "Decision Maker"): Analyzes historical data to forecast future behavior, such as which customers are likely to click an ad, when they are most receptive, and which channels will yield the highest ROI.Generative AI (The "Creator"): Uses learned patterns to produce original assets, including ad copy, visuals, and personalized emails.Box 3: YesYes - A generative AI chatbot can engage customers in personalized conversations and recommend products.Generative AI chatbots have transformed customer service from static, rule-based scripting into dynamic, personalized conversations that can significantly enhance customer engagement, satisfaction, and sales. By leveraging Large Language Models (LLMs) and analyzing customer data in real-time, these AI assistants can understand context, sentiment, and intent to provide tailored product recommendations and immediate, human- like assistance.
https://notegpt.io/ai-image-translatorhttps://www.snowflake.com/en/fundamentals/predictive-analytics-marketing/ https://dialonce.ai/en/blog-ai/trends/building-customer-loyalty-generative-ai-chatbot.html
Post your Comments and Discuss Microsoft AB-731 exam prep with other Community members:
Our website is free, but we have to fight against AI bots and content theft. We're sorry for the inconvenience caused by these security measures. You can access the rest of the AB-731 content, but please register or login to continue.