Assignment two

02_activities/assignments/DC_Cohort/Assignment2.md

@@ -56,7 +56,18 @@ The store wants to keep customer addresses. Propose two architectures for the CU
 **HINT:** search type 1 vs type 2 slowly changing dimensions. 
 
 ```
-Your answer...
+There are two possible architectures for the `CUSTOMER_ADDRESS` table, depending on whether the bookstore wants to keep a history of address changes.
+
+Type 1 — Overwrite changes
+
+In a Type 1 design, the customer's existing address is updated when they provide a new address. The table could contain columns such as `customer_address_id`, `customer_id`, `street`, `city`, `province`, `postal_code`, and `country`. If a customer moves, their old address is replaced with the new address. This approach is simple and only keeps the customer's current address, but it does not preserve any history of previous addresses.
+
+Type 2 — Retain changes
+
+In a Type 2 design, a new record is created whenever a customer's address changes, while the previous address remains in the table. The table could contain `customer_address_id`, `customer_id`, `street`, `city`, `province`, `postal_code`, `country`, `start_date`, `end_date`, and `is_current`. When the customer moves, the previous record can be given an end date and marked as no longer current, while a new record is created with the new address and a new start date. This allows the bookstore to maintain a complete history of the customer's addresses.
+
+Therefore, Type 1 overwrites the existing address and does not retain historical changes, while Type 2 creates new records and retains the history of address changes. Type 2 is useful when the bookstore needs to know where a customer lived at a particular point in time, while Type 1 is sufficient when only the customer's current address is important.
+
 ```
 
 ***
@@ -191,5 +202,13 @@ Consider, for example, concepts of labour, bias, LLM proliferation, moderating c
 
 
 ```
-Your thoughts...
+The article “Neural nets are just people all the way down” shows that machine learning systems are not completely independent or objective. Even though we often think of AI as something that is powered mainly by computers and algorithms, the article explains that humans play an important role in creating the data used to train these systems. People manually label images, organize categories, and make decisions about how information should be classified. These human decisions can introduce bias into the data and, as a result, into the machine learning system.
+
+One important ethical issue is that biased or inappropriate training data can lead to unfair outcomes. For example, the article discusses ImageNet and how some categories used to label people were considered offensive or sensitive. If an AI system learns from these labels, it may reproduce the same biases when making predictions about new people. This demonstrates that even if an algorithm is technically working as designed, the results can still be harmful if the data or categories used to train it are problematic.
+
+Another ethical issue is the labour involved in creating AI systems. Much of the work behind machine learning happens through people collecting, cleaning, labeling, and moderating data. This work can be hidden from the people who use AI systems, even though it is essential for making those systems function. This shows that AI is not simply created by computers; many human workers contribute to the development of these technologies.
+
+I think this means that people who create and use machine learning systems have a responsibility to think carefully about the data they use. They should check training data for bias, remove harmful or inappropriate categories, and consider how different groups of people could be affected by the system. It is also important to have diverse people involved in designing and reviewing these systems because different perspectives can help identify problems that one group might overlook.
+
+Overall, the article changed the way I think about AI because it shows that technology is not automatically neutral just because a computer is making the decision. Human choices exist throughout the process, from collecting and labeling data to designing the model and deciding how its results are used. Therefore, building responsible AI requires not only good technology but also careful human judgment, fairness, and accountability.
 ```