Gains:
- Ability to explain the limits of the use of artificial intelligence in MIS within the framework of KVKK and basic information security principles
- Ability to safely apply anonymization, least privilege and data classification principles to artificial intelligence workflows
- Ability to responsibly manage an end-to-end MIS-AI project within the framework of ethics, compliance and competent approval.
Module Exam
1. Which statement most accurately describes the role of artificial intelligence output in the work of a Management Information Systems specialist?
- A) AI produces drafts and proposals; ✔ Business rule is human's responsibility to verify with real data and competent approval
- B) AI output is written fluently and technically so it usually does not require verification
- C) AI can finalize requirements directly without business unit approval
- D) In finance-critical decisions, AI output replaces the approval of a competent expert
Description: Artificial intelligence quickly drafts intermediate products such as requirements, data models, SQL and reports; However, it is the human responsibility to verify these drafts with the business rule, actual data, and stakeholder confirmation. Particularly in financial and compliance-critical decisions, AI is not a substitute for competent expert approval.
2. What is the main problem with the statement 'the system must be fast' in a requirements analysis study?
- A) It is written as non-functional even though it is a functional requirement
- B) Cannot be tested and verified because it does not contain a measurable threshold ✔
- C) Stakeholders do not understand because it contains too many technical details
- D) It is invalid because it is not written in user story format
Explanation: This statement describes a non-functional requirement, but it is not measurable. 'Fast' is subjective; defines an acceptable requirement response time by number, condition and load (e.g. 'query response less than 2 seconds at average load'). A requirement that cannot be measured cannot be tested.
3. Which of the following is the main purpose of normalization in data modeling?
- A) Simplifying the query by collecting all the data in one large table
- B) Prevent update anomalies by reducing data duplication and maintaining integrity ✔
- C) Always minimize the size of backup files
- D) Directly guaranteeing graphics and board performance
Description: Normalization is the process of logically splitting tables to reduce redundancy and preserve the integrity of data. Keeping the same information in one place prevents update anomalies and inconsistency. The goal is not to speed up the report; deliberate denormalization is sometimes done for speed.
4. What is the key risk of inferring 'sales increased as ad spend increased, so advertising increases sales' on a business intelligence dashboard?
- A) Confusing correlation with causation; Acting together does not prove that one creates the other ✔
- B) Incorrect selection of chart type
- C) The data is not normalized
- D) The board contains too many colors
Explanation: This is confusing correlation with causation. The two metrics may move together, but a third factor (e.g. season, campaign period) may be driving both up. The MIS expert should also check for this trap in AI-generated narrative texts and test causal claims with evidence.
5. In smart automation, where RPA (robotic process automation) and artificial intelligence are used together, which step type is RPA most suitable for?
- A) Steps that are unclear, based on interpretation, and require different decisions each time
- B) Steps working on regular, repetitive and structured data ✔
- C) Steps that require empathy with the customer
- D) Steps in which strategic investment decisions are made
Explanation: RPA is strong at regular, repetitive and structured data steps: copying data from one screen and entering it into another system. Artificial intelligence comes into play in steps that require ambiguous judgment, classification, or natural language understanding; High-risk decisions remain subject to human approval.
6. Why do AI outputs need to be validated particularly carefully in ERP systems?
- A) ERP data is always anonymous so additional caution is required
- B) ERP processes are financially and operationally critical; Mistake has direct financial and legal consequences ✔
- C) ERP systems cannot be integrated with artificial intelligence at all
- D) AI does not understand ERP reports because they are not visual
Description: ERP; Conducts financial and operational critical processes such as accounting, inventory, purchasing and payroll. An error here is directly reflected in the financial recording, payment or inventory decision. Therefore, AI recommendations are not processed without the approval of a competent expert; Audit trail is mandatory.
7. What is the first thing an MIS specialist should pay attention to when using artificial intelligence when working with CRM data?
- A) Choosing the color palette of the panel
- B) Personal data privacy: anonymization, minimum authorization and ensuring KVKK compliance first ✔
- C) Loading the largest possible data set into the vehicle as is
- D) It is sufficient to use the newest version of the model
Description: CRM data is largely personal data (name, contact, purchase history). Before giving this data to an artificial intelligence tool, anonymization, minimum authorization and KVKK compliance should be evaluated. Additionally, automated decisions that affect the customer require human oversight and the right to object.
8. What type of chart is generally best suited for showing rate (percentage) change over time on a dashboard?
- A) Line chart for time trend; scale and axis must be set correctly ✔
- B) A pie chart is best in every situation
- C) Graphics with three-dimensional effects increase readability
- D) Chart type does not affect the message, they are all equivalent
Explanation: A line chart is suitable for showing a trend over time; If the axes and scale are set correctly, the change can be read clearly. The pie chart shows the share at a single point in time, it is not suitable for time trending. Graphic selection is made according to the message to be conveyed.
9. Which of the following is the most important risk of no-code / low-code platforms in terms of governance?
- A) Applications always run very slowly
- B) Undocumented, unaudited 'shadow IT' solutions create data and security risks ✔
- C) They never solve real business problems
- D) They can only be used in large companies
Description: No-code/low-code allows non-technical employees (citizen developers) to quickly install applications. However, when it grows uncontrolled; Undocumented, unsecured and orphaned 'shadow IT' solutions create data leakage and technical debt. Governance requires recording and monitoring these solutions.
10. What does 'consistency', one of the data quality dimensions, mean?
- A) The data fits the real world value exactly
- B) The same data does not conflict with each other in different systems and records ✔
- C) All fields in the data are filled
- D) Finding the data in its most current form
Explanation: Consistency means that the same data does not conflict with each other in different systems or records; For example, a customer's address is the same in the CRM and billing system. Accuracy means conformity to reality, completeness means completeness, and up-to-dateness means timeliness of data. These are separate dimensions.
11. Within the scope of KVKK, what is the most basic principle that an MIS expert should apply before processing personal data with an artificial intelligence tool?
- A) Collecting all personal data that can be obtained
- B) Data minimization and purpose limitation: process only necessary data for the defined purpose ✔
- C) Putting the data in a public folder that everyone can access
- D) Leaving the privacy decision entirely to artificial intelligence
Description: Data minimization and purpose limitation are essential: processing only necessary data for an identified and legitimate purpose. Where possible, data is anonymized or pseudonymized; Areas that identify the person are kept away from the vehicle. The 'more data, the better' approach is against KVKK principles.
12. Why is the sentence 'the system automatically approves when appropriate' a risky requirement in an AI-powered requirement draft?
- A) Because it is too long
- B) 'Feasible cases' are undefined, open to interpretation, untestable and error prone ✔
- C) Automation is always prohibited
- D) Because it is not in user story format
Explanation: 'Suitable situations' is a vague expression; It is not defined what conditions trigger approval. The ambiguous requirement is misinterpreted by the developer and cannot be tested. The MIS expert should clarify such ambiguous statements produced by AI with clear rules and thresholds and have them confirmed by the stakeholder.
13. What is the main purpose of an ETL (Extract-Transform-Load) process in a data warehouse?
- A) Generating new customer data
- B) Extracting and transforming data from different sources and loading it into a consistent structure ready for analysis ✔
- C) Automatically selecting the colors of the panels
- D) Sharing personal data regardless of the law
Description: ETL; It extracts data from sources (Extract), cleans it and converts it into a standard structure (Transform) and loads it into the data warehouse for analysis (Load). The aim is to bring dispersed and different formats of operational data into a consistent structure ready for reporting and analysis. ETL does not produce data, it moves and organizes existing data.
14. Why is 'exception handling' critical in process automation design?
- A) Because it always makes automation work faster
- B) Because it provides safe stopping and guidance to people instead of silent errors in illegal situations ✔
- C) Because it improves graphic design
- D) Because it automatically deletes personal data
Explanation: Automations always encounter situations that go against the rules (missing data, unexpected format, system error). Exception handling allows the robot to safely stop the process and redirect it to a human in these cases, rather than silently taking the wrong action. If no exceptions are designed, errors will propagate silently and the audit trail will be lost.