The process of teaching employees to use AI systems
A technical process where AI learns patterns from historical data to make future predictions or decisions
Creating physical models of business processes
Developing organizational training materials

What is training data?

Data used to back up an application
A list of business rules codec into an AI system
Historical data used to teach an AI model how to recognize patterns
Data collected after a model is deployed

In the context of LLMs, what is a token?

A security credential for accessing the API
A fundamental unit of text that the model processes (word parts, punctuation, etc.)
A form of cryptocurrency used to pay for AI services
A complete sentence processed by the model

In the context of AI, what is an inference?

The process of training an AI model with labeled data
Using a trained model to make predictions or decisions on new data
The development of AI hardware
The extraction of features from raw data

Fill in the blank: ___ is when a machine learning model performs well on training data, but poorly on new data.

Underfitting
Normalization
Inference
Overfitting

Which technique helps protect sensitive data when training AI models?

Real-time inference
Data anonymization
Using raw personal data
Maximizing batch size

True or false? It is unethical to use AI-generated content in marketing without clearly indicating it was created by AI.

True
False

What is a model card in responsible AI development?

A physical ID card for AI developers
Documentation that describes a model's intended use, limitations, performance metrics, and ethical considerations
A visual diagram of neural network architecture
A credit card used to purchase AI services

In the context of machine learning (ML), what is a model?

A flowchart that maps out business logic
A collection of training data
A mathematical representation trained to recognize patterns or make predictions.
A user interface for interacting with AI

Fill in the blank: Deep learning is a subset of____.

human intelligence
data engineering
machine learning
robotics

Which of the following best describes the difference between AI and traditional software?

Traditional software automates task using logic, while AI systems adapt and learn from data.
AI systems require no data, but traditional systems do.
AI runs only on cloud platforms, while traditional software runs locally.
Traditional software uses neural networks and AI does not.

A client’s support team is overwhelmed with incoming tickets. They want to automate triage and respond to common issues. Which AI solution should you recommend?

Forecasting model
Conversational AI chatbot integrated with a knowledge base
Image recognition
Recommendation system

What does facial recognition typically involve?

Identifying speech patterns
Extracting emotions from text
Matching faces in images to known individuals
Recommending products to users

Fill in the blank: Clustering techniques are used when you ____.

have labels for your data
want to group data points with similar features
want to train a chatbot
are analyzing time-series data

True or false? A pre-trained AI model can often be customized with a client’s data to improve performance.

True
False

When building a machine learning model, what's the purpose of a test dataset?

To generate training labels
To improve model performance during training
To evaluate the model's performance on new, unseen data
To increase the size of the training dataset

True or false? The field of AI and regulatory guidance around it are static.

True
False

What's a key difference between traditional AI and foundation models?

Foundation models sure self-supervised learning.
Traditional AI can write entire documents.
Foundation models are limited to narrow task.
Traditional AI is trained on unlabeled data.

What is a common reason AI projects fail to deliver value?

Lack of executive sponsorship and business alignment
Too much data
Use of cloud services
Too much automation

Which of these is the least likely to benefit from AI?

Repetitive data entry
Personalizing marketing emails
Prediction customer churn
Replacing critical legal decision-making

A consulting client wants to personalize product recommendations for their customers. What AI capability would best support this?

Optical character recognition
Recommendation system
Image segmentation
Chatbot framework

True or false? Tools like Microsoft Copilot and Notion AI are examples of AI applications built on top of foundation models.

True
False

Your team wants to use ChatGPT to help draft internal documents. What critical limitation should you be aware of?

It can only generate very short texts.
It may confidently present incorrect information as factual.
It can only work with predetermined templates.
It cannot understand industry terminology.

What is Artificial General Intelligence (AGI)?

AI that can write marketing copy
AI that excels in processing visual data
A still-theoretical form of AI that can understand and learn any intellectual task a human can
A common term for no-code AI tools

What's a neural network?

A type of data science project
A complex mathematical equation for learning A to B mappings
A hardware component used in AI
A tool for automating tasks

A client wants to generate marketing copy at scale using AI. Which of the following is the most accurate statement about this use case?

Generative AI can assist, but outputs should be reviewed for accuracy and tone.
Generative AI generates factually correct information.
This isn't a recommended use case for AI.
Generative AI requires proprietary datasets to work.

Check all that apply. Which of the following is a potential application of AI?

Boosting productivity in HR
Customer service
Cybersecurity
Recommending products based on past purchases

Which of the following best defines artificial intelligence (AI)?

A set of hardware components that simulate human movement
A software system that mimics human intelligence to perform task
A programming language used for automation
A cloud service used to store machine data

What's a key difference between Retrieval-Augmented Generation (RAG) and traditional LLM implementation?

RAG models are always smaller and faster.
RAG incorporates external knowledge sources to supplement model responses.
RAG only works for coding tasks.
RAG doesn't require prompts.

Which type of machine learning involves training a model using labeled data?

Unsupervised learning
Reinforcement learning
Supervised learning
Deep learning

What is named entity recognition (NER) used for?

Extracting names, locations, or organizations from text
Recognizing specific objects in videos
Identifying images in a dataset
Translating documents

What’s the main benefit of using a platform like OpenAI’s API when building tools?

It builds full applications automatically.
It provides access to general-purpose language models for a wide range of text-based tasks.
It only works for image classifications.
It requires advanced data science knowledge to use.

What's a defining feature of generative AI models like GPT?

They classify images.
They generate new content such as text or images.
They compress large files.
They only perform regression.

True or false? AI systems trained on biased data can unintentionally produce unfair or discriminatory outcomes.

True

False

How is machine learning (ML) different from traditional programming?

ML uses hard-coded rules to make decisions.
ML writes its own software from scratch.
ML learns from data to make predictions or decisions without being explicitly programmed.
ML only works for image processing tasks.

True or false? Prompt engineering is a specific skill used by technical developers, not consultants.

True
False

A client’s AI model is unintentionally producing biased results for a specific demographic. What should you recommend?

Stop using AI entirely.
Remove the demographic data to make it fair.
Ignore the issue unless complaints are received.
Conduct a bias audit and retrain the model with more diverse data.

True or false? In supervised learning, the model learns from data without any labels or target outputs.

True
False

True or false? Generative AI tools like ChatGPT can NOT help consultants draft client reports, proposals, and marketing content.

True
False

A client is unsure whether AI is appropriate for their process automation needs. As a consultant, what’s the best first step?

Recommended an AI tool.
Ask about the available data and the problem's predictability.
Begin coding a prototype.
Suggest hiring more staff.

What is a deep fake?

A complex AI algorithm
A synthesized video/image of someone doing something they didn't
A type of adversarial attack
A biased AI output

A client asks you if AI can help improve their monthly sales forecasting. Which machine learning technique should you recommend?

Clustering
Classification
Regression
Reinforcement learning

Why is prompt engineering important when using LLMs?

It allows you to reduce data size.
It improves model training speed.
It helps get more accurate and relevant outputs from the model.
It makes the model learn new knowledge.

True or false? Every AI project should begin with a clear understanding of the business goal it's trying to solve.

True
False

True or false? Generative AI models like GPT can only summarize existing content and can NOT create anything new.

True
False

Which of the following is NOT a typical use case for conversational AI?

Internal knowledge base access
Customer support
FAQ automation
Image classification

Which of the following best describes ChatGPT and its competitors like Gemini and Claude?

Spreadsheet analysis tools
Rule-based expert systems
Large language models designed for natural language understanding and generation
Data encryption platforms

True or false? Hugging Face is a tool for creating low-code AI dashboards.

True
False

True or false? Natural language processing allows AI systems to understand, interpret, and generate human language.

True
False

Which task would best be handled using Natural Language Processing (NLP)?

Predicting weather patterns
Detecting anomalies in iOT sensor data
Classifying images of cats and dogs
Analyzing customer support transcripts

What should businesses do to protect their competitive advantage in AI?

Outsource their data.
Protect their data.
Share their models openly.
Ignore transparency.

What makes conversational AI different from simple rule-based bots?

It uses NLP and machine learning to understand and generate text dynamically.
It's unable to understand natural language.
It uses fixed responses only.
It's limited to answering yes or no questions.

A B2B SaaS client wants to improve how their SDR team prioritizes leads. Which AI capability should they use?

Image classification
Lead scoring using predictive analytics
Facial recognition
Anomaly detection

In the context of AI, what makes a system intelligent?

It uses a user interface.
It can store large amounts of data.
It mimics human reasoning and decision-making.
It connects to the internet.

What's one practical way a consultant can help a client increase transparency in an AI system?

Use proprietary models.
Recommend regular data deletion.
Document how the model makes decisions and provide explanations for users.
Limit who can see outputs.

In the context of AI, what is hallucination?

AI generating creative content
AI generating incorrect answers with confidence
AI processing data very quickly
AI having emotional responses

Why is accountability important in responsible AI?

It designates human oversight for AI actions and outcomes.
It ensures AI will always be accurate.
It reduces data storage costs.
It eliminates the need for AI ethics.

Which of the following is NOT a commonly used AI tool or framework?

PyTorch
TensorFlow
Hugging Face
Python

True or false? Claude is an open-source AI tool anyone can download and run locally.

True
False

You've been tasked with leading your department's first AI initiative. What should be your first step?

Purchase and AI platform.
Restructure your department to focus on AI.
Hire data scientists.
Define the specific business problem you're trying to solve.

Which AI platform offers access to open-source foundation models and a large model hub, often used for fine-tuning or experimentation?

Salesforce Einstein
GitHub
Hugging Face
Webflow

Which of these is a common limitation of generative AI models?

They can generate content in multiple formats.
They struggle to cite sources in a verifiable way.
They may generate plausible-sounding but incorrect or biased information.
They're incapable of learning from prompts.

What is a key feature of Artificial Narrow Intelligence (ANI)?

It can understand and perform any intellectual task a human can.
It specializes in a single task or domain.
It's more powerful than human intelligence.
It can replicate human emotions.

Which step involves putting an AI model into use in a real-world environment?

Training
Validation
Deployment
Feature extraction

Which of the following best improves a prompt to generate a detailed response from an LLM?

Break the prompt into multiple unrelated questions.
Keep the prompt extremely short.
Include emojis.
Give context and ask for step-by-step reasoning.

How can AI bias be reduced?

By using more a complex algorithms
By using more historical data without changes
By avoiding cloud services
By diversifying training data and testing model fairness

True or false? Bias in AI doesn’t occur as long as the model itself is technically accurate.

True
False

True or false? All AI models are interpretable and can easily explain how they make decisions.

True
False

What is algorithmic bias?

An error in coding
A problem found in robotics
A tendency for an AI system to produce skewed or unfair outcomes
A feature of deep learning models

A client asks you to build a model that selects job applicants based on historical hiring data. As a consultant, what’s your first ethical concern?

Whether the model can scale across departments
If historical data may contain bias that could be perpetuated by the model
If the client prefers a cloud or on-prem solution
Whether it's faster to use a rule-based system

Which of the following is NOT an example of Artificial Narrow Intelligence (ANI)?

A chatbot that answers FAQs
A self-driving car trained for road navigation
An AI that generates email responses
An AI that can independently invent scientific theories and understand human emotions

True or false? ChatGPT can write code, answer questions, summarize text, and simulate conversation.

True
False

True or false? Microsoft and Google offer AI APIs that allow businesses to analyze language, extract key information, and translate text.

True
False

Which of the following is true of Artificial General Intelligence (AGI)?

It has already been achieved and deployed.
It specializes in single tasks.
It's capable of learning and performing any intellectual task a human can.
It's the same as robotic process automation.

What’s one key question to ask when assessing whether a task is suitable for AI?

Can this task be completed with no human involvement at all?
Does this task require constant manual rule writing?
Is there enough data to learn patterns from?
Does it involve high-performance hardware?

What are the three modes of AI consumption?

Embedded AI, API calls, platform model
Cloud-based AI, on-premise AI, hybrid AI
Free AI, paid AI, subscription AI
Research AI, development AI, production AI

Download Now!

Which technique helps protect sensitive data when training AI models?

By vmartinez

Which technique helps protect sensitive data when training AI models?

Real-time inference
Data anonymization
Using raw personal data
Maximizing batch size

Fill in the blank: ___ is when a machine learning model performs well on training data, but poorly on new data.

By vmartinez

Fill in the blank: ___ is when a machine learning model performs well on training data, but poorly on new data.

Underfitting
Normalization
Inference
Overfitting

In the context of AI, what is an inference?

By vmartinez

In the context of AI, what is an inference?

The process of training an AI model with labeled data
Using a trained model to make predictions or decisions on new data
The development of AI hardware
The extraction of features from raw data

In the context of LLMs, what is a token?

By vmartinez

In the context of LLMs, what is a token?

A security credential for accessing the API
A fundamental unit of text that the model processes (word parts, punctuation, etc.)
A form of cryptocurrency used to pay for AI services
A complete sentence processed by the model

What is training data?

By vmartinez

What is training data?

Data used to back up an application
A list of business rules codec into an AI system
Historical data used to teach an AI model how to recognize patterns
Data collected after a model is deployed

When explaining model training to a business stakeholder, which description is most appropriate?

By vmartinez

When explaining model training to a business stakeholder, which description is most appropriate?

The process of teaching employees to use AI systems
A technical process where AI learns patterns from historical data to make future predictions or decisions
Creating physical models of business processes
Developing organizational training materials