CertArea
All EC-Council exams

Free Certified AI Program Manager (C|AIPM) practice test: 30 questions with answers

30 original questions for Certified AI Program Manager (C|AIPM), drawn from every topic area. Answer each one to see whether you got it right and why. No sign-up.

Free questions
30
Full bank
112 questions

Certified AI Program Manager (C|AIPM) exam guide: topics, cost and format

All 30 Certified AI Program Manager (C|AIPM) questions with answers

The same questions as the test above, grouped by exam topic. Open any question to see the answer and explanation.

Pilot Execution, Measurement and Sustainment (6 questions)

  1. Question 1

    Which pilot design best tests whether an AI use case creates value?

    1. AA pilot with no comparison group, where the team reports that the results felt better than before the pilot began
    2. BA controlled pilot with a defined comparison group, baseline metrics and clear success criteria agreed before it starts
    3. CA pilot launched to all customers at once, so that the largest possible sample is available for the analysis afterwards
    4. DA pilot run only in the team's demo environment, with synthetic data chosen to show the best results that are possible
    Show answer and explanation
    Correct answer: B

    A controlled pilot with a comparison group and agreed baselines and criteria separates the effect of the AI from other changes. Uncontrolled pilots, mass launches and demo-only tests cannot show real value.

  2. Question 2

    Which metric would best show the financial impact of an AI-driven process change?

    1. AThe number of AI tokens consumed last month, which shows how busy the model was but not what it achieved for the business
    2. BA before-and-after comparison of cost per transaction or cycle time, adjusted for volume changes and the cost of running the AI
    3. CThe number of meetings held about the AI programme, which indicates the team's commitment to the initiative over the year
    4. DThe total number of lines of code written by the AI tool, which measures the tool's productivity directly for the business
    Show answer and explanation
    Correct answer: B

    Comparing cost per transaction or cycle time before and after, adjusted for volume and running costs, shows financial impact. Token use, meeting counts and code lines measure activity rather than outcome.

  3. Question 3

    After a pilot succeeds, what should the programme manager do to scale it?

    1. APlan the scaling with operating model changes, support, monitoring and funding, and decide the next phase on the evidence
    2. BRoll the pilot out to every business unit the next day without changing the process, the training or the controls in place
    3. CStop measuring the results, so that the scaled version is judged only by its novelty in the business during its first year
    4. DHand the pilot to a vendor with no internal ownership or governance in place for the scaled service after the handover
    Show answer and explanation
    Correct answer: A

    Scaling needs operating model changes, support, monitoring and funding, with decisions taken on evidence. Immediate rollout, stopping measurement and vendor handover without ownership each remove the controls that scale depends on.

  4. Question 4

    Which practice helps sustain an AI capability over several years?

    1. AAssigning a business owner, funding ongoing operations, reviewing the model regularly, and planning for retraining and retirement
    2. BBuilding the model once and leaving it unchanged for ten years, so that the investment is never revisited by the team again
    3. CRelying on the original project team to remain in place indefinitely, with no handover to operations at any point in time
    4. DRemoving monitoring after the first year, since the model will not change in production once it has been deployed for use
    Show answer and explanation
    Correct answer: A

    Ownership, funding, regular review and a lifecycle plan keep an AI capability working over time. Freezing the model, relying on the original team and removing monitoring all allow quality to decline unnoticed.

  5. Question 5

    Which practice gives early warning that an AI system's quality is slipping in production?

    1. ARelying on a single user complaint each quarter as the only signal that something might be wrong with the model in service
    2. BChecking the model only at the annual audit, which is the single moment that the team reviews the quality of the model each year
    3. COngoing monitoring of accuracy, error rates, user feedback and drift indicators, with thresholds that trigger a review
    4. DMeasuring only the speed of the model's responses, since fast responses are always accurate responses to the users of the system
    Show answer and explanation
    Correct answer: C

    Continuous monitoring of accuracy, errors, feedback and drift, with thresholds, gives early warning. Annual checks, occasional complaints and speed measurement alone miss degradation.

  6. Question 6

    What is an appropriate step when a pilot fails to meet its success criteria?

    1. AHide the failed result from the leadership team, so that the programme is not seen to have lost momentum with the sponsors
    2. BReview the evidence, decide whether to adjust, pivot or stop, and record the lessons learned for the programme
    3. CRerun the pilot with the same design and the same criteria until it succeeds, regardless of what the evidence has shown so far
    4. DContinue spending on the pilot without any new decision being taken, since stopping would be an admission of failure by the team
    Show answer and explanation
    Correct answer: B

    A failed pilot should lead to an evidence-based decision to adjust, pivot or stop, with the lessons recorded. Hiding results, rerunning until success and drifting on without a decision each misuse the evidence.

AI Platforms, Governance and Ethics (6 questions)

  1. Question 7

    Which framework provides a structured approach to managing AI risk across the AI lifecycle?

    1. AThe Six Sigma method, which reduces defects in manufacturing processes through statistical process control techniques on the line
    2. BThe ITIL framework, which organises IT service management practices into life-cycle stages for the delivery of IT services
    3. CThe PCI DSS standard, which sets requirements for the protection of payment card data in merchant and service environments
    4. DThe NIST AI Risk Management Framework, which organises AI risk management into functions such as govern, map, measure and manage
    Show answer and explanation
    Correct answer: D

    The NIST AI RMF organises AI risk management into govern, map, measure and manage functions. ITIL, PCI DSS and Six Sigma address service management, payment security and manufacturing quality.

  2. Question 8

    Which governance control matters most for a high-impact AI use case, such as credit decisions?

    1. AHuman oversight and documented review of the model's decisions, with the ability to explain and contest the outcomes
    2. BFully automated decisions with no review, since the model has been trained on a large dataset and is therefore assumed to be fair
    3. CA single developer approving the model alone, with no record of the approval or of the testing that was performed on it
    4. DPublishing the model's training data on a public website so that anyone can check the model for bias on their own
    Show answer and explanation
    Correct answer: A

    High-impact decisions need human oversight, documentation and a way for people to explain and contest outcomes. Unreviewed automation, single-person approval and public data dumps each remove the controls such decisions need.

  3. Question 9

    What does model drift describe?

    1. AThe gradual increase in the model's size as more features are added to it by the team over successive versions
    2. BThe movement of the model's files between two storage locations on the company's network during a planned migration
    3. CA decline in model performance over time, as the data or the operating environment changes away from the training data
    4. DThe change in the colour of the model dashboard as the team updates the theme for the executive view each quarter
    Show answer and explanation
    Correct answer: C

    Model drift is the decline in performance as the data or environment shifts from what the model learned. File moves, model size and dashboard colours are not drift.

  4. Question 10

    Which practice reduces the risk of a generative AI assistant giving confident but wrong answers to customers?

    1. ATurning off the logs of the assistant's answers, so that the team does not have to review them for errors each week
    2. BAllowing the assistant to answer any question from the internet without restriction, so that it appears knowledgeable to users
    3. CRemoving all human review, since customers prefer instant answers even when those answers may turn out to be inaccurate
    4. DGrounding responses in approved company content through retrieval, with citations and an escalation path for uncertain cases
    Show answer and explanation
    Correct answer: D

    Grounding in approved content with citations and escalation reduces confident errors, because answers can be traced and uncertain cases handled. Unrestricted answers, removing review and disabling logs each increase the risk.

  5. Question 11

    Which data protection principle matters when an AI tool processes customers' personal data?

    1. AIndefinite retention, since the model may need the data in the future to improve its performance over the coming years
    2. BMaximum data collection, so that the model is trained on as many personal details as possible in order to be accurate
    3. CPublic sharing of the training data, which is seen as the most transparent approach for the AI programme in the market
    4. DData minimisation and purpose limitation, so that only the data needed for the stated purpose is used and kept
    Show answer and explanation
    Correct answer: D

    Data minimisation and purpose limitation are core data protection principles that apply to AI processing. Maximum collection, public sharing and indefinite retention each conflict with them.

  6. Question 12

    Which two elements should an AI use policy for employees include? (Choose two.)

    1. AWhich tools are approved for use, and what types of data may and may not be entered into them by staff
    2. BA ban on any use of AI, even for approved tasks, so as to remove all risk from the organisation entirely
    3. CA requirement to verify AI outputs before they are relied on, especially for customer-facing or legal content
    4. DPermission for staff to share confidential client data with any public AI tool they find useful during their work
    5. EA rule that AI outputs must never be documented, so that the work stays informal and quick to produce for the team
    Show answer and explanation
    Correct answer: A, C

    Approved tools with data rules, and output verification for customer and legal content, make everyday AI use safer. A total ban, sharing confidential data publicly and skipping documentation each create risk or hide accountability.

AI Fundamentals and Business Readiness (6 questions)

  1. Question 13

    An executive asks why the company should invest in AI at all. What is the most useful framing for the programme manager to give?

    1. ALink AI investments to specific business outcomes, such as cost reduction, revenue growth or risk reduction, with measurable targets
    2. BExplain that AI is fashionable and that competitors are using it, so the company must follow the trend to stay visible in the market
    3. CPresent the model's technical architecture diagrams first, so that executives understand which algorithms the team will use
    4. DRecommend buying the largest model available, since larger models always deliver better business results whatever the need
    Show answer and explanation
    Correct answer: A

    Tying AI investment to measurable business outcomes gives executives a basis for decisions. Trend-following, architecture-first presentations and size-based purchasing do not answer the question of value.

  2. Question 14

    Which measure best describes an organisation's readiness for AI adoption?

    1. AThe total number of AI conference talks attended by the leadership team in the last twelve months of the financial year
    2. BThe number of AI tools that employees have installed on their personal devices without any review by the IT function
    3. CAn assessment of data quality, skills, infrastructure, governance and leadership sponsorship against a maturity scale
    4. DThe number of AI-generated slides in the quarterly board deck, which shows how interested the leadership team is in AI
    Show answer and explanation
    Correct answer: C

    Readiness is assessed across data, skills, infrastructure, governance and sponsorship, which shows where gaps lie. Tool installs, conference attendance and slide counts measure activity rather than readiness.

  3. Question 15

    Which data condition is the most common blocker for early AI projects?

    1. AA surplus of skilled data scientists who have no projects to work on, leaving them idle for months at a time in the team
    2. BToo much clean, well-labelled data, which makes the projects take too long to scope and get approved by the governance board
    3. CFragmented, inconsistent or poorly governed data that cannot be accessed reliably for training and evaluation purposes
    4. DExcessive storage capacity, which makes it expensive to keep the datasets available for the teams' use over the long term
    Show answer and explanation
    Correct answer: C

    Fragmented and poorly governed data is the most frequent obstacle, because models and evaluations depend on reliable access. Excess data, idle staff and spare storage are not typical blockers.

  4. Question 16

    What does a maturity model for AI adoption help a programme manager do?

    1. APosition the organisation on a scale, identify the gaps, and plan the steps needed to reach the next level
    2. BGuarantee that the AI pilots will succeed without any further planning or measurement of their results over the coming year
    3. CReplace the business case with a single score that the board can compare across competitors each year in its review
    4. DRemove the need for executive sponsorship, since the maturity score shows whether leadership support is already present
    Show answer and explanation
    Correct answer: A

    A maturity model positions the organisation, shows gaps and guides the path forward. It does not guarantee success, replace the business case or remove the need for sponsorship.

  5. Question 17

    Which two actions are typical in the readiness phase before an AI pilot? (Choose two.)

    1. AConfirming data access, privacy approvals and the skills needed to run the pilot successfully within the time planned
    2. BAgreeing a baseline measurement and the success criteria before any build or deployment work begins on the pilot
    3. CDeploying the model to all customers before the pilot has produced any evidence of value for the business at all
    4. DSkipping the baseline measurement so that the pilot results can be presented as dramatic improvements to the board
    5. EHiring a large team of contractors before the scope and the success criteria of the pilot have been defined and agreed
    Show answer and explanation
    Correct answer: A, B

    Confirming data, privacy and skills, and agreeing a baseline and success criteria, prepare a pilot that can be judged. Mass deployment, skipped baselines and hasty hiring each weaken the evidence or the control of the work.

  6. Question 18

    Which step should come first in an AI readiness assessment?

    1. ADefining the business objectives and the decisions that AI is expected to support, so that the assessment has a clear purpose
    2. BBuying the GPU cluster the team has been promised by a vendor, so that the infrastructure exists when the data arrives
    3. CChoosing the programme's name and logo for the internal launch in the communications plan that is drawn up each quarter
    4. DTraining every employee on prompt writing before anyone has a defined use for AI in their own daily work
    Show answer and explanation
    Correct answer: A

    A readiness assessment needs objectives and decisions to test against, otherwise it measures the wrong things. Buying hardware, naming the programme and mass training all happen before there is a defined use to support.

Use Case Identification and Strategy (6 questions)

  1. Question 19

    Which method helps prioritise AI use cases across a portfolio?

    1. AStarting with every idea on the list at once, so that no team feels left out of the programme or its early benefits
    2. BChoosing whichever use case the most senior manager mentioned first at the last meeting of the year, to keep the sponsor happy
    3. CRanking use cases by how novel the technology sounds to the board members at each presentation during the year
    4. DA scoring model that weighs business value against feasibility, data readiness, risk and cost for each use case
    Show answer and explanation
    Correct answer: D

    A scoring model that balances value, feasibility, data, risk and cost gives a transparent basis for prioritisation. Seniority, novelty and inclusion-for-all ignore the trade-offs that decide value.

  2. Question 20

    A team proposes an AI chatbot for customer service. What should be assessed first?

    1. AThe brand colour of the chat window, which must match the website's theme before the project can be approved to proceed
    2. BThe problem it solves, the volume and value of the interactions, the risk of wrong answers, and the data it would need
    3. CThe number of emojis the bot will use in its replies, since customers prefer a friendly tone over accuracy in support
    4. DWhether competitors have a similar chatbot, which is the only factor that matters when the project is presented for approval
    Show answer and explanation
    Correct answer: B

    The problem, interaction value, error risk and data needs determine whether a chatbot is a sound use case. Colour, emojis and competitor presence are secondary and do not establish value or safety.

  3. Question 21

    Which element belongs in an AI strategy roadmap?

    1. AA single slide that says the company will be an AI leader, with no initiatives, owners or timing given at all
    2. BA list of every AI tool on the market, ranked by the vendor's marketing budget for the year ahead in the sector
    3. CSequenced initiatives with dependencies, owners, milestones and the outcomes each initiative is expected to deliver
    4. DThe colours and fonts for the internal AI newsletter that the communications team publishes each month for staff
    Show answer and explanation
    Correct answer: C

    A roadmap sequences initiatives with owners, milestones and expected outcomes, which makes it actionable. Vendor rankings, slogans and newsletter styling do not form a roadmap.

  4. Question 22

    What is the main purpose of value-tracking metrics in an AI programme?

    1. ATo count how many prompts employees send to the AI tool each day, as a proxy for the programme's popularity in the business
    2. BTo show whether each deployed use case delivers its expected business outcome, and to decide whether to scale or stop it
    3. CTo show how many graphics processors the programme has purchased, which proves that the programme is serious about its investment
    4. DTo compare the programme with a competitor's press releases each quarter, in the board report that is circulated to members
    Show answer and explanation
    Correct answer: B

    Value-tracking metrics test whether each use case delivers the expected outcome, which informs the scale or stop decision. Prompt counts, hardware purchases and press comparisons do not show value.

  5. Question 23

    Which two factors make an AI use case a poor candidate for early investment? (Choose two.)

    1. AThe problem is not well defined, and no one has agreed what success would look like in measurable terms
    2. BThe data required is not available and would take a long time to assemble, with unclear rights to use it
    3. CThe use case has a clear, measurable outcome, with an accountable business owner and the data it needs available now
    4. DThe team has already run several small pilots that produced clear, measurable results for the business over the year
    5. EThe business owner has made a clear commitment to use the outputs in daily operations once the work is complete
    Show answer and explanation
    Correct answer: A, B

    Undefined problems and unavailable data are common reasons early investments fail. Clear outcomes, proven pilots and committed owners are signs of a good candidate.

  6. Question 24

    Which technique helps a team test whether an AI idea is desirable, feasible and viable before building it?

    1. AWriting a long technical specification and sending it to the vendor for a quote, without any user involvement at any stage
    2. BBuilding the full system first and then asking users whether they like the result that the engineers produced at the end
    3. CDesign thinking with lightweight prototypes and tests, which check whether users want the idea and whether it can work
    4. DHolding a one-hour brainstorm and recording the first idea that the most senior person in the room proposes for the project
    Show answer and explanation
    Correct answer: C

    Prototypes and tests check desirability, feasibility and viability early and cheaply. Building first, quotes without users and deference to seniority all skip the evidence that the idea needs.

Change Management and Enablement (6 questions)

  1. Question 25

    Which change management model structures individual change through awareness, desire, knowledge, ability and reinforcement?

    1. AThe ADKAR model, which structures individual change through awareness, desire, knowledge, ability and reinforcement
    2. BA Gantt chart, which displays tasks and their timing on a timeline for the project team to follow each week of the work
    3. CA SWOT analysis, which lists the strengths, weaknesses, opportunities and threats of the organisation as a whole
    4. DThe balanced scorecard, which tracks financial, customer, process and learning measures for the organisation over time
    Show answer and explanation
    Correct answer: A

    ADKAR is an individual change model covering awareness, desire, knowledge, ability and reinforcement. Gantt charts, SWOT analysis and the balanced scorecard serve planning and strategy rather than individual change.

  2. Question 26

    Why does an AI programme need employee enablement, not only technical training?

    1. ABecause enablement is only a marketing exercise that makes the programme look good to the members of the board
    2. BBecause the AI tools are so simple that no training or guidance is needed by anyone in the business at all
    3. CBecause the training shows staff how to build the models themselves without any support from the data science teams
    4. DBecause people must understand when to use AI, how to check its outputs, and how their roles change as a result
    Show answer and explanation
    Correct answer: D

    Enablement covers when to use AI, how to verify its outputs and how roles change, which technical training alone does not address. The other options either deny the need for guidance or misstate what enablement is.

  3. Question 27

    A department resists an AI tool that changes its workflow. What is the most effective first step?

    1. AIgnore the resistance and proceed with rollout, since resistance is always a sign that the staff are unwilling to change at all
    2. BMandate the tool immediately and remove the old process on the same day, so that staff have no choice but to adapt
    3. CEngage the affected people early, explain the benefits and concerns, and involve them in shaping the new workflow
    4. DReplace the affected team with contractors who have no experience of the old workflow, and so have no resistance to the change
    Show answer and explanation
    Correct answer: C

    Early engagement, honest discussion of benefits and concerns, and involvement in design address resistance at its source. Forced mandates, ignoring resistance and replacing staff each damage trust and adoption.

  4. Question 28

    Which measure best indicates successful adoption of an AI tool by staff?

    1. AThe number of times the tool's login page was opened by staff during the week, regardless of what they did afterwards
    2. BRegular, active use of the tool in the intended workflows, combined with user feedback showing that they find it useful
    3. CThe number of licences purchased for the tool, whether or not any of them is actually used by the people who hold them
    4. DThe number of training certificates issued for the tool, which shows that everyone has completed the course on it
    Show answer and explanation
    Correct answer: B

    Active use in intended workflows, with positive feedback, shows that adoption is real. Page views, licence counts and certificates measure access or activity rather than adoption in practice.

  5. Question 29

    Which practice makes employees more confident using AI in their daily work?

    1. AA single long video of several hours that nobody is required to watch, with no examples relevant to any particular role
    2. BPractical, role-specific training with examples, supported by a place to ask questions and share good practice
    3. CA rule that AI must never be used, which makes staff afraid to try the tools even for approved and safe tasks at work
    4. DPublic criticism of any employee who makes a mistake while using AI for their work, in the presence of their colleagues
    Show answer and explanation
    Correct answer: B

    Role-specific training with examples and a place to ask questions builds confidence through practice. Unwatched videos, bans and public criticism all reduce confidence and use.

  6. Question 30

    Which change risk is most likely when an AI tool automates a repetitive task?

    1. AThe staff will stop using their computers altogether, which makes the change difficult to measure in the business
    2. BThe tool will never make a mistake, so the checks on its output can be dropped from the process entirely
    3. CThe tool will speed up the work so much that the team will have to be given a long holiday to rest from it
    4. DStaff anxiety about their roles, which can lead to disengagement or to workarounds if it is not addressed openly
    Show answer and explanation
    Correct answer: D

    Anxiety about roles is the most common change risk when automation arrives, and it shows up as disengagement or workarounds. The other options either assume error-free tools or describe unrealistic outcomes.