Gains:
- Ability to design an operation that combines content, search, recommendation, service and analytics steps in a single auditable workflow
- Ability to transform KVKK/GDPR, consumer law and marketplace agreements into a compliance framework embedded in daily artificial intelligence use
- Ability to institutionalize ethics, transparency and responsibility principles as artificial intelligence governance and create a rule set for the team
Module Exam
1. A store owner directly applies the new selling price suggested by artificial intelligence to all products without any cost and margin control. What is the fundamental mistake in this approach?
- A) Transforming the artificial intelligence output into a decision without subjecting it to cost, margin and commercial verification; Ignoring that the responsibility lies with people ✔
- B) It is strictly forbidden to use artificial intelligence in pricing
- C) Artificial intelligence always shows the price lower than it is
- D) The price offer is not presented in a table
Description: Artificial intelligence produces a statistical recommendation for the price; However, the final price depends on cost, margin, campaign and competition rules and the responsibility belongs to the store owner. Unverified output is like an unchecked product and can cause harm.
2. A customer service officer pastes the order list, including customers' names, phone numbers and address information, into a publicly available artificial intelligence tool and says 'write campaign text to these people'. What is the main problem with this behavior?
- A) The campaign text is too long
- B) Artificial intelligence cannot write campaign text
- C) The campaign can only be sent by e-mail
- D) Violating KVKK/GDPR by transferring personal data to a public tool without anonymizing it ✔
Explanation: Name, telephone, address and order information are personal data within the scope of KVKK/GDPR. Uploading it as is to a public tool is a data breach. The right way is to anonymize personal data or not share it at all, only to have the text itself produced.
3. Which of the following is the most correct approach to get the most reliable result from artificial intelligence when writing a product description?
- A) Give the product name and let artificial intelligence guess the rest
- B) Asking the user to use as many exaggerated adjectives as possible
- C) Giving verified specifications, target audience and tone rules as input and checking the output on a claim basis ✔
- D) Copying the competitor's product description exactly as it is
Explanation: Artificial intelligence does not know the real features of the product; The printed text may contain fabricated claims without providing it with verified technical specifications (specs) and target audience. The correct approach is to give actual features as input and validate the output.
4. Why is 'keyword stuffing' harmful in e-commerce SEO?
- A) Because it reduces readability and conversion and can be penalized by the search/marketplace algorithm ✔
- B) Because it lengthens the text and causes the page to open slower
- C) Because keywords are always paid
- D) Because it only causes problems in English stores
Explanation: Keyword stuffing is filling the text with the same term unnecessarily. It not only reduces readability and conversion, but can also reduce visibility by being penalized by search and marketplace algorithms. Sustainable SEO requires intent-driven and natural text.
5. A fraud detection model flagged an order as 'high risk'. What is the correct operational approach?
- A) Cancel the order immediately and close the account permanently
- B) Ignore the score and automatically confirm all orders
- C) Take the score as a signal and make the final decision with human review, additional verification and justification ✔
- D) Refusing the refund without giving any explanation to the customer
Explanation: Risk score is a statistical signal, not conclusive evidence. There is a risk of false positives (unfairly blocking honest customers). High-impact decisions (order cancellation, account closure) should be subject to human review, justification, and additional verification if necessary.
6. Why is 'automatic recommendation without considering stock and margin' risky in recommendation systems?
- A) Because recommendation systems always work slowly
- B) Because it can disrupt the experience and profitability by highlighting out-of-stock or unprofitable products ✔
- C) Because recommendation systems only work in large stores
- D) Because customers stop shopping when they see recommendations
Explanation: If the recommendation engine only looks at behavioral data, it may highlight out-of-stock or low-margin/harmful products. This disrupts both customer experience and profitability. That's why automatic recommendations should be tied to inventory, margin and business rule control.
7. Which of the following sets of factors most affects winning the 'Buy Box' in marketplaces?
- A) Length of product description only
- B) How many years the store has been open alone
- C) File size of the product image
- D) Combination of competitive and operational factors such as price, delivery performance, seller rating and inventory ✔
Description: Buy Box is the selection of a seller by clicking on the 'Add to Cart' button among many sellers selling the same product. Competitive and operational factors such as price, delivery time/performance, seller rating and stock availability are decisive. Just the lowest price alone is not enough.
8. Why is it problematic to show an accessory that is not actually in the product in a product image produced with artificial intelligence?
- A) Because the resolution of the image decreases
- B) Because artificial intelligence images are always royalty-free
- C) Because it may lead to consumer law violations and platform sanctions by representing the product in a misleading way ✔
- D) Because images with accessories are clicked less
Description: The image must fairly represent the product. Showing an item that is not included in the product is a misleading representation; In terms of consumer law, it is considered a misleading commercial practice and results in returns/complaints and platform sanctions. The image must be exactly compatible with the real product.
9. What is the best attitude towards artificial intelligence saying 'two metrics increased together, one is the cause of the other' in e-commerce analytics?
- A) Accept correlation as direct causality and take action
- B) Treat the two metrics as independent from each other and abandon the analysis completely
- C) Knowing that correlation is not causation and testing the claim with a controlled test or additional data ✔
- D) Always accepting every causality claim of artificial intelligence as true
Explanation: Two metrics that move together are correlated; This does not imply causality. A common external factor (e.g. campaign period) may have increased both. The correct attitude is to test causality with controlled testing (A/B) or additional data.
10. What is the best way to ensure that the customer service chatbot provides reliable answers about 'return policy'?
- A) Basing the answers on an up-to-date and approved knowledge base and transferring them to humans outside the scope ✔
- B) Allowing the chatbot to freely generate responses from its general knowledge
- C) Fixing the 'non-refundable' answer to every question
- D) Chatbot automatically approves all return requests
Description: Chatbot can 'make up' policy from its own memory (hallucination). The right solution is to base the answers on the store's current and approved knowledge base, and to transfer it to humans in out-of-scope and risky situations (escalation).
11. Why should 'setting a common price by agreeing with competitors' not be done while monitoring competitor prices?
- A) Because the common price can be noticed by customers
- B) Because price fixing/cartel is prohibited in competition law and leads to heavy penalties ✔
- C) Because observing competitor prices is prohibited under all circumstances
- D) Because artificial intelligence cannot calculate the competitor's price
Explanation: Agreeing on prices with competitors (price fixing/cartel) is prohibited in competition law and leads to heavy administrative fines. It is legitimate to observe competitor pricing; but coordination with competitors is illegal. The price decision must be made independently and unilaterally.
12. Which of the following is true in terms of complying with marketplace rules when producing a product title?
- A) The same title can be used in all marketplaces without any changes
- B) Checking and adapting the title according to the platform's character limit, format and prohibited expression rules ✔
- C) Adding as many exclamation points and capital letters as possible to the title
- D) It is always free to use the brand name in the title without permission
Description: Each marketplace has its own character limit, prohibited phrases, and formatting rules for the title (e.g. shouting in all caps, promotional phrases, brand clearance). AI output must be audited according to these rules and adapted to the platform; One text may not fit all.
13. If the AI gave you a confident number saying 'last quarter our conversion rate was 4.2 percent', what is the right approach?
- A) Putting the number directly into the report and presentation
- B) Ask the artificial intelligence a few more times and take the average
- C) Accepting the number as correct because it is presented fluently
- D) Know that the number may be fake and verify it from your own analytical source ✔
Explanation: Unless the artificial intelligence has access to your data, it has made up this number (hallucination). Metrics such as conversions are validated only from your own analytics dashboard (Google Analytics, marketplace dashboard, ecommerce engine). It is wrong to use the number without confirming it with its source.
14. What is the most critical measure for the 'fair decision' principle when establishing an artificial intelligence-supported system in the extradition process?
- A) Keeping decisions justifiable and objectionable and linking high-impact situations to human approval ✔
- B) Automatically reject all returns without exception
- C) Not recording return decisions at all
- D) Never explaining the reason for the decision to the customer
Explanation: Automatic refund/rejection decisions may systematically disadvantage certain customer groups (bias). For fair decision making, decisions must be justifiable, contestable, and high-impact cases must be subject to human review. Transparency and auditability are essential.
15. Which principle is essential when using artificial intelligence in terms of brand safety in product images and banner designs?
- A) Publishing every image produced without checking it
- B) Assuming that it is free to add logos of well-known brands to the image
- C) Thinking that it is unimportant that the visual is compatible with the real product
- D) Check images for copyright/trademark infringement and misleading representation and keep them consistent with the brand identity ✔
Description: Artificial intelligence may produce images, logos of other brands, copyrighted characters or unauthorized work elements. For brand safety, output must be inspected for copyright and trademark infringement, and the image produced must honestly represent the actual product and be consistent with your own brand identity.