C_AIG_2412 SAP Certified Associate - SAP Generative AI Developer Sample Questions

C_AIG_2412 SAP Certified Associate - SAP Generative AI Developer

Questions and Answers

Question 1

Match the components of a Retrieval Augmented Generation architecture to the diagram.

• (✗) Box - Models- Vector Database   Box-> Applications- Backend and Frontend   Box-> Retrieval- Embedding Model and LLM
• (✗) Box - Models- Embedding Model and LLM   Box-> Applications- Vector Database   Box- Retrieval- Backend and Frontend
• (✗) Box - Models- Backend and Frontend   Box-> Applications- Embedding Model and LLM   Box-> Retrieval- Vector Database
• (✓) Box - Models- Embedding Model and LLM   Box- Applications- Backend and Frontend   Box- Retrieval- Vector Database

Question 2

Which of the following steps must be performed to deploy LLMs in the generative Al hub?

• (✗) 1. Provision SAP Al   2. Core Create a configuration   3. Run the booster
• (✗) 1. Run the booster   2. Create service keys   3. Select the executable ID
• (✓) 1. Provision SAP Al Core   2. Check for foundation model scenario   3. Create a configuration   4. Create a deployment
• (✗) 1. Check for foundation model scenario   2. Create a deployment   3. Configuring entitlements

Question 3

You want to extract useful information from customer emails to augment existing applications in your company. How can you use generative-ai-hub-sdk in this context?

• (✗) Generate random email content and send them to customers.
• (✓) Generate JSON strings based on extracted information.
• (✗) Generate a new SAP application based on the mail data.
• (✗) Train custom models based on the mail data.

Question 4

What are some benefits of the SAP Al Launchpad? Note: There are 2 correct answers to this question.

• [✓] Centralized Al lifecycle management for all Al scenarios.
• [✗] Direct deployment of Al models to SAP HANA.
• [✗] Integration with non-SAP platforms like Azure and AWS.
• [✓] Simplified model retraining and performance improvement.

Question 5

Why would a user include formatting instructions within a prompt?

• (✗) To increase the faithfulness of the output
• (✗) To force the model to separate relevant and irrelevant output
• (✗) To redirect the output to another software program
• (✓) To ensure the model's response follows a desired structure or style

Question 6

Where can you configure language models in generative Al hub?

• (✗) The Configuration tab of the SAP BTP cockpit
• (✗) The Orchestration tab in SAP AI Launchpad
• (✓) The Configuration tab within ML Operations in SAP AI Launchpad
• (✗) The Models tab in Prompt Editor

Question 7

Which of the following are features of the SAP Al Foundation? Note: There are 2 correct answers to this question.

• [✓] Al runtimes and lifecycle management
• [✓] Ready-to-use Al services
• [✗] Joule integration in SAP SuccessFactors
• [✗] Open source Al model repository

Question 8

How does SAP deal with vulnerability risks created by generative Al? Note: There are 2 correct answers to this question.

• [✓] By implementing responsible Al use guidelines and strong product security standards.
• [✗] By relying on external vendors to manage security threats.
• [✓] By identifying human, technical, and exfiltration risks through an Al Security Taskforce.
• [✗] By focusing on technological advancement only.

Question 9

What contract type does SAP offer for Al ecosystem partner solutions?

• (✗) Annual subscription-only contracts
• (✗) Pay-as-you-go for each partner service
• (✓) All-in-one contracts, with services that are contracted through SAP
• (✗) Bring Your Own License (BYOL) for embedded partner solution

Question 10

How does the Al API support SAP Al scenarios? Note: There are 2 correct answers to this question.

• [✓] By integrating Al services into business applications
• [✓] By providing a unified framework for operating Al services
• [✗] By managing Kubernetes clusters automatically
• [✗] By integrating Al models into third-party platforms like AWS

Question 11

Which of the following steps is NOT a requirement to use the Orchestration service?

• (✗) Create a deployment for orchestration
• (✗) Create an instance of an Al model
• (✓) Modify the underlying Al models
• (✗) Get an auth token for orchestration

Question 12

What is a significant risk associated with using LLMs?

• (✗) Immediate accuracy without fine-tuning.
• (✗) Lack of scalability in business applications.
• (✗) Reduced computational requirements.
• (✓) Potential biases in generated content.

Question 13

What are some benefits of using an SDK for evaluating prompts within the context of generative AI? Note: There are 3 correct answers to this question.

• [✓] Creating custom evaluators that meet specific business needs
• [✓] Automating prompt testing across various scenarios
• [✓] Providing metrics to quantitatively assess response quality
• [✗] Supporting low code evaluations using graphical user interface
• [✗] Maintaining data privacy by using data masking techniques

Question 14

Which of the following statements accurately describe the RAG process? Note: There are 2 correct answers to this question.

• [✓] The retrieved content is combined with the LLM's capabilities to generate a response.
• [✓] The user's question is used to search a knowledge base or a set of documents.
• [✗] The embedding model stores the generated answers for future reference.
• [✗] The LLM directly answers the user's question without accessing external information.

Question 15

You want to use the orchestration service through SAP's generative-Al-hub-sdk. What does the following code do?

python

from gen_ai_hub.orchestration.models.11m import LLM 

llm = LLM(name="gpt-40", version="latest", parameters={"max_tokens": 256, "temperature": 0.2})

• (✗) Create the Orchestration Configuration
• (✓) Define the LLM
• (✗) Define the Template and Default Input Values
• (✗) Run the Orchestration Request

Question 16

What are some advantages of using agents in training models? Note: There are 2 correct answers to this question.

• [✗] To eliminate the need for human oversight
• [✓] To streamline LLM workflows
• [✓] To improve the quality of results
• [✗] To guarantee accurate decision making in complex scenarios

Question 17

What is the primary function of the generative Al hub in SAP's Al Foundation?

• (✗) To provide ready-to-use Al services for document processing
• (✗) To serve as an abstraction layer to access a range of foundation Al models
• (✗) To store embeddings of unstructured data for semantic data retrieval
• (✓) To manage the Al lifecycle efforts end-to-end

Question 18

What does the Prompt Management feature of the SAP Al Iaunchpad allow users to do?

• (✗) Create and edit prompts
• (✗) Provide personalized user interactions
• (✗) Interact with models through a conversational interface
• (✓) Access and manage saved prompts and their versions

Question 19

Which statement best describes the Chain-of-Thought (COT) prompting technique?

• (✗) Writing a series of connected prompts creating a chain of related information.
• (✗) Linking multiple Al models in sequence, where each model's output becomes the input for the next model in the chain.
• (✗) Connecting related concepts by having the LLM generate chains of ideas.
• (✓) Concatenating multiple related prompts to form a chain, guiding the model through sequential reasoning steps.

Question 20

How do resource groups in SAP Al Core improve the management of machine learning workloads? Note: There are 2 correct answers to this question.

• [✗] They enable simultaneous orchestration of Kubernetes clusters.
• [✓] They ensure workload separation for different tenants or departments.
• [✗] They enhance pipeline execution speeds through workload distribution.
• [✓] They provide isolation for datasets and Al artifacts.

Question 21

Which of the following describes Large Language Models (LLMs)?

• (✓) They utilize deep learning to process and generate human-like text.
• (✗) They are rule-based systems designed for specific tasks.
• (✗) They cannot process large datasets efficiently.
• (✗) They rely on predefined scripts for decision-making.

Question 22

How does SAP ensure the enterprise-readiness of its Al solutions?

• (✗) By using generic Al models without business context complying with Al ethics standards
• (✗) By ensuring that Al models make bias-free decisions without human input
• (✓) By implementing rigorous product standards for Al capabilities

Question 23

Which of the following are functionalities provided by the generative-Al-hub-SDK? Note: There are 2 correct answers to this question.

• [✓] Create chat responses and embeddings
• [✓] Interact with LLMs
• [✗] Configure SAP BTP credentials
• [✗] Customize SAP Al Launchpad

Question 24

What is one primary benefit of using LLMs in business applications?

• (✗) They eliminate the need for data security measures.
• (✓) They enhance automation and scalability of processes.
• (✗) They are only applicable for customer support scenarios.
• (✗) They require minimal computing power for training.

Question 25

Which of the following must you do before connecting to a dataset in order to train a machine learning model in SAP Al Core? Note: There are 2 correct answers to this question.

• [✗] Store the dataset in the SAP HANA Vector Engine.
• [✗] Grant access rights to the SAP BTP cockpit.
• [✓] Store the dataset in a hyperscaler object store.
• [✓] Provide the storage secret to access the dataset.

Question 26

Which of the following capabilities does the generative Al hub provide to developers? Note: There are 2 correct answers to this question.

• [✗] Code generation to extend SAP BTP applications
• [✓] Tools for prompt engineering and experimentation
• [✗] Proprietary LLMs exclusively
• [✓] Integration of foundation models into applications

Question 27

What can be done once the training of a machine learning model has been completed in SAP Al Core? Note: There are 2 correct answers to this question.

• [✓] The model can be deployed for inferencing.
• [✗] The model can be deployed in SAP HANA.
• [✓] The model can be registered in the hyperscaler object store.
• [✗] The model's accuracy can be optimized directly in SAP HANA.

Question 28

Which of the following sequence of steps does SAP recommend you use to solve a business problem using generative Al hub?

• (✓) 1. Create a basic prompt in SAP Al Launchpad   2. Scale the solution using generative-ai-hub-sdk   3. Create a baseline evaluation method for the simple prompt   4. Enhance the prompts   5. Evaluate various models for the problem using generative-ai-hub-sdk
• (✗) 1. Create a basic prompt in SAP AI Launchpad   2. Evaluate various models for the problem using generative-ai-hub-sdk   3. Scale the solution using generative-ai-hub-sdk   4. Create a baseline evaluation method for the simple prompt   5. Enhance the prompts.
• (✗) 1. Create a basic prompt in SAP Al Launchpad   2. Enhance the prompts   3. Create a baseline evaluation method for the simple prompt   4. Evaluate various models for the problem using generative-ai-hulb-sdk   5. Scale the solution using generative-ai-hulb-sdk

Question 29

What is Machine Learning (ML)?

• (✗) A technology that equips machines with human-like capabilities such as problem-solving, visual perception, and decision-making.
• (✓) A subset of Al that focuses on enabling computer systems to learn and improve from experience or data.
• (✗) A form of Al that only focuses on creating new content, including text, images, sound, and videos.
• (✗) A statistical method for data processing that does not involve any Al techniques.

Question 30

What must be defined in an executable to train a machine learning model using SAP Al Core? Note: There are 2 correct answers to this question.

• [✓] Infrastructure resources such as CPUs or GPUs
• [✓] Pipeline containers to be used
• [✗] Deployment templates for SAP Al Launchpad
• [✗] User scripts to manually execute pipeline steps

Question 31

Which of the following is a principle of effective prompt engineering?

• (✗) Combine multiple complex tasks into a single prompt.
• (✓) Use precise language and providing detailed context in prompts.
• (✗) Write vague and open-ended instructions to encourage creativity.
• (✗) Keep prompts as short as possible to avoid confusion.

Question 32

Why is generative Al gaining significant attention and investment in the current business landscape? Note: There are 2 correct answers to this question.

• [✓] It only requires natural language skills to use.
• [✗] It can run entire business operations without human intervention.
• [✓] It lowers barriers to adoption.
• [✗] It can replicate complex technical skills without training or quality control.

Question 33

Which of the following are grounding principles included in SAP's Al Ethics framework? Note: There are 3 correct answers to this question.

• [✓] Human agency and oversight
• [✗] Maximize business profits
• [✗] Store all user data for legal proceedings
• [✓] Avoid bias and discrimination
• [✓] Transparency and explainability

Question 34

What advantage can you gain by leveraging different models from multiple providers through the SAP's generative Al hub?

• (✓) Enhance the accuracy and relevance of Al applications that use SAP's data assets
• (✗) Design new product interfaces for SAP application
• (✗) Get more training data for new models
• (✗) Train new models using SAP and non-SAP data

Question 35

Which of the following techniques uses a prompt to generate or complete subsequent prompts (streamlining the prompt development process), and to effectively guide Al model responses?

• (✗) Chain-of-thought prompting
• (✗) Few-shot prompting
• (✗) One-shot prompting
• (✓) Meta prompting

Question 36

What is a Large Language Model (LLM)?

• (✗) A rule-based expert system to analyze and generate grammatically correct sentences.
• (✗) A gradient boosted decision tree algorithm for predicting text.
• (✓) An Al model that specializes in processing, understanding, and generating human language.
• (✗) A database system optimized for storing large volumes of textual data.

Question 37

Which of the following executables in generative Al hub works with Anthropic models?

• (✗) Azure OpenAl Service
• (✗) GCP Vertex Al
• (✓) AWS Bedrock
• (✗) SAP Al Core

Question 38

What are the applications of generative Al that go beyond traditional chatbot applications? Note: There are 2 correct answers to this question.

• [✓] To produce outputs based on software input.
• [✓] To interpret human instructions and control software systems always producing output for human consumption.
• [✗] To follow a specific schema - human input, Al processing, and output for human consumption.
• [✗] To interpret human instructions and control software systems without necessarily producing output for human consumption.

Question 39

What is the purpose of splitting documents into smaller overlapping chunks in a RAG system?

• (✓) To enable the matching of different relevant passages to user queries
• (✗) To reduce the storage space required for the vector database
• (✗) To simplify the process of training the embedding model
• (✗) To improve the efficiency of encoding queries into vector representations

Question 40

What are some use cases for fine-tuning of a model? Note: There are 2 correct answers to this question.

• [✓] To sanitize model outputs
• [✗] To introduce new knowledge to a model in a resource-efficient way
• [✗] To quickly create iterations on a new use case
• [✓] To customize outputs for specific types of inputs

Question 41

How can few-shot learning enhance LLM performance?

• (✗) By reducing overfitting through regularization techniques
• (✗) By providing a large training set to improve generalization
• (✓) By offering input-output pairs that exemplify the desired behavior
• (✗) By enhancing the model's computational efficiency

Question 42

What is the primary function of the embedding model in a RAG system?

• (✓) To encode queries and documents into vector representations for comparison
• (✗) To store vector representations of documents and search for relevant passages
• (✗) To generate responses based on retrieved documents and user queries
• (✗) To evaluate the faithfulness and relevance of generated answers

Question 43

How can Joule improve workforce productivity? Note: There are 2 correct answers to this question.

• [✗] By offering generic task recommendations unrelated to specific roles.
• [✗] By resolving hardware malfunctions.
• [✓] By maintaining strict adherence to data privacy regulations.
• [✓] By providing context-based role-specific task assistance.

Question 44

What are some metrics to evaluate the effectiveness of a Retrieval Augmented Generation system? Note: There are 2 correct answers to this question.

• [✓] Relevance
• [✗] Speed
• [✗] Carbon footprint
• [✓] Faithfulness

Question 45

You want to assign urgency and sentiment categories to a large number of customer emails. You want to get a valid json string output for creating custom applications. You decide to develop a prompt for the same using generative Al hub. What is the main purpose of the following code in this context?

python

prompt_test = """Your task is to extract and categorize messages. Here are some examples:

{{?technique_examples}}

Use the examples when extract and categorize the following message:

{{?input}}

Extract and return a json with the following keys and values:

- "urgency" as one of {{?urgency}}

- "sentiment" as one of {{?sentiment}}

"categories" list of the best matching support category tags from: {{?categories}}

Your complete message should be a valid json string that can be read directly and onlycontains the keys mentioned in t

import random

random.seed(42)

k = 3

examples random. sample (dev_set, k)

example_template = """<example> {example_input}

examples

'\n---\n'.join([example_template.format(example_input=example ["message"],example_output=json.dumps (example[

f_test = partial (send_request, prompt=prompt_test, technique_examples examples,**option_lists)

response = f_test(input=mail["message"])

 

• (✓) Evaluate the performance of a language model using few-shot learning
• (✗) Generate random examples for language model training
• (✗) Preprocess a dataset for machine learning
• (✗) Train a language model from scratch

Question 46

What capabilities does the Exploration and Development feature of the generative Al hub provide? Note: There are 2 correct answers to this question.

• [✓] Al playground and chat
• [✓] Prompt editor and management
• [✗] Automatic model selection
• [✗] Develop and debug ABAP code

Question 47

What are some examples of generative Al technologies? Note: There are 2 correct answers to this question.

• [✗] Rule-based algorithms
• [✗] Robotic process automation
• [✓] Al models that generate new content based on training data
• [✓] Foundation models

Question 48

What defines SAP's approach to LLMs?

• (✗) Using proprietary transformer-based models exclusively.
• (✓) Ensuring ethical AI practices and seamless business integration.
• (✗) Prioritizing only the performance of open-source models.
• (✗) Avoiding partnerships with external AI providers.

Question 49

Which of the following is unique about SAP's approach to Al?

• (✓) SAP's deep integration of Al with business processes and analytics.
• (✗) Utilizing Al mainly for marketing purposes.
• (✗) Offering Al capabilities in their future products as of 2025.
• (✗) Focusing Al solely on customer support services.

Question 50

What are the benefits of SAP's generative Al hub? Note: There are 2 correct answers to this question.

• [✗] Send your data to various LLM providers for training feedback
• [✓] Build custom Al solutions and extend SAP applications
• [✗] Provide libraries for no-code development
• [✓] Accelerate Al development with flexible access to a broad range of models

Question 51

What are some drivers for the rapid adoption of generative AI? Note: There are 2 correct answers to this question.

• [✓] Ease of use
• [✗] Significant hardware cost savings
• [✓] Wide availability
• [✗] Availability of skilled developers

Question 52

What are some features of Joule? Note: There are 3 correct answers to this question.

• [✗] Downloading and processing data.
• [✓] Streamlining tasks with an Al assistant that knows your unique role.
• [✗] Generating standalone applications.
• [✓] Maintaining data privacy while offering generative Al capabilities.
• [✓] Providing coding assistance and content generation.

Question 53

Which neural network architecture is primarily used by LLMs?

• (✗) Sequential encoder-decoder architecture
• (✗) Recurrent neural network architecture
• (✗) Convolutional Neural Networks (CNNs)
• (✓) Transformer architecture with self-attention mechanisms

Question 54

What are some characteristics of the SAP generative Al hub? Note: There are 2 correct answers to this question.

• [✓] It provides instant access to a wide range of large language models (LLMSs).
• [✗] It operates independently of SAP's partners and ecosystem.
• [✓] It ensures relevant, reliable, and responsible business Al.
• [✗] It only supports traditional machine learning models.

Question 55

What is a part of LLM context optimization?

• (✗) Enhancing the computational speed of the model
• (✗) Adjusting the model's output format and style
• (✓) Providing the model with domain-specific knowledge needed to solve a problem
• (✗) Reducing the model's size to improve efficiency

Question 56

What are some components of the training pipeline in SAP AI Core? Note: There are 2 correct answers to this question.

• [✗] The SAP HANA database for model storage
• [✓] Input datasets stored in a hyperscaler object store
• [✗] Automated deployment to Kubernetes clusters
• [✓] Executables that define the training process

Question 57

What are some SAP recommendations to evaluate pricing and rate information of model usage within SAP's generative Al hub? Note: There are 2 correct answers to this question.

• [✓] Weigh the cost of using advanced models against the expected return on investment
• [✗] Avoid subscription-based pricing models
• [✗] Use pricing models that have fixed rates irrespective of the usage patterns
• [✓] Adopt best practice pricing strategies, such as outcome-based pricing

Question 58

Which technique is used to supply domain-specific knowledge to an LLM?

• (✗) Prompt template expansion
• (✗) Fine-tuning the model on general data
• (✓) Retrieval-Augmented Generation
• (✗) Domain-adaptation training

Question 59

What are some functionalities provided by SAP Al Core? Note: There are 3 correct answers to this question.

• [✓] Integration of Al services with business applications using a standardized API
• [✗] Monitoring and retraining models in SAP Al Core
• [✗] Management of SAP S/4HANA cloud infrastructure
• [✓] Orchestration of Al workflows such as model training and inference
• [✓] Continuous delivery and tenant isolation for scalability

Question 60

What is the goal of prompt engineering?

• (✓) To craft inputs that guide Al systems in generating desired outputs
• (✗) To optimize hardware performance for Al computations
• (✗) To develop new neural network architectures for Al models
• (✗) To replace human decision-making with automated processes

Question 61

What are some benefits of SAP Business Al? Note: There are 3 correct answers to this question.

• [✓] Intelligent business document processing
• [✗] Automatic human emotion recognition
• [✓] Personalized recommendations based on Al algorithms
• [✓] Al-powered forecasting and predictions
• [✗] Face detection and face recognition

Question 62

What does SAP recommend you do before you start training a machine learning model in SAP Al Core? Note: There are 3 correct answers to this question.

• [✓] Define the required infrastructure resources for training.
• [✓] Register the input dataset in SAP Al Core.
• [✗] Perform manual data integration with SAP HANA.
• [✓] Configure the training pipeline using templates.
• [✗] Configure the model deployment in SAP AI Launchpad.

Question 63

Which of the following is a benefit of using Retrieval Augmented Generation?

• (✗) It enables LLMs to learn new languages without additional training.
• (✗) It reduces the computational resources required for language modeling.
• (✗) It eliminates the need for fine-tuning LLMs for specific tasks.
• (✓) It allows LLMs to access and utilize information beyond their initial training data.

Question 64

You want to download a json output for a prompt and the response.

Which of the following interfaces can you use in SAP's generative Al hub in SAP AI Launchpad?

• (✗) Prompt Editor
• (✗) Prompt management
• (✓) Chat
• (✗) Administration