The
Google AI Essentials certificate on Coursera covers five modules: Introduction to AI, Maximize Productivity With AI Tools, Discover the Art of Prompting, Use AI Responsibly, and Stay Ahead of the AI Curve, each ending in a graded quiz. The 118+ solved MCQs below follow the course module by module, with the correct option marked and a short explanation under every question, plus a mixed final-exam practice set. The prompting module’s TCREI framework (Task, Context, References, Evaluate, Iterate) is the most tested idea and has its own section. A
free PDF of the full set is at the end of the page.
ย
๐ค
Module 1 โ Introduction to AI (Questions 1โ24)
Foundations of AI, machine learning types, generative AI and Google tools
Question 01
What is Artificial Intelligence (AI)?
AA robot that performs physical tasks
BA set of technologies that enable computers to perform tasks that typically require human intelligence โ
CA type of software used only for gaming
DA hardware component inside computers
๐ก ExplanationThe course defines AI as computer programs that complete cognitive tasks typically associated with human intelligence, such as learning, reasoning and problem solving.
Question 02
What does the term ‘machine learning’ mean?
ATeaching robots to walk
BA branch of AI where systems learn patterns from data and improve without being explicitly programmed โ
CProgramming a machine to follow fixed instructions
DInstalling software updates automatically
๐ก ExplanationML lets systems identify patterns in large amounts of data and make predictions or decisions from them.
Question 03
Which of the following is NOT a type of machine learning?
ASupervised learning
BUnsupervised learning
CReinforcement learning
DInstructed learning โ
๐ก ExplanationThe course covers supervised, unsupervised and reinforcement learning; ‘instructed learning’ is not a category.
Question 04
In supervised learning, the model is trained on:
AUnlabelled data
BLabelled data with known answers โ
CRewards and penalties
DRandom noise
๐ก ExplanationExample from the course: training a model on labelled images of cats and dogs so it can classify new images.
Question 05
In unsupervised learning, the model:
ANeeds labelled examples
BFinds patterns and groupings in unlabelled data โ
CLearns from rewards
DCannot learn at all
๐ก ExplanationClustering customers by purchasing behaviour is a typical example.
Question 06
Reinforcement learning trains a model by:
AGiving it labelled examples
BRewarding desired actions and penalising undesired ones โ
CShowing it unlabelled data
DCopying another model
๐ก ExplanationThe model learns through trial and error to maximise reward, as in game-playing AI.
Question 07
What is generative AI?
AAI that only sorts data
BAI that creates new content such as text, images, audio or code โ
CAI that controls robots
DA spreadsheet tool
๐ก ExplanationGen AI, the focus of the course, produces new outputs based on patterns in its training data.
Question 08
What is the relationship between AI, machine learning and deep learning?
AThey are unrelated
BDeep learning is a subset of machine learning, which is a subset of AI โ
CAI is a subset of machine learning
DMachine learning is a subset of deep learning
๐ก ExplanationAI is the broad field; ML is one approach within it; deep learning is ML using multi-layered neural networks.
Question 09
Which of the following best describes a neural network?
AA computer network of servers
BA model with layers of interconnected nodes inspired by the human brain that learns patterns in data โ
CA social network for AI
DA type of database
๐ก ExplanationNeural networks are the basis of deep learning and modern generative models.
Question 10
What is a ‘foundation model’?
AA small single-purpose model
BA large model trained on vast data that can be adapted to many tasks โ
CA model used in construction
DA model that cannot be changed
๐ก ExplanationGemini is built on a foundation model; such models underpin most gen AI tools.
Question 11
What is a large language model (LLM)?
AA dictionary
BA type of foundation model trained on huge amounts of text to understand and generate language โ
CA grammar app
DA translation table
๐ก ExplanationLLMs predict likely sequences of words, which lets them answer questions, summarise and write.
Question 12
Which Google product is powered by a large language model?
AGoogle Maps
BGoogle Drive
CGemini โ
DGoogle Calendar
๐ก ExplanationGemini is Google’s AI assistant built on Gemini models.
Question 13
What does ‘multimodal AI’ mean?
AAI that runs on many devices
BAI that can process and generate more than one type of data, such as text, images and audio โ
CAI with many users
DAI with multiple servers
๐ก ExplanationGemini can take an image and text as input and respond in text, for example.
Question 14
In AI terminology, what does ‘inference’ mean?
ATraining the model
BThe process of a trained model producing an output from new input โ
CCollecting data
DLabelling data
๐ก ExplanationEvery response a chatbot gives is an inference.
Question 15
What is ‘training data’?
AThe output of a model
BThe data a model learns from during development โ
CA user’s prompt
DThe model’s settings
๐ก ExplanationThe quality and scope of training data shape what a model can do and its biases.
Question 16
Which is an example of AI you might use every day?
AA mechanical calculator
BEmail spam filtering and smartphone autocorrect โ
CA paper map
DA printed dictionary
๐ก ExplanationThe course opens with everyday examples: recommendations, navigation, spam filters.
Question 17
How does generative AI differ from traditional software?
AIt runs faster
BTraditional software follows explicit rules; gen AI learns patterns from data and produces new, non-deterministic outputs โ
CIt needs no computer
DThere is no difference
๐ก ExplanationThe same prompt can produce different outputs because gen AI is probabilistic.
Question 18
What is the difference between AI and machine learning?
AThey are the same
BAI is the broad goal of intelligent machines; ML is a method of achieving it by learning from data โ
CML is broader than AI
DAI only means robots
๐ก ExplanationML is one of several approaches within the wider field of AI.
Question 19
What is ‘transfer learning’?
AMoving a model between computers
BApplying knowledge a model learned on one task to a new, related task โ
CTransferring data to the cloud
DLearning without data
๐ก ExplanationIt is why foundation models can be adapted quickly to new uses.
Question 20
What is a ‘parameter’ in an AI model?
AA user-defined setting
BA value the model learns during training that determines its behaviour โ
CThe length of a prompt
DA type of data
๐ก ExplanationLLMs have billions of parameters.
Question 21
Which of these tasks is generative AI good at?
ADrafting an email from bullet points โ
BPhysically moving objects
CReplacing human judgement entirely
DGuaranteeing factual accuracy
๐ก ExplanationGen AI excels at drafting, summarising, brainstorming and translating; it still needs human review.
Question 22
What is ‘human-in-the-loop’?
AA fitness programme
BKeeping a person involved to review, approve or correct AI outputs โ
CA looping prompt
DAn AI that acts like a human
๐ก ExplanationThe course stresses that AI should augment human work, with people making final decisions.
Question 23
Which statement about AI is TRUE?
AAI understands the world like a human
BAI finds patterns in data; it does not have understanding or intent โ
CAI is always correct
DAI needs no data
๐ก ExplanationRecognising this limitation helps you use AI tools appropriately.
Question 24
What is ‘augmentation’ in the context of AI at work?
AReplacing workers
BUsing AI to enhance human abilities and speed up tasks while people stay in control โ
CAdding more servers
DIncreasing screen size
๐ก ExplanationGoogle frames AI as a tool to augment, not replace, human skills.
Next: Maximize Productivity With AI Tools โ
ย
โก
Module 2 โ Maximize Productivity With AI Tools (Questions 25โ44)
Writing, summarising, brainstorming and planning with Gemini and Workspace
Question 25
Which of the following is a productivity task AI tools can help with?
ADrafting documents and emails
BSummarising long reports
CBrainstorming ideas
DAll of the above โ
๐ก ExplanationModule 2 covers writing, summarising, brainstorming, planning and analysing with AI.
Question 26
What is ‘Gemini for Workspace’?
AA video game
BAI features built into Google Docs, Gmail, Sheets and Slides to help write, organise and summarise โ
CA new operating system
DA spreadsheet formula
๐ก ExplanationExamples: ‘Help me write’ in Docs and Gmail, ‘Help me organise’ in Sheets.
Question 27
A manager wants a first draft of a project update email. Which AI approach fits?
AAsk the AI to send it directly to the team
BGive the AI the key points and audience, generate a draft, then edit it โ
CAvoid AI entirely
DLet the AI decide the content
๐ก ExplanationDraft with AI, review and personalise before sending.
Question 28
Which is an appropriate use of AI for data in a spreadsheet?
AAsking it to generate a formula or summarise trends in the data โ
BLetting it change financial records unsupervised
CUsing it to delete data
DNothing; AI cannot work with tables
๐ก ExplanationAI can suggest formulas, categorise entries and explain trends; you verify the results.
Question 29
What is a good practice when using AI to summarise a document?
ATrust the summary without reading
BCheck the summary against the source for omissions and errors โ
CShare it immediately
DUse only the first sentence
๐ก ExplanationSummaries can miss nuance or hallucinate; verification is part of the workflow.
Question 30
Which of these is NOT a strength of current AI tools?
AGenerating many ideas quickly
BRephrasing text for a different tone
CKnowing your company’s private, unshared information โ
DTranslating between languages
๐ก ExplanationAI only knows what it was trained on or what you give it in the prompt.
Question 31
How can AI help with brainstorming?
ABy making the final decision
BBy generating a wide range of ideas you can refine and select from โ
CBy replacing team meetings
DIt cannot help
๐ก ExplanationUse AI for volume and variety, then apply human judgement.
Question 32
What is the best way to get a tone-appropriate email from an AI tool?
AGive no instructions
BSpecify the audience, purpose and tone in the prompt โ
CUse all capital letters
DAsk for a long essay
๐ก ExplanationContext about audience and tone is one of the prompt elements taught in the course.
Question 33
Which Google tool can turn a text prompt into an image?
AGoogle Sheets
BImagen (in Gemini and Workspace) โ
CGoogle Calendar
DGoogle Forms
๐ก ExplanationImagen powers image generation in Gemini and ‘Help me visualise’ in Slides.
Question 34
What is NotebookLM?
AA paper notebook
BA Google AI tool that answers questions and summarises using documents you upload โ
CA laptop model
DA coding language
๐ก ExplanationNotebookLM grounds its answers in your own sources, reducing hallucination.
Question 35
A team uses AI to transcribe and summarise meetings. What should they do first?
ANothing
BMake sure participants know and consent, and check the company’s AI policy โ
CDelete the recording
DPublish the summary externally
๐ก ExplanationResponsible use includes transparency and following organisational policies.
Question 36
Which workflow is recommended when using AI for a task?
APrompt once and accept the result
BPrompt, evaluate the output, refine the prompt, repeat until satisfied โ
COnly use AI at the end
DNever edit AI output
๐ก ExplanationThe course calls this iterative, ‘human-in-the-loop’ working.
Question 37
What is a limitation to remember when AI drafts a legal or financial document?
AIt writes too slowly
BIt may be wrong, outdated or non-compliant and must be checked by a qualified person โ
CIt cannot write in English
DIt is always correct
๐ก ExplanationHigh-stakes documents need expert review.
Question 38
How can AI help with learning a new skill at work?
ABy creating practice questions, explanations and study plans โ
BBy taking the exam for you
CBy replacing training entirely
DIt cannot help
๐ก ExplanationThe course gives examples of AI as a tutor for upskilling.
Question 39
Which prompt is best for scheduling help?
A‘Help’
B‘Create a weekly plan for a 3-person team to launch a newsletter in 4 weeks, with milestones’ โ
C‘Plan’
D‘Do my work’
๐ก ExplanationSpecific, contextual prompts produce useful plans.
Question 40
AI-generated code should be:
ADeployed immediately
BReviewed and tested by a developer before use โ
CIgnored
DTrusted blindly
๐ก ExplanationCode assistants speed up work but can introduce bugs or insecure patterns.
Question 41
What is the benefit of using AI inside the apps you already use (like Docs and Gmail)?
AIt is slower
BIt reduces switching between tools and keeps work in context โ
CIt removes the need to read
DThere is no benefit
๐ก ExplanationIntegrated AI features fit into existing workflows.
Question 42
Which is an example of AI improving accessibility at work?
ALive captions and automatic transcription โ
BFaster printers
CBigger monitors
DLonger meetings
๐ก ExplanationAI captioning and summarisation help colleagues with hearing or attention needs.
Question 43
When AI output is good but not perfect, you should:
ADiscard it entirely
BEdit and improve it; AI output is a starting point โ
CSend it anyway
DAsk AI to send it
๐ก ExplanationTreat AI as a collaborator whose drafts you refine.
Question 44
What does ‘productivity’ gain from AI mostly come from?
AReplacing all tasks
BReducing time on repetitive or first-draft work so people focus on higher-value tasks โ
CLonger working hours
DFewer meetings only
๐ก ExplanationThe course emphasises freeing time for judgement, creativity and relationships.
Next: Discover the Art of Prompting โ
ย
โ๏ธ
Module 3 โ Discover the Art of Prompting (Questions 45โ66)
The TCREI framework, zero/few-shot, chain-of-thought, iteration and multimodal prompts
Question 45
What is a ‘prompt’ in the context of generative AI?
AA hardware button
BThe text, image or other input you give an AI tool to get a response โ
CThe AI’s answer
DA type of password
๐ก ExplanationPrompting is the main skill the course teaches in Module 3.
Question 46
What is ‘prompt engineering’?
ABuilding prompt hardware
BDesigning and refining prompts to get the best output from an AI tool โ
CWriting software code
DFixing AI bugs
๐ก ExplanationClear, specific prompts with context and examples give better results.
Question 47
What does ‘iterative prompting’ mean?
AAsking the same question repeatedly
BRefining your prompt step by step based on the output you get โ
CPrompting several AIs at once
DUsing numbered prompts
๐ก ExplanationIteration is expected; the first output is rarely the final one.
Question 48
According to the course, which framework helps you write effective prompts?
AABCD
BTask, Context, References, Evaluate, Iterate (the TCREI framework) โ
CPROMPT
DAI-5
๐ก ExplanationGoogle’s AI Essentials teaches: specify the Task, give Context, add References, Evaluate the output, Iterate.
Question 49
In the TCREI framework, what does ‘Task’ mean?
AThe AI’s job title
BWhat you want the AI to do, including persona and output format โ
CThe deadline
DThe device used
๐ก ExplanationExample: ‘Act as a marketing expert and write three slogans in a table.’
Question 50
In the TCREI framework, ‘Context’ refers to:
AThe AI’s training data
BBackground details that help the AI tailor its response, such as audience and purpose โ
CThe weather
DThe prompt length
๐ก ExplanationContext narrows the output to what you actually need.
Question 51
In the TCREI framework, ‘References’ means:
ACitations in the output
BExamples or samples you include so the AI understands the style or format you want โ
CLinks to Google
DThe AI’s memory
๐ก ExplanationProviding a sample email, for instance, helps the AI match your tone.
Question 52
What should you do in the ‘Evaluate’ step?
ATrust the output
BCheck the output for accuracy, relevance, bias and whether it meets the task โ
CShare it immediately
DDelete the prompt
๐ก ExplanationEvaluation is where human judgement enters.
Question 53
What is ‘zero-shot’ prompting?
AA prompt with no examples โ
BA prompt with one example
CA prompt with many examples
DA blank prompt
๐ก ExplanationThe model relies only on the instruction.
Question 54
What is ‘few-shot’ prompting?
AA prompt with no instructions
BA prompt that includes several examples to guide the output โ
CPrompting a few times
DA short prompt
๐ก ExplanationExamples improve consistency of format and style.
Question 55
What is ‘chain-of-thought’ prompting?
ALinking many chats
BAsking the AI to explain its reasoning step by step before answering โ
CRepeating prompts
DA prompt about chains
๐ก ExplanationIt improves answers to reasoning and maths problems.
Question 56
What is a ‘persona’ in a prompt?
AThe user’s name
BA role you ask the AI to adopt, such as ‘act as a financial analyst’ โ
CThe AI’s mood
DA password
๐ก ExplanationPersonas shape tone, vocabulary and expertise.
Question 57
What is ‘prompt chaining’?
AUsing many AIs
BBreaking a complex task into a sequence of prompts, each building on the last โ
CRepeating one prompt
DA security attack
๐ก ExplanationExample: outline first, then draft each section, then refine the tone.
Question 58
Which prompt is most effective?
A‘Write something about marketing’
B‘Write a 150-word LinkedIn post announcing our new eco-friendly water bottle to young professionals, in an upbeat tone’ โ
C‘Marketing’
D‘Help’
๐ก ExplanationSpecific task, audience, length and tone.
Question 59
What is the purpose of a ‘system prompt’ when using tools like Gemini?
ATo restart the computer
BTo set overall instructions, role or constraints that apply to the whole conversation โ
CTo delete chat history
DTo change the language
๐ก ExplanationSystem-level instructions persist across turns.
Question 60
What is ‘multimodal prompting’?
APrompting in many languages
BIncluding images, audio or files along with text in a prompt โ
CPrompting many models
DPrompting many times
๐ก ExplanationExample: uploading a chart and asking Gemini to explain it.
Question 61
Why should you provide an output format in the prompt?
AIt is required by law
BIt helps the AI deliver results you can use directly, like a table or bullet list โ
CIt makes the AI faster
DIt is not useful
๐ก Explanation‘Give the answer as a table with three columns’ saves editing time.
Question 62
What is ‘hallucination’ and how does prompting relate to it?
AAn AI error that prompting cannot affect
BFalse output; clear context, references and asking for sources reduce it โ
CA feature of good prompts
DA hardware fault
๐ก ExplanationGrounded prompts and verification are the course’s answer to hallucination.
Question 63
Which is an example of a good iteration after a weak output?
AGiving up
BAdding more context, a reference example or a clearer format request โ
CTyping the same prompt again
DSwitching off the computer
๐ก ExplanationEach iteration should change something specific.
Question 64
What is ‘tree-of-thought’ prompting?
APrompting about trees
BAsking the AI to explore several reasoning paths before choosing the best โ
CA prompt with branches drawn
DA tree diagram tool
๐ก ExplanationAn advanced technique for complex problems.
Question 65
Why include the intended audience in a prompt?
AIt is optional decoration
BIt lets the AI adjust complexity, tone and vocabulary for the reader โ
CAudience does not matter
DIt makes the prompt longer only
๐ก ExplanationWriting for executives differs from writing for new hires.
Question 66
What is ‘AI agent’ prompting in the course’s advanced section?
APrompting a human agent
BSetting up an AI to simulate a scenario or role, such as a practice interviewer โ
CPrompting a spy
DAgent-based printing
๐ก ExplanationAgents can be set up as simulators or expert feedback givers.
Next: Use AI Responsibly โ
ย
๐ก๏ธ
Module 4 โ Use AI Responsibly (Questions 67โ86)
Hallucination, bias, privacy, security, transparency and accountability
Question 67
What is an ‘AI hallucination’?
AA visual glitch
BWhen an AI confidently produces false or made-up information โ
CWhen the AI refuses to answer
DWhen the AI becomes conscious
๐ก ExplanationHallucinations are a core limitation discussed in the Responsible AI module.
Question 68
What does it mean to say an AI model is ‘biased’?
AIt has political views
BIts outputs systematically favour or disadvantage certain groups because of patterns in training data โ
CIt runs on one computer
DIt is slow
๐ก ExplanationBias enters through data, design and use.
Question 69
What is the best way to verify whether AI-generated information is accurate?
AAsk the same AI again
BCross-check with reliable, independent sources and subject-matter experts โ
CTrust it if it sounds confident
DCheck the word count
๐ก ExplanationVerification is a required step, especially for facts, figures and citations.
Question 70
What should you consider before sharing sensitive work data with an AI tool?
ANothing; AI tools are always private
BYour organisation’s data and privacy policies, and whether the tool stores or trains on your input โ
CWhether the AI is fast
DThe colour of the interface
๐ก ExplanationNever paste confidential or personal data into tools not approved for it.
Question 71
Which is a principle of responsible AI use?
AHide that AI was used
BBe transparent about AI use, review outputs and take accountability โ
CLet AI make all decisions
DIgnore bias
๐ก ExplanationTransparency, accountability and human oversight run through the module.
Question 72
What is ‘drift’ in the context of AI models?
AA model moving between servers
BA model’s performance changing over time as the real world changes from its training data โ
CA design style
DA prompt error
๐ก ExplanationModels need monitoring and updating.
Question 73
How can businesses use AI responsibly in decisions that affect people?
ALet AI decide hiring alone
BKeep humans in the loop, audit for bias and explain decisions โ
CHide the AI’s role
DUse AI only at night
๐ก ExplanationHigh-impact decisions require human review.
Question 74
What is ‘privacy’ concern with AI tools?
AAI can read minds
BInputs may be stored, reviewed or used to train models, exposing personal or confidential data โ
CAI has no privacy issues
DPrivacy only matters for images
๐ก ExplanationCheck each tool’s data policy.
Question 75
What is a ‘deepfake’?
AA deep neural layer
BAI-generated fake media that depicts people saying or doing things they did not โ
CA strong password
DA fake dataset
๐ก ExplanationDeepfakes raise misinformation and consent concerns.
Question 76
What should you do if an AI tool produces harmful or offensive content?
AShare it
BStop, do not use it, report it through the tool’s feedback, and refine or abandon the prompt โ
CIgnore it
DPost it online
๐ก ExplanationReporting helps providers improve safeguards.
Question 77
Why is diverse, representative training data important?
AIt makes models slower
BIt reduces bias and improves fairness for different groups โ
CIt is not important
DIt increases cost only
๐ก ExplanationUnrepresentative data leads to unfair outputs.
Question 78
Who is accountable for decisions made with AI assistance?
AThe AI
BThe humans and organisations using the AI โ
CNobody
DThe internet
๐ก ExplanationAccountability always rests with people.
Question 79
What is a ‘knowledge cutoff’?
AWhen the AI stops working
BThe date after which a model has no training data, so it may not know recent events โ
CA password expiry
DA prompt limit
๐ก ExplanationVerify anything time-sensitive.
Question 80
Which is an example of responsible transparency?
ALabelling AI-generated images in a report โ
BPretending AI content is human-written
CDeleting AI logs
DHiding prompts
๐ก ExplanationDisclosure builds trust.
Question 81
What is ‘security’ risk with AI prompts?
ANone
BPrompt injection or leaking data through prompts and outputs โ
COnly password theft
DAI viruses
๐ก ExplanationTreat prompts and outputs as data that can be exposed.
Question 82
What is the recommended approach to AI and intellectual property?
AUse any content freely
BCheck licensing, avoid reproducing copyrighted material and follow organisational policy โ
CIgnore copyright
DAI content has no owner
๐ก ExplanationIP questions around gen AI are still evolving; caution is advised.
Question 83
What does ‘fairness’ in AI mean?
AAll outputs are identical
BAI treats individuals and groups equitably without unjust discrimination โ
CAI is free
DAI is fast
๐ก ExplanationFairness requires testing outputs across groups.
Question 84
How should you handle AI output that includes a statistic?
AUse it as is
BFind the original source and confirm the number and context โ
CRound it
DDelete the whole output
๐ก ExplanationStatistics are a common hallucination.
Question 85
What is ‘inclusive design’ for AI?
ADesigning for one user
BDesigning AI so it works well for people of different abilities, languages and backgrounds โ
CAdding colours
DLarger fonts only
๐ก ExplanationInclusion is part of responsible development and use.
Question 86
Which is a safe practice when using a public AI chatbot?
APasting customer records
BUsing anonymised or non-confidential information only โ
CSharing passwords
DUploading contracts
๐ก ExplanationAssume anything you type could be stored.
Next: Stay Ahead of the AI Curve โ
ย
๐
Module 5 โ Stay Ahead of the AI Curve (Questions 87โ98)
AI literacy, keeping up with developments and building an AI-enhanced workflow
Question 87
What does ‘stay ahead of the AI curve’ mean in the course?
ABuy new hardware
BKeep learning about AI developments and adapting your skills as tools change โ
CIgnore AI news
DUse only one tool
๐ก ExplanationModule 5 is about lifelong learning and adaptability.
Question 88
Which is a reliable way to keep up with AI developments?
ARandom social media posts
BOfficial product blogs, reputable tech news and courses โ
CRumours
DIgnoring updates
๐ก ExplanationThe course recommends trustworthy sources.
Question 89
What is ‘AI literacy’?
AReading AI’s code
BUnderstanding what AI can and cannot do, and using it effectively and responsibly โ
CWriting AI
DMemorising AI terms
๐ก ExplanationThe course aims to build AI literacy for any role.
Question 90
Which skill becomes more valuable as AI handles routine tasks?
ATyping speed
BCritical thinking, creativity and judgement โ
CMemorising facts
DManual data entry
๐ก ExplanationHuman skills complement AI.
Question 91
What is a good way to introduce AI in a team?
AMandate it overnight
BStart with low-risk tasks, share learnings and set guidelines โ
CBan discussion
DUse it secretly
๐ก ExplanationGradual, transparent adoption works best.
Question 92
Which is an ’emerging’ AI capability mentioned in the course?
AAI agents that complete multi-step tasks โ
BPrinting
CEmail
DSpreadsheets
๐ก ExplanationAgents and multimodal tools are growing areas.
Question 93
What should you do when a new AI tool is released at work?
AUse it immediately with client data
BCheck company policy, test with non-sensitive tasks, then evaluate โ
CIgnore it
DDelete other tools
๐ก ExplanationPolicy first, then safe experimentation.
Question 94
How can you measure whether AI is helping your work?
AGuess
BTrack time saved, quality of output and errors before and after adopting it โ
CCount prompts
DAsk the AI
๐ก ExplanationEvaluate benefits honestly.
Question 95
What is the role of ‘experimentation’ in AI adoption?
AIt is risky and should be avoided
BTrying tools on real but low-stakes tasks to learn what works โ
COnly for engineers
DUnnecessary
๐ก ExplanationThe course encourages hands-on practice.
Question 96
Which mindset does the course recommend toward AI?
AFear
BCuriosity with critical evaluation โ
CBlind trust
DIndifference
๐ก ExplanationExplore, verify, adapt.
Question 97
What is a practical next step after finishing Google AI Essentials?
AStop learning
BApply AI to a real task at work and keep refining your prompting skills โ
CDelete all AI tools
DTake no further courses
๐ก ExplanationThe course ends with building an AI-enhanced workflow.
Question 98
What certificate do you receive after completing Google AI Essentials?
AA university degree
BA Google career certificate (shareable on LinkedIn) issued via Coursera โ
CA cash prize
DA job at Google
๐ก ExplanationCompletion grants the Google AI Essentials certificate.
Next: Mixed Questions โ
ย
๐
Final Exam Practice โ Mixed Questions (Questions 99โ118)
Exam-style questions across all five modules
Question 99
A colleague says ‘AI told me, so it must be true.’ The best response is:
AAgree
BExplain that AI can hallucinate and the claim should be verified with a reliable source โ
CIgnore them
DAsk AI again
๐ก ExplanationVerification is the course’s recurring message.
Question 100
Which prompt best follows the TCREI framework?
A‘Write a report’
B‘Act as an HR specialist (task). Our startup has 20 remote staff (context). Here is our old policy as a reference. Draft a 1-page remote-work policy in plain English.’ โ
C‘Policy please’
D‘HR’
๐ก ExplanationTask, context, references and format are all present.
Question 101
You need a chart explained. Which Gemini capability do you use?
AMultimodal input: upload the chart image and ask for an explanation โ
BText only
CAudio only
DNone
๐ก ExplanationGemini handles image plus text prompts.
Question 102
Which situation is an inappropriate use of a public AI tool?
ABrainstorming blog topics
BPasting a customer’s personal data to draft a reply โ
CSummarising a public article
DDrafting a meeting agenda
๐ก ExplanationPersonal data must not go into unapproved tools.
Question 103
Supervised learning example:
AGrouping songs by similarity without labels
BTraining on emails labelled spam/not spam โ
CA game AI learning by reward
DRandom guessing
๐ก ExplanationLabels make it supervised.
Question 104
The first draft from AI is weak. The correct next step is:
AAccept it
BIterate: add context, a reference example or specify format โ
CGive up
DSend it anyway
๐ก ExplanationIteration is expected.
Question 105
What does ‘AI augments human work’ mean?
AAI replaces humans
BAI speeds up or improves tasks while people keep judgement and responsibility โ
CHumans serve AI
DAI works alone
๐ก ExplanationAugmentation is the course’s framing.
Question 106
What is the main risk of using AI output in a legal document without review?
AIt is too short
BErrors, outdated or non-compliant content that could cause harm โ
CIt is in English
DNone
๐ก ExplanationExpert review is essential.
Question 107
Which is TRUE about an LLM’s knowledge?
AIt is always current
BIt has a knowledge cutoff and may not know recent events โ
CIt knows your private files
DIt is perfect
๐ก ExplanationCheck anything time-sensitive.
Question 108
Few-shot prompting is useful when:
AYou want random output
BYou need the output in a consistent style or format shown by examples โ
CYou have no examples
DYou want a short answer
๐ก ExplanationExamples anchor style.
Question 109
Which is an example of bias in AI?
AA model recommending products
BA hiring model rating candidates lower because of gender patterns in past data โ
CA spell checker
DA calculator
๐ก ExplanationBias from historical data.
Question 110
What is the safest way to use AI to summarise a confidential internal report?
APublic chatbot
BAn organisation-approved tool whose data policy you have checked, then verify the summary โ
CEmail it to AI
DPost it online
๐ก ExplanationApproved tools plus verification.
Question 111
What does ‘human-in-the-loop’ protect against?
AFaster work
BUnchecked errors, bias and inappropriate decisions โ
CShorter prompts
DNothing
๐ก ExplanationHuman review catches what AI misses.
Question 112
Gen AI is probabilistic. This means:
AIt always gives the same answer
BThe same prompt can produce different outputs โ
CIt is deterministic
DIt cannot write
๐ก ExplanationExpect variation and iterate.
Question 113
What is the best description of Gemini?
AA search engine
BGoogle’s AI assistant built on Gemini models, usable in the Gemini app and Workspace โ
CA spreadsheet
DA browser
๐ก ExplanationGemini is the main tool used in the course.
Question 114
Which is a good evaluation question for AI output?
AIs it long?
BIs it accurate, relevant, unbiased and does it complete the task? โ
CIs it colourful?
DIs it fast?
๐ก ExplanationThe Evaluate step in TCREI.
Question 115
An AI image tool produces a picture of a real public figure doing something they did not do. This is:
AFine to publish
BA deepfake and a responsible-use violation โ
CArt
DAllowed with credit
๐ก ExplanationDo not create or share such content.
Question 116
A chain-of-thought prompt is best for:
AWriting a haiku
BSolving a multi-step maths or logic problem โ
CTranslating a word
DNaming a pet
๐ก ExplanationReasoning tasks benefit from step-by-step prompting.
Question 117
Which is NOT part of responsible AI use?
ATransparency
BVerification
CConcealing AI involvement โ
DAccountability
๐ก ExplanationConcealment undermines trust.
Question 118
Why should you keep learning about AI after the course?
ATools and capabilities change quickly; skills must be refreshed โ
BThe certificate expires daily
CIt is required by law
DNo reason
๐ก ExplanationStaying current is Module 5’s theme.
โ๏ธ The TCREI prompting framework โ what the quizzes ask
Task
What you want, plus persona and output format
Context
Audience, purpose, constraints, background
References
Examples or samples that show the style you want
Evaluate
Check accuracy, relevance, bias and fit to the task
Iterate
Refine the prompt and repeat until the output works
Extras
Zero/few-shot, chain-of-thought, prompt chaining, multimodal
๐ฅ Download all 118+ Google AI Essentials MCQs as PDF (free)
๐ฏ More AI certification MCQs
Question banks updated regularly to match the latest course content
Last updated: 15 November 2026 โ expanded from 20 to 118 questions, reorganised by the five course modules with a final-exam practice set.
Have a question from the actual quiz that is not here, or a wrong answer? Leave a comment with the question number and it goes into the next update.
[…] Prepare with free MCQs: 600 Computer MCQs (PDF) ยท IBM AI Fundamentals MCQs ยท Google AI Essentials MCQs […]
[…] ๐ Google AI Essentials MCQs ๐ Microsoft AI-900 MCQs โ๏ธ AWS AI Practitioner MCQs […]