IBM AI Fundamentals MCQs with Answers 2026

IBM AI Fundamentals MCQs with Answers 2026

159+
MCQs Covered
7
Course Modules
100%
Free Access
2026
Updated For

The IBM AI Fundamentals course on IBM SkillsBuild is a free credential covering the foundations of artificial intelligence, machine learning and deep learning, natural language processing, computer vision, running AI models with IBM Watson Studio, AI ethics and the AI job landscape. Pass the module quizzes and the final assessment to earn a digital badge you can share on LinkedIn. The 159+ solved MCQs below follow the course module by module, with the correct option marked and a short explanation under every question, so you can revise before each quiz and the final exam. A free PDF of the full set is at the end of the page.

Question 01

Which of the following best describes the goal of Artificial Intelligence?

ATo replace all human workers with machines permanently
BTo create systems that can perform tasks requiring human-like intelligence such as reasoning, learning, and problem solving ✅
CTo build faster computers with more processing power
DTo develop new programming languages for developers
💡 ExplanationAI is the field of computer science focused on building systems that perform tasks normally requiring human intelligence: understanding language, recognizing images, making decisions and solving problems.

Question 02

Who coined the term ‘Artificial Intelligence’ and when?

AAlan Turing, 1950
BJohn McCarthy, 1956 ✅
CArthur Samuel, 1959
DMarvin Minsky, 1969
💡 ExplanationJohn McCarthy used the term in the proposal for the 1956 Dartmouth Summer Research Project, considered the birth of AI as a field.

Question 03

Alan Turing’s 1950 paper ‘Computing Machinery and Intelligence’ introduced:

AThe first neural network
BThe Turing Test ✅
CThe first chess program
DMachine learning
💡 ExplanationTuring proposed the ‘imitation game’, now called the Turing Test, to judge whether a machine can exhibit intelligent behaviour indistinguishable from a human.

Question 04

What does the Turing Test evaluate?

AHow fast a computer can calculate
BWhether a machine’s conversation is indistinguishable from a human’s ✅
CHow much data a model can store
DWhether a robot can walk
💡 ExplanationIf a human judge cannot reliably tell the machine from a person in text conversation, the machine is said to pass the test.

Question 05

Which IBM computer defeated world chess champion Garry Kasparov in 1997?

AWatson
BDeep Blue ✅
CSummit
DGranite
💡 ExplanationDeep Blue’s 1997 victory was the first time a computer beat a reigning world champion in a match under tournament conditions.

Question 06

IBM Watson became famous in 2011 by winning which TV game show?

AWheel of Fortune
BWho Wants to Be a Millionaire
CJeopardy! ✅
DFamily Feud
💡 ExplanationWatson beat champions Ken Jennings and Brad Rutter on Jeopardy!, showing that a machine could understand natural-language clues.

Question 07

Gerald Tesauro’s TD-Gammon program at IBM was significant because it:

AWas the first chatbot
BTaught itself to play backgammon through reinforcement learning ✅
CTranslated languages
DRecognized faces
💡 ExplanationTD-Gammon (1992) learned by playing against itself, an early success of reinforcement learning that IBM’s course highlights.

Question 08

What is the difference between Narrow AI and General AI?

ANarrow AI performs specific tasks while General AI could perform any intellectual task a human can ✅
BNarrow AI is more powerful than General AI
CGeneral AI already exists in smartphones
DNarrow AI needs more data than General AI
💡 ExplanationAll AI in use today is Narrow (weak) AI, built for one task. Artificial General Intelligence with human-level reasoning across domains does not yet exist.

Question 09

Which of the following is an example of Narrow AI?

AA spam filter in your email ✅
BA machine with human-level consciousness
CA robot that can do every job a person can
DA superintelligent system
💡 ExplanationA spam filter does one task well. That is the definition of narrow AI.

Question 10

Artificial Superintelligence (ASI) refers to:

AAI used in supermarkets
BAI that surpasses human intelligence in virtually all fields ✅
CAI running on supercomputers
DAny AI that uses deep learning
💡 ExplanationASI is a hypothetical future stage beyond general AI; it does not exist today.

Question 11

Which statement about AI is TRUE?

AAI systems are conscious
BAI can only work with numbers
CMost AI today is narrow AI trained for specific tasks ✅
DAI never makes mistakes
💡 ExplanationToday’s AI is narrow, data-driven and can make errors, especially when data is biased or incomplete.

Question 12

What are the three main subfields that fall under Artificial Intelligence highlighted in the IBM course?

ACoding, databases and networking
BMachine Learning, Natural Language Processing and Computer Vision ✅
CCloud computing, cybersecurity and blockchain
DHTML, CSS and JavaScript
💡 ExplanationThe course organises AI around machine learning (learning from data), NLP (understanding language) and computer vision (interpreting images).

Question 13

Which of the following is a real-world example of AI being used in healthcare?

AUsing spreadsheets to record appointments
BSending automated email reminders
CUsing deep learning models to detect cancer in medical images ✅
DPrinting patient records
💡 ExplanationDeep learning models trained on X-rays and MRI scans can detect cancers and diabetic retinopathy, often as accurately as specialists.

Question 14

A recommendation system on a streaming service is an example of AI that:

AGenerates random suggestions
BLearns from your viewing history to predict what you will like ✅
COnly shows the newest content
DNeeds no data
💡 ExplanationRecommendation engines use machine learning on user behaviour data to predict preferences.

Question 15

Which of these is NOT a typical application of AI?

AFraud detection in banking
BVirtual assistants like Siri
CManually typing documents ✅
DSelf-driving cars
💡 ExplanationManual typing involves no learning or decision-making by a machine.

Question 16

What is an ‘algorithm’?

AA type of computer
BA step-by-step set of instructions for solving a problem ✅
CA programming language
DA database
💡 ExplanationAlgorithms are the procedures AI systems follow; machine learning algorithms learn their rules from data.

Question 17

What is ‘big data’ in the context of AI?

AData that takes up more than 1 GB
BExtremely large, fast and varied datasets that traditional tools cannot process easily ✅
COnly video data
DData stored on paper
💡 ExplanationBig data is often described by the three Vs: volume, velocity and variety. Modern AI depends on it for training.

Question 18

Why has AI advanced so rapidly in recent years?

AComputers became cheaper to buy
BAvailability of big data, powerful GPUs and better algorithms ✅
CProgramming became easier
DGovernments banned older software
💡 ExplanationThe combination of massive datasets, cheap parallel computing (GPUs) and advances such as deep learning drove the current AI boom.

Question 19

What is ‘structured data’?

AData organised in rows and columns, such as a spreadsheet ✅
BFree-form text
CImages and videos
DAudio recordings
💡 ExplanationStructured data fits neatly into tables; unstructured data (text, images, audio) does not and needs AI to interpret.

Question 20

Which of the following is unstructured data?

AA customer table in a database
BA sales spreadsheet
CA collection of customer emails ✅
DA list of product prices
💡 ExplanationEmails are free text with no fixed format, which is unstructured data.

Question 21

In the IBM course, ‘augmented intelligence’ means:

AAI that replaces humans
BAI designed to enhance and support human decision-making ✅
CAI with virtual reality
DAI without data
💡 ExplanationIBM prefers the term augmented intelligence to stress that AI should extend human capability rather than replace it.

Question 22

What is a ‘chatbot’?

AA robot that cleans floors
BA software program that simulates conversation with users ✅
CA spam email
DA hardware device
💡 ExplanationChatbots use NLP to understand user messages and respond, as in IBM watsonx Assistant.

Question 23

What is a ‘virtual assistant’ such as Siri or Alexa an example of?

ARobotics only
BAI combining speech recognition and natural language understanding ✅
CA search engine
DManual customer service
💡 ExplanationThese assistants convert speech to text, interpret intent with NLP and respond, often with speech synthesis.

Question 24

Which of the following is an example of AI in finance?

ACounting cash by hand
BDetecting unusual transactions that may be fraud ✅
CPrinting bank statements
DOpening a branch
💡 ExplanationFraud-detection models learn normal patterns of spending and flag anomalies in real time.

Question 25

Robotics combined with AI allows robots to:

AOnly follow fixed instructions
BPerceive their environment and make decisions ✅
CWork without electricity
DReplace all programmers
💡 ExplanationAI gives robots perception (computer vision), planning and learning instead of fixed scripted movements.

Next: Machine Learning and Deep Learning →

🧠

Topic 2 — Machine Learning and Deep Learning (Questions 26–59)

Supervised, unsupervised and reinforcement learning, neural networks, foundation models

Question 26

What is machine learning?

AProgramming every rule by hand
BA subset of AI where systems learn patterns from data to make predictions without being explicitly programmed ✅
CA type of computer hardware
DA spreadsheet function
💡 ExplanationArthur Samuel defined ML in 1959 as giving computers the ability to learn without being explicitly programmed.

Question 27

Who coined the term ‘machine learning’?

AAlan Turing
BArthur Samuel ✅
CJohn McCarthy
DGeoffrey Hinton
💡 ExplanationArthur Samuel, an IBM researcher, coined the term in 1959 while building a checkers-playing program that improved with experience.

Question 28

What are the three main types of machine learning?

AFast, slow and medium
BSupervised, unsupervised and reinforcement learning ✅
CText, image and audio
DOnline, offline and hybrid
💡 ExplanationSupervised learning uses labelled data, unsupervised finds structure in unlabelled data, and reinforcement learns from rewards.

Question 29

In supervised learning, the training data is:

AUnlabelled
BLabelled with the correct answers ✅
CRandom
DNot needed
💡 ExplanationEach example comes with the correct output (label), such as an email marked ‘spam’ or ‘not spam’.

Question 30

Predicting house prices from size and location is an example of:

AClustering
BRegression (supervised learning) ✅
CReinforcement learning
DUnsupervised learning
💡 ExplanationPredicting a continuous number from labelled examples is regression.

Question 31

Sorting emails into ‘spam’ and ‘not spam’ is an example of:

ARegression
BClassification ✅
CClustering
DDimensionality reduction
💡 ExplanationClassification assigns inputs to discrete categories.

Question 32

What is clustering in unsupervised machine learning?

AGrouping similar models for deployment
BSplitting training data into batches
CAutomatically grouping data points with similar characteristics without predefined labels ✅
DCombining neural networks
💡 ExplanationClustering finds natural groupings, such as customer segments, without human-labelled categories.

Question 33

Which algorithm is commonly used for clustering?

ALinear regression
BK-means ✅
CDecision tree
DLogistic regression
💡 ExplanationK-means groups data into k clusters around centroids.

Question 34

In reinforcement learning, an agent learns by:

AReading labelled examples
BReceiving rewards or penalties for its actions ✅
CCopying another model
DMemorising the data
💡 ExplanationThe agent interacts with an environment and adjusts its behaviour to maximise cumulative reward.

Question 35

Which of the following is a classic application of reinforcement learning?

AEmail spam filtering
BGame-playing agents such as AlphaGo or TD-Gammon ✅
CSorting customer records
DTranslating documents
💡 ExplanationGames provide clear rewards, making them ideal for reinforcement learning.

Question 36

What is the purpose of a decision tree algorithm?

ATo store files hierarchically
BTo make decisions by splitting data into branches based on feature values until a prediction is reached ✅
CTo grow trees in a simulation
DTo organise neural-network layers
💡 ExplanationDecision trees are interpretable supervised models used in fraud detection, diagnosis and customer segmentation.

Question 37

What is a ‘feature’ in machine learning?

AA software bug
BAn input variable used by the model to make predictions ✅
CThe model’s output
DA type of chart
💡 ExplanationFeatures are the measurable properties of the data, such as age, income or pixel values.

Question 38

What is a ‘label’ in supervised learning?

AThe model’s name
BThe correct output the model should predict for an example ✅
CA data type
DA chart title
💡 ExplanationLabels are the answers the model learns to predict.

Question 39

What is ‘training’ a model?

AInstalling the software
BFeeding data to the algorithm so it learns patterns and adjusts its parameters ✅
CRunning the model in production
DDeleting bad data
💡 ExplanationTraining adjusts the model’s internal parameters to reduce prediction error on the training data.

Question 40

Why is data split into training and test sets?

ATo save storage space
BTo evaluate how well the model performs on data it has not seen ✅
CTo make training faster
DIt is not necessary
💡 ExplanationTesting on held-out data shows whether the model generalises or has merely memorised the training set.

Question 41

What is ‘overfitting’?

AThe model is too simple
BThe model learns the training data too closely, including noise, and performs poorly on new data ✅
CThe model runs too fast
DThe dataset is too large
💡 ExplanationAn overfitted model has high training accuracy but low test accuracy.

Question 42

What is ‘underfitting’?

AThe model is too complex
BThe model is too simple to capture the patterns in the data ✅
CThe model has too much data
DThe model is overtrained
💡 ExplanationUnderfitting shows poor performance on both training and test data.

Question 43

What is ‘accuracy’ as a model metric?

AThe speed of predictions
BThe proportion of predictions the model gets right ✅
CThe size of the model
DThe number of features
💡 ExplanationAccuracy = correct predictions / total predictions. It can mislead on imbalanced data.

Question 44

What is a neural network?

AA computer network of servers
BA model made of interconnected layers of nodes (neurons) inspired by the human brain ✅
CA spreadsheet of neurons
DA type of database
💡 ExplanationArtificial neural networks pass data through layers of weighted connections to learn complex patterns.

Question 45

What are the three basic layer types in a neural network?

AFront, middle, back
BInput, hidden and output layers ✅
CData, model and result
DTraining, testing and validation
💡 ExplanationInput receives the data, hidden layers transform it, and the output layer produces the prediction.

Question 46

What is deep learning?

ALearning that takes a long time
BMachine learning using neural networks with many hidden layers ✅
CLearning from deep-sea data
DA type of supervised learning only
💡 Explanation‘Deep’ refers to the many layers that let the network learn hierarchical features.

Question 47

Which of the following encompasses the use of neural networks with multiple hidden layers to analyze complex data and make predictions?

ALinear regression
BDeep learning ✅
CClustering
DDecision trees
💡 ExplanationThis is the course’s definition of deep learning.

Question 48

What is the role of ‘weights’ in a neural network?

AThey measure the computer’s mass
BThey determine how strongly each input influences a neuron’s output and are adjusted during training ✅
CThey store the output
DThey label the data
💡 ExplanationTraining is the process of finding weights that minimise error.

Question 49

What is ‘backpropagation’?

ASending data backwards to the user
BThe algorithm that adjusts a network’s weights by propagating the error backward from the output ✅
CA way to delete layers
DA data-cleaning technique
💡 ExplanationBackpropagation with gradient descent is the standard method for training neural networks.

Question 50

Which type of learning would you use when you have thousands of unlabelled customer records and want to find natural groups?

ASupervised
BUnsupervised ✅
CReinforcement
DTransfer
💡 ExplanationNo labels plus a need for structure means unsupervised learning (clustering).

Question 51

What is ‘bias’ in a dataset?

AA technical error in the code
BSystematic distortion that makes the data unrepresentative of the real population ✅
CThe model’s accuracy
DA hidden layer
💡 ExplanationBiased training data leads to biased predictions, which is why data quality matters so much.

Question 52

What does ‘garbage in, garbage out’ mean for machine learning?

AModels should be deleted
BPoor-quality training data produces poor-quality predictions ✅
CData should be recycled
DModels produce random output
💡 ExplanationA model is only as good as the data it learns from.

Question 53

What is a ‘model’ in machine learning?

AThe computer used for training
BThe mathematical representation learned from data that makes predictions ✅
CThe dataset itself
DThe person who builds AI
💡 ExplanationAfter training, the model is the function that maps new inputs to outputs.

Question 54

What is ‘transfer learning’?

AMoving a model between computers
BReusing a model trained on one task as the starting point for a related task ✅
CTransferring data to the cloud
DLearning without data
💡 ExplanationTransfer learning saves time and data by building on pre-trained models.

Question 55

What is a ‘generative AI’ model?

AA model that only classifies data
BA model that creates new content such as text, images or code ✅
CA spreadsheet model
DA rule-based system
💡 ExplanationGenerative models like large language models and image generators produce new outputs rather than just labels.

Question 56

What is a ‘large language model’ (LLM)?

AA dictionary
BA deep-learning model trained on huge amounts of text to understand and generate language ✅
CA translation table
DA speech recorder
💡 ExplanationLLMs such as IBM Granite and GPT models are foundation models for text.

Question 57

What is a ‘foundation model’?

AA model that cannot be changed
BA large model trained on broad data that can be adapted to many downstream tasks ✅
CA model for building construction
DA basic rule-based program
💡 ExplanationIBM watsonx.ai provides foundation models that businesses can fine-tune for their own use cases.

Question 58

What does ‘hallucination’ mean for a generative AI model?

AThe model shuts down
BThe model confidently produces incorrect or made-up information ✅
CThe model refuses to answer
DThe model becomes conscious
💡 ExplanationHallucinations are a known limitation; outputs should be verified, especially in business use.

Question 59

Which best describes ‘fine-tuning’ a foundation model?

ADeleting its parameters
BFurther training it on a smaller, task-specific dataset ✅
CMaking it run slower
DConverting it to a chatbot
💡 ExplanationFine-tuning adapts a general model to a specific domain or task.

Next: Natural Language Processing →

💬

Topic 3 — Natural Language Processing (Questions 60–82)

How machines understand, interpret and generate human language, chatbots, prompting

Question 60

What does Natural Language Processing (NLP) enable a computer to do?

ADisplay text in multiple fonts
BUnderstand, interpret and generate human language ✅
CStore text files
DPrint documents faster
💡 ExplanationNLP bridges human language and machine understanding; it powers chatbots, translation and sentiment analysis.

Question 61

What is tokenization in NLP?

AEncrypting text
BBreaking text into smaller units such as words or sub-words ✅
CTranslating text
DDeleting punctuation
💡 ExplanationTokenization is usually the first step in processing text.

Question 62

What is ‘sentiment analysis’?

ACounting words in a document
BDetermining whether text expresses a positive, negative or neutral opinion ✅
CTranslating emotions into code
DSpelling correction
💡 ExplanationCompanies use it to analyse reviews and social-media posts.

Question 63

What is ‘named entity recognition’ (NER)?

ANaming a new AI model
BIdentifying and classifying names of people, organisations, places and dates in text ✅
CRecognising faces
DCounting entities in a database
💡 ExplanationNER extracts structured information from unstructured text.

Question 64

What is ‘speech recognition’?

AConverting spoken words into text ✅
BConverting text into speech
CRecognising a speaker’s face
DPlaying audio files
💡 ExplanationSpeech-to-text is used by virtual assistants and transcription services.

Question 65

What is ‘text-to-speech’ (speech synthesis)?

AConverting speech into text
BConverting written text into spoken audio ✅
CTranslating text
DSummarising text
💡 ExplanationIBM Watson Text to Speech generates natural-sounding voices from text.

Question 66

What is ‘machine translation’?

ATranslating code between programming languages
BAutomatically translating text from one human language to another ✅
CMoving data between machines
DConverting speech to text
💡 ExplanationNeural machine translation powers tools like Google Translate and Watson Language Translator.

Question 67

What is ‘intent’ in a chatbot?

AThe chatbot’s personality
BThe goal or purpose behind a user’s message ✅
CThe chatbot’s name
DThe length of the message
💡 ExplanationFor example, ‘I want to book a flight’ has the intent ‘book_flight’.

Question 68

What is an ‘entity’ in a chatbot conversation?

AThe user’s name only
BA specific piece of information in the message, such as a date, city or product ✅
CThe chatbot’s reply
DThe server
💡 ExplanationIn ‘book a flight to Paris on Friday’, Paris and Friday are entities.

Question 69

What is ‘stemming’ in NLP?

ARemoving stop words
BReducing words to their root form, e.g. ‘running’ to ‘run’ ✅
CTranslating words
DCounting word frequency
💡 ExplanationStemming and lemmatization normalise words so variations are treated alike.

Question 70

What are ‘stop words’?

AWords that end a sentence
BCommon words like ‘the’, ‘is’, ‘and’ that are often removed before analysis ✅
COffensive words
DWords in another language
💡 ExplanationRemoving stop words reduces noise in text processing.

Question 71

What is the ‘bag of words’ model?

AA storage format
BRepresenting text by the counts of words it contains, ignoring order ✅
CA translation method
DA speech model
💡 ExplanationIt is a simple way to turn text into numbers for machine learning.

Question 72

What are ‘word embeddings’?

AFonts for words
BNumerical vector representations of words that capture meaning and relationships ✅
CWord lists
DEncrypted text
💡 ExplanationWords with similar meanings have similar vectors, enabling models to understand context.

Question 73

What is a ‘transformer’ in NLP?

AAn electrical device
BA neural-network architecture using attention that powers modern language models ✅
CA translation table
DA speech recorder
💡 ExplanationTransformers, introduced in 2017, are the basis of GPT, BERT and IBM Granite models.

Question 74

What does the ‘attention mechanism’ allow a model to do?

APay attention to the user
BFocus on the most relevant parts of the input when producing each output ✅
CStay awake
DIgnore all input
💡 ExplanationAttention lets models weigh relationships between words regardless of distance.

Question 75

What is ‘prompt engineering’?

ABuilding hardware prompts
BDesigning the input text given to a generative model to get better outputs ✅
CWriting error messages
DProgramming a chatbot’s buttons
💡 ExplanationHow you phrase the prompt strongly affects an LLM’s answer.

Question 76

In prompt engineering, ‘zero-shot’ prompting means:

AGiving no instructions at all
BAsking the model to perform a task without any examples ✅
CGiving many examples
DPrompting with images only
💡 ExplanationThe model relies entirely on its training to complete the task.

Question 77

In prompt engineering, ‘few-shot’ prompting means:

APrompting with few words
BProviding a few examples in the prompt to show the model the expected format ✅
CRunning the prompt a few times
DUsing a small model
💡 ExplanationExamples guide the model’s output style and accuracy.

Question 78

‘One-shot’ prompting provides how many examples?

ANone
BOne ✅
CTwo
DMany
💡 ExplanationOne example is given before the actual request.

Question 79

What is ‘text summarization’?

ACounting characters
BAutomatically producing a shorter version of a text that keeps the key points ✅
CTranslating text
DCorrecting grammar
💡 ExplanationSummarization can be extractive (selecting sentences) or abstractive (generating new sentences).

Question 80

Which of the following is NOT an NLP task?

ASentiment analysis
BMachine translation
CImage classification ✅
DQuestion answering
💡 ExplanationImage classification is a computer-vision task.

Question 81

What makes human language difficult for computers?

AIt has too few words
BAmbiguity, context, slang and multiple meanings ✅
CIt cannot be stored
DIt has no grammar
💡 ExplanationThe same word can mean different things, and meaning depends on context.

Question 82

What is a ‘conversational AI’ system?

AA system that talks to itself
BSoftware that can hold natural dialogue with people via text or voice ✅
CA system for phone calls only
DA recorded message
💡 ExplanationIBM watsonx Assistant is a platform for building conversational AI.

Next: Computer Vision →

👁️

Topic 4 — Computer Vision (Questions 83–100)

Image classification, object detection, CNNs and vision applications

Question 83

What is computer vision?

AComputers with cameras attached
BThe field of AI that enables machines to interpret and understand visual information from images and video ✅
CScreen display technology
DImage storage
💡 ExplanationComputer vision lets systems recognise objects, faces, text and scenes.

Question 84

What is a Convolutional Neural Network (CNN) primarily used for?

AText translation
BImage and video analysis ✅
CSound recording
DDatabase queries
💡 ExplanationCNNs detect edges, shapes and objects through layers of filters, making them ideal for vision tasks.

Question 85

What is ‘image classification’?

AResizing an image
BAssigning a label to an entire image, such as ‘cat’ or ‘dog’ ✅
CFinding where objects are
DRemoving noise
💡 ExplanationClassification answers ‘what is in this image?’.

Question 86

What is ‘object detection’?

AStoring objects in a database
BIdentifying objects in an image AND locating them with bounding boxes ✅
CBlurring an image
DLabelling an entire image
💡 ExplanationDetection answers ‘what and where’, used in self-driving cars and security cameras.

Question 87

What is ‘image segmentation’?

ASplitting an image file into parts
BClassifying each pixel of an image to outline exact object boundaries ✅
CCompressing an image
DRotating an image
💡 ExplanationSegmentation is used in medical imaging to outline tumours precisely.

Question 88

What is ‘facial recognition’?

ADrawing faces
BIdentifying or verifying a person from facial features in an image ✅
CDetecting smiles only
DPhoto editing
💡 ExplanationIt raises privacy concerns, which is why IBM stopped offering general-purpose facial-recognition products in 2020.

Question 89

What is OCR?

AOnline Content Review
BOptical Character Recognition: converting images of text into machine-readable text ✅
CObject Colour Rendering
DOpen Camera Recording
💡 ExplanationOCR lets you search scanned documents and digitise forms.

Question 90

In a CNN, what does a ‘filter’ (kernel) do?

ARemoves bad images
BSlides over the image to detect features such as edges or textures ✅
CCompresses the image
DLabels the image
💡 ExplanationEarly layers detect simple features; deeper layers combine them into complex shapes.

Question 91

How do computers represent images?

AAs words
BAs grids of pixels with numeric values for colour intensity ✅
CAs sound waves
DAs vectors of letters
💡 ExplanationA colour image is a 3-D array of pixel values (height x width x RGB channels).

Question 92

Which of the following is an application of computer vision in healthcare?

AScheduling appointments
BAnalysing X-rays and MRI scans to detect disease ✅
CBilling patients
DPrinting prescriptions
💡 ExplanationCNNs trained on medical images assist radiologists in detecting cancers and fractures.

Question 93

Self-driving cars use computer vision to:

APlay music
BDetect lanes, pedestrians, traffic signs and other vehicles ✅
CBook service appointments
DCalculate fuel prices
💡 ExplanationVision is combined with radar and lidar for perception.

Question 94

What is ‘video analysis’ in computer vision?

AEditing videos
BApplying vision techniques across frames to track objects and recognise actions over time ✅
CCompressing videos
DAdding subtitles
💡 ExplanationUsed in sports analytics, surveillance and manufacturing quality control.

Question 95

What is ‘data augmentation’ in image training?

ABuying more images
BCreating modified copies of images (rotated, flipped, cropped) to expand the training set ✅
CDeleting duplicates
DIncreasing resolution
💡 ExplanationAugmentation helps prevent overfitting when real images are limited.

Question 96

What is ‘pose estimation’?

AGuessing a photo’s date
BDetecting the position of a person’s body joints in an image ✅
CMeasuring image size
DDetecting emotions
💡 ExplanationUsed in fitness apps, animation and physiotherapy.

Question 97

Why does computer vision need large labelled datasets such as ImageNet?

ATo store photos
BTo train models to recognise many object categories accurately ✅
CTo compress images
DTo test cameras
💡 ExplanationImageNet’s millions of labelled images drove the deep-learning breakthrough in 2012.

Question 98

Which is an example of computer vision in retail?

APrinting receipts
BCashier-less checkout that recognises products picked from shelves ✅
CEmail newsletters
DLoyalty cards
💡 ExplanationCameras and vision models track items so customers can walk out without scanning.

Question 99

What is ‘image generation’ in generative AI?

AScanning photos
BCreating new images from text prompts or other images ✅
CResizing images
DPrinting images
💡 ExplanationDiffusion models and GANs can create realistic images from descriptions.

Question 100

Which combination describes a ‘multimodal’ AI model?

AWorks with only images
BProcesses more than one type of data, such as text and images together ✅
CRuns on multiple computers
DHas many layers
💡 ExplanationMultimodal models can describe a picture in words or generate images from text.

Next: IBM Watson, watsonx and Running AI Models →

🔬

Topic 5 — IBM Watson, watsonx and Running AI Models (Questions 101–122)

Watson services, watsonx.ai, AutoAI in Watson Studio, deployment and monitoring

Question 101

What is IBM Watson?

AA chess computer
BIBM’s portfolio of AI products and services for business ✅
CA programming language
DAn operating system
💡 ExplanationWatson includes tools for NLP, speech, assistants, data science and governance.

Question 102

What is IBM watsonx?

AA video game
BIBM’s enterprise AI and data platform for building, training, deploying and governing AI models ✅
CA smartphone
DA database only
💡 ExplanationLaunched in 2023, watsonx has three components: watsonx.ai, watsonx.data and watsonx.governance.

Question 103

What is IBM watsonx.ai?

AA tool for storing data
BA studio for training, tuning and deploying foundation models and machine learning ✅
CA video-editing tool
DA chat app for consumers
💡 ExplanationIt gives access to IBM Granite and open-source foundation models with tools for prompt engineering and tuning.

Question 104

What is IBM watsonx.data?

AA chatbot
BA data store built on an open lakehouse architecture for AI workloads ✅
CA spreadsheet
DA translation service
💡 ExplanationIt lets organisations manage data across clouds for analytics and AI.

Question 105

What is IBM watsonx.governance?

AA legal firm
BA toolkit to monitor, manage and govern AI models for risk, bias and compliance ✅
CA cloud server
DA chatbot builder
💡 ExplanationIt tracks model lineage, bias and drift so AI stays transparent and compliant.

Question 106

What are IBM Granite models?

ARock samples
BIBM’s family of open, enterprise-ready foundation models for language, code and more ✅
CHardware chips
DSpreadsheet templates
💡 ExplanationGranite models are trained on curated business data and offered through watsonx.ai.

Question 107

What is IBM Watson Assistant (watsonx Assistant) primarily designed to do?

AStore files
BBuild AI-powered chatbots and virtual agents for customer service ✅
CAnalyse images
DManage servers
💡 ExplanationIt uses NLP to understand intents and entities and respond across channels.

Question 108

What is IBM Watson Discovery?

AA travel app
BAn AI search and text-analytics service that finds insights in large collections of documents ✅
CA chat app
DA storage drive
💡 ExplanationDiscovery can read contracts, manuals and reports to answer questions.

Question 109

What is IBM Watson Studio?

AA music studio
BA platform for data scientists to build, train and deploy machine-learning models ✅
CA video editor
DAn email client
💡 ExplanationWatson Studio includes AutoAI, notebooks and tools for the full model lifecycle.

Question 110

What is IBM AutoAI in Watson Studio?

AA tool that writes Python for you
BA tool that automatically prepares data, selects algorithms and builds the best ML pipeline with minimal manual effort ✅
CA customer-support chatbot
DAn app deployer
💡 ExplanationAutoAI handles feature engineering, algorithm selection and hyperparameter tuning so non-experts can build models.

Question 111

In the SkillsBuild lab ‘Run AI models with IBM Watson Studio’, learners use AutoAI to:

AWrite code from scratch
BUpload a dataset and automatically generate and compare machine-learning pipelines ✅
CDesign a website
DTranslate text
💡 ExplanationThe lab shows how AutoAI produces ranked pipelines and lets you deploy the best one.

Question 112

What does Watson Natural Language Understanding (NLU) do?

ATranslates speech
BExtracts sentiment, entities, keywords, categories and emotion from text ✅
CGenerates images
DStores documents
💡 ExplanationNLU is used to analyse reviews, support tickets and social media.

Question 113

What is Watson Speech to Text?

AA typing tool
BA service that converts spoken audio into written text ✅
CA text translator
DA document scanner
💡 ExplanationUsed for call-centre transcription and voice interfaces.

Question 114

What is IBM Watson OpenScale (now part of watsonx.governance)?

AA weighing device
BA tool for monitoring deployed AI models for fairness, bias, drift and explainability ✅
CA database
DA spreadsheet add-in
💡 ExplanationOpenScale was IBM’s AI monitoring tool; its capabilities now live in watsonx.governance.

Question 115

What is IBM watsonx Orchestrate?

AA music app
BA tool that uses AI agents and automation to carry out business tasks across apps ✅
CA data store
DA chip design
💡 ExplanationOrchestrate automates workflows in HR, sales and procurement using AI assistants.

Question 116

What is IBM Cloud Pak for Data?

AA packaging service
BAn integrated data and AI platform that runs on any cloud ✅
CA payment app
DA photo library
💡 ExplanationIt bundles Watson services and data tools on Red Hat OpenShift.

Question 117

IBM Watson Health was known for applying AI to:

AGaming
BHealthcare, including oncology and clinical research ✅
CRetail checkout
DAviation
💡 ExplanationIBM sold Watson Health assets to Francisco Partners in 2022; the course still references its healthcare use cases.

Question 118

What is an ‘API’ in the context of Watson services?

AA type of AI
BAn interface that lets developers call Watson capabilities from their own applications ✅
CA password
DA chatbot
💡 ExplanationWatson services are consumed through REST APIs and SDKs.

Question 119

What is ‘model deployment’?

ADeleting a model
BMaking a trained model available so applications can use it to make predictions ✅
CTraining a model again
DDrawing a model diagram
💡 ExplanationIn Watson Studio, deployment publishes the model as an endpoint.

Question 120

What is ‘model drift’?

AA model moving between servers
BDecline in model accuracy over time as real-world data changes from the training data ✅
CA faster model
DA visual effect
💡 Explanationwatsonx.governance monitors drift and alerts teams to retrain.

Question 121

IBM’s open-source toolkit for detecting and mitigating bias in machine learning is called:

AAI Fairness 360 ✅
BWatson Bias
CGranite Fair
DOpenBias
💡 ExplanationAI Fairness 360 (AIF360) is IBM’s open-source library of fairness metrics and algorithms.

Question 122

IBM’s open-source toolkit for explaining AI model decisions is called:

AWatson Explain
BAI Explainability 360 ✅
CGranite XAI
DOpenScale 360
💡 ExplanationAI Explainability 360 provides algorithms to interpret model predictions.

Next: AI Ethics and Responsible AI →

⚖️

Topic 6 — AI Ethics and Responsible AI (Questions 123–147)

IBM’s principles, the five pillars of trust, bias, privacy and governance

Question 123

What does IBM consider the central purpose of AI, according to its Principles for Trust and Transparency?

ATo replace human workers
BTo augment human intelligence ✅
CTo maximise profit
DTo automate everything
💡 ExplanationIBM’s first principle: the purpose of AI is to augment human intelligence, not replace it.

Question 124

According to IBM’s principles, data and insights generated from it belong to:

AIBM
BThe government
CTheir creator (the client) ✅
DNobody
💡 ExplanationIBM’s second principle: data and insights belong to their creator; IBM does not use client data to train its models without permission.

Question 125

IBM’s third Principle for Trust and Transparency states that AI systems must be:

ASecret and proprietary
BTransparent and explainable ✅
CFast and cheap
DFully autonomous
💡 ExplanationPeople must be able to understand how and why AI reaches its decisions.

Question 126

Which of the following are IBM’s five pillars of trustworthy AI?

ASpeed, cost, size, power, scale
BExplainability, fairness, robustness, transparency, privacy ✅
CAccuracy, profit, growth, security, design
DAutomation, efficiency, speed, data, cloud
💡 ExplanationThese five pillars underpin IBM’s approach to AI ethics and are heavily tested.

Question 127

Which of the following is NOT one of IBM’s pillars of trustworthy AI?

AFairness
BRobustness
CProfitability ✅
DTransparency
💡 ExplanationProfitability is not an ethics pillar. The five are explainability, fairness, robustness, transparency and privacy.

Question 128

Which principle is NOT part of IBM’s AI ethics framework?

AExplainability
BPrivacy
CFairness
DHuman replacement ✅
💡 ExplanationIBM’s framework is built on augmenting humans, not replacing them.

Question 129

What does ‘explainable AI’ (XAI) mean?

AAI that explains jokes
BAI whose decisions can be understood and interpreted by humans ✅
CAI that needs no data
DAI with a user manual
💡 ExplanationExplainability is essential in high-stakes areas such as lending and healthcare.

Question 130

What does the ‘fairness’ pillar address?

AEqual pay for AI engineers
BEnsuring AI treats individuals and groups equitably and does not discriminate ✅
CFair pricing of software
DEqual server loads
💡 ExplanationFairness requires detecting and mitigating bias in data and models.

Question 131

What does the ‘robustness’ pillar mean?

AThe AI is physically strong
BThe AI performs reliably and resists manipulation, attacks and unexpected inputs ✅
CThe AI is expensive
DThe AI uses large servers
💡 ExplanationRobust AI is secure against adversarial attacks and handles edge cases safely.

Question 132

What does the ‘transparency’ pillar require?

AClear screens
BOpenness about how the AI is built, what data it uses and how it works ✅
CFree software
DPublic source code for everything
💡 ExplanationUsers should know when they are interacting with AI and how it was developed.

Question 133

What does the ‘privacy’ pillar require?

AHiding the AI
BProtecting personal data and respecting users’ data rights throughout the AI lifecycle ✅
CPrivate servers only
DNo data collection ever
💡 ExplanationAI must comply with data-protection laws such as GDPR.

Question 134

What is ‘algorithmic bias’?

AA bug in the code
BSystematic unfair outcomes produced by an AI system, often from biased training data ✅
CA slow algorithm
DA preference for one programming language
💡 ExplanationExample: a hiring model trained on past hires that favoured one gender.

Question 135

Who, according to IBM’s AI ethics framework, is ultimately responsible for an AI system’s decisions?

AThe AI itself
BThe humans and organisations that design, deploy and use the system ✅
CThe cloud provider
DNobody
💡 ExplanationAccountability stays with people; AI is a tool.

Question 136

What is ‘AI governance’?

AGovernment-run AI
BThe policies, processes and tools that ensure AI is developed and used responsibly ✅
CA type of neural network
DAI voting systems
💡 ExplanationIBM’s AI Ethics Board and watsonx.governance support governance.

Question 137

Who co-chairs IBM’s AI Ethics Board?

AArvind Krishna and Ginni Rometty
BFrancesca Rossi and Christina Montgomery ✅
CAlan Turing and John McCarthy
DSam Altman and Satya Nadella
💡 ExplanationThe board is led by IBM’s AI Ethics Global Leader Francesca Rossi and Chief Privacy & Trust Officer Christina Montgomery.

Question 138

Why did IBM announce in 2020 that it would stop offering general-purpose facial-recognition software?

AIt was unprofitable
BConcerns about bias, mass surveillance and human-rights violations ✅
CLack of cameras
DCompetition from Apple
💡 ExplanationIBM’s CEO wrote to the US Congress citing racial profiling and civil-liberty concerns.

Question 139

What is a ‘human-in-the-loop’ approach?

AA human standing in a circle
BKeeping a human involved in AI decisions, especially high-stakes ones, to review or override outputs ✅
CA loop in code
DA robot that looks human
💡 ExplanationIt supports accountability and catches errors the AI makes.

Question 140

What is ‘data privacy’ in AI?

AHiding the dataset from the model
BProtecting individuals’ personal information from unauthorised access and misuse ✅
CPrivate clouds
DEncrypting the model
💡 ExplanationTechniques include anonymisation, consent and data minimisation.

Question 141

What is a ‘deepfake’?

AA deep neural network
BAI-generated fake video, audio or images that appear real ✅
CA hidden layer
DA strong password
💡 ExplanationDeepfakes raise concerns about misinformation and consent.

Question 142

Which of the following is an example of an ethical risk of AI in hiring?

AFaster CV screening
BA model rejecting candidates based on gender or ethnicity patterns in past data ✅
CScheduling interviews
DSending confirmation emails
💡 ExplanationThis is algorithmic bias, which fairness tools like AI Fairness 360 aim to detect.

Question 143

What is the EU AI Act?

AA film
BThe European Union’s regulation that classifies AI systems by risk and sets rules for them ✅
CAn IBM product
DA programming standard
💡 ExplanationIt is the first comprehensive AI law; high-risk systems face strict obligations.

Question 144

What does GDPR primarily regulate?

AGaming
BPersonal-data protection and privacy in the EU ✅
CAI hardware
DCloud pricing
💡 ExplanationGDPR affects how AI systems collect and process personal data.

Question 145

What is ‘informed consent’ in data collection?

ACollecting data silently
BPeople understanding and agreeing to how their data will be used ✅
CA legal contract with IBM
DData stored with a password
💡 ExplanationConsent is a cornerstone of privacy and trustworthy AI.

Question 146

Why is diverse representation in AI development teams important?

AFor marketing photos
BIt helps identify biases and blind spots that a homogeneous team might miss ✅
CIt lowers costs
DIt is required by IBM Cloud
💡 ExplanationDiverse teams build fairer systems.

Question 147

Which statement reflects IBM’s view on AI and jobs?

AAI will eliminate all jobs
BAI will change jobs, automating tasks while creating new roles that need new skills ✅
CAI has no effect on jobs
DOnly engineers will have jobs
💡 ExplanationThe course’s final module focuses on preparing for an AI-shaped job market.

Next: Your Future in AI: The Job Landscape →

🚀

Topic 7 — Your Future in AI: The Job Landscape (Questions 148–159)

AI careers, skills, SkillsBuild and the digital badge

Question 148

Which job role focuses on collecting, cleaning and analysing data to find insights?

AData scientist ✅
BGraphic designer
CNetwork engineer
DAccountant
💡 ExplanationData scientists use statistics and machine learning to turn data into decisions.

Question 149

Which role builds and maintains the systems that train and deploy machine-learning models?

AMachine learning engineer ✅
BHR manager
CCopywriter
DSales executive
💡 ExplanationML engineers bridge data science and software engineering.

Question 150

A ‘prompt engineer’ specialises in:

ABuilding hardware
BCrafting inputs for generative AI to produce reliable, high-quality outputs ✅
CRepairing printers
DWriting legal contracts
💡 ExplanationIt is one of the new roles created by generative AI.

Question 151

Which role ensures AI systems comply with ethics principles and regulations?

AAI ethicist / AI governance specialist ✅
BTruck driver
CChef
DPhotographer
💡 ExplanationDemand for AI governance skills is rising with laws like the EU AI Act.

Question 152

Which of these is a ‘durable’ skill valuable in an AI-driven workplace?

AMemorising facts
BCritical thinking and problem solving ✅
CTyping speed
DKnowing one software version
💡 ExplanationHuman skills such as creativity, communication and judgement complement AI.

Question 153

What does ‘upskilling’ mean?

AClimbing a hill
BLearning new skills to keep up with changes in your job or industry ✅
CGetting promoted
DChanging careers entirely
💡 ExplanationIBM SkillsBuild offers free courses for upskilling in AI and technology.

Question 154

What is IBM SkillsBuild?

AA construction company
BIBM’s free online learning platform offering courses and digital credentials in technology and workplace skills ✅
CA paid certification body
DA job-search website
💡 ExplanationThe AI Fundamentals course and badge are delivered through SkillsBuild.

Question 155

What do you receive after completing the IBM AI Fundamentals course?

AA university degree
BA digital badge issued via Credly that you can share on LinkedIn ✅
CA cash reward
DA job at IBM
💡 ExplanationIBM digital credentials are verified badges hosted by Credly.

Question 156

Which tasks are most likely to be automated by AI?

ARepetitive, rule-based tasks ✅
BCreative leadership
CEmotional counselling
DStrategic negotiation
💡 ExplanationRoutine tasks are easiest to automate; human judgement remains valuable.

Question 157

What is a ‘citizen data scientist’?

AA government employee
BA non-specialist who uses low-code tools like AutoAI to build models ✅
CA data scientist with citizenship
DA statistics professor
💡 ExplanationTools like AutoAI let business users create models without deep coding.

Question 158

Which industry is NOT being transformed by AI?

AHealthcare
BFinance
CManufacturing
DNone; AI affects virtually every industry ✅
💡 ExplanationAI applications span every sector, from agriculture to entertainment.

Question 159

The best way to prepare for an AI-influenced career, according to the course, is to:

AAvoid technology
BBuild AI literacy, keep learning and develop skills that complement AI ✅
CWait for AI to replace you
DStudy only one tool
💡 ExplanationLifelong learning and adaptability are the course’s closing message.

🔬 IBM AI Tools Quick Reference

🤖 watsonx Assistant
AI chatbots and virtual agents
🔍 Watson Discovery
AI-powered document search and insights
🧪 Watson Studio + AutoAI
Build, train and deploy ML models
🌐 watsonx.ai
Foundation models incl. IBM Granite
🗄️ watsonx.data
Open lakehouse data store for AI
🛡️ watsonx.governance
Bias, drift, risk and compliance monitoring
🗣️ Watson Speech to Text / Text to Speech
Voice interfaces and transcription
🧩 watsonx Orchestrate
AI agents that automate business tasks

⚖️ IBM AI Ethics – what the final exam asks

Three Principles for Trust and Transparency: (1) the purpose of AI is to augment human intelligence; (2) data and insights belong to their creator; (3) AI systems must be transparent and explainable.

Five Pillars of Trustworthy AI: Explainability · Fairness · Robustness · Transparency · Privacy. Any option such as “profitability”, “speed” or “human replacement” is not a pillar.

💡 Top Tips to Pass IBM AI Fundamentals in 2026

🆓
Free on IBM SkillsBuild
The course, quizzes and badge are completely free; you can retake the final assessment
🔬
Know the Watson tools
Match each Watson / watsonx product to its use case; Topic 5 covers them all
⚖️
Memorise the 5 pillars
IBM’s ethics principles and pillars are the most repeated final-exam questions

📥 Download all 159+ IBM AI Fundamentals MCQs as PDF (free)

🎯 More AI certification MCQs

Question banks updated regularly to match the latest course content

IBM AI Engineering Professional Certificate Quiz Answers

IBM AI Fundamentals MCQs with Answers 2026 – Infographic

Frequently Asked Questions

Is the IBM AI Fundamentals certification free?

Yes — the IBM AI Fundamentals course and digital badge are completely free through IBM SkillsBuild. You simply create a free account, complete the learning modules, pass the assessment, and earn your shareable digital badge instantly.

How long does the IBM AI Fundamentals course take?

The IBM AI Fundamentals course typically takes 6 to 8 hours to complete. It is entirely self-paced so you can spread it across several days or complete it in a single focused session depending on your schedule.

Is the IBM AI badge recognized by employers?

Yes — IBM digital badges are globally recognized Credly-verified credentials. They carry IBM’s brand reputation and are widely accepted by Fortune 500 companies, technology firms, and government organizations as proof of foundational AI knowledge.

Do I need programming skills for IBM AI Fundamentals?

No programming knowledge is required. The IBM AI Fundamentals certification is designed for everyone — including business professionals, students, and career changers — who want to build foundational AI knowledge without writing any code.

About the author

MCQS TOP

4 Comments

  • […] 🤖 IBM AI Fundamentals150+ MCQs by module: intro to AI, ML and deep learning, NLP, computer vision…✨ IBM Generative AI Fundamentals75+ MCQs: foundation models, LLMs, prompt engineering, use cases, risks, watsonxSkillsBuild badge · beginnerOpen answers →🌐 AI Fundamentals with IBM SkillsBuild (Cisco NetAcad)Module-by-module quiz answers and final exam practice for the Cisco Networking Academy versionNetAcad course · same IBM badgeOpen answers →🛠️ IBM AI Engineering Professional Certificate100+ MCQs: ML with Python, Keras, TensorFlow, PyTorch, LLMs, fine-tuning, RAG and LangChainCoursera · 13 courses · intermediateOpen answers →📈 IBM Machine Learning Professional Certificate90+ MCQs: EDA, regression, classification, clustering, PCA, deep learning, reinforcement learningCoursera · 6 courses · intermediateOpen answers → […]

Leave a Comment