Get All Weeks Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Week 01: Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Module 1 Quiz
Q.1 What does big data offer for computational social science?
- Empirical methods
- Integrated induction
- Theoretical analysis
- Computer simulations
Q.2 In lecture, we worked with a list of several characteristics of “big data”. Which one was NOT part of our characterization?
- Machine learning is often the only way we have to make sense of it
- It often is not sampled, but it is still representative of society
- The data is often an unavoidable byproduct of digital interaction
- Different sources are often used in a complementary way
Q.3When we try to integrate messy data from different sources and see the big picture from it, this aspect of big data is known as:
- Data-fusion
- Real-time fusion
- Digital footprint
- No sampling
Q.4 More than 2 out of 7 (some 30%) of people on Earth are on Facebook. Therefore, Facebook gives a very representative sample of the human population.
- True
- False
Q.5 Why did machine learning become so effective after we had ‘big data’?
- Massive data allowed machines to learn from data
- A series of serendipitous coincidences
- Companies had the funds to invent it
- National Security Agencies had the funds to invent it
Q.6 Sometimes, in the industry, ‘big data’ is characterized by ‘the four Vs’. Which one is NOT one of them:
- Velocity
- Volume
- Verifiability
- Variety
Q.7 As early as in the 2012 Presidential campaign, the Obama campaign spent more money in data management projects than TV ads. Which of the following data sources did they collect to create some 16 million unique voter profiles?
- Employment documents matched with household surveys
- Citizen profiles from different National Security Agencies
- Telephone survey results matched with census data
- Tweets, Facebook postings, TV setup boxes
Q.8 What is a so-called “filter bubble”?
- An opinion bubble that only shows what agrees with government filtering
- An opinion bubble where you only see the things you want to see
- A network bubble where companies control which friends you can interact with
- A news bubble where you only get information from one social network
Q,.9 What is a so-called “echo chamber”?
- Echoing the most important aspects that result from data fusion
- An information isolation where your opinion is reinforced by similar opinions
- Echoing how people with different opinions see the world
- The technique to identify fake news by echoing them back to you
Q.10 What is a big data approach to calculate the inflation rate?
- Data scientists collect the receipts from consumers in different counties
- Data scientists model the empirical data and calibrate the inflation simulation
- US Bureau of Labor Statistics sent a large number of staff to collect prices manually
- Data scientists collect the prices of hundreds of online retailers
Q.11 We have seen an example in lecture about what call centers are training when they announce that ‘the call might be recorded for quality and training purposes’. What is being trained?
- Computational social scientists in training, who learn natural language processing
- Professionals who detect unprofitable clients
- Algorithms that match the caller’s personality with the call center representative
- Government officials who design regulations for cybersecurity
Q.12 How does today’s machine learning translation get so good?
- They feed machine a lot of textbooks on translation and let it learn the patterns
- Linguists incorporated ever more comprehensive grammatical rules into the algorithms
- Linguists incorporated ever more comprehensive vocabulary words into the algorithms
- Machine learning algorithms identified relations between humanly translated texts
Q.13 You are part of an online retail company that has the goal to grow quickly. Recommending new clients exactly what they need is an essential part of the successful growth strategy. What kind of recommender system do you recommend?
- Cooperative message filtering
- Cooperative prediction filtering
- Collaborative filtering
- Content-based filtering
Q.14 What is the difference between “content-based filtering” and “collaborative filtering” for recommender systems?
- The former is based on content about others and the latter on collectively programmed machine learning
- The former is based on past data from the individual and the latter on data from other individuals
- The former is based on content and the latter on collaboration of the company with the client
- The former is based on an indexed table of content and the latter on the combination of different filters
Week 02: Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Natural Language Processing (NLP) Assignment Task
Q.1. For this assignment, you will gain insight on two ‘personalities’ by doing Natural Language Processing on a sample speech from each personality. This quiz is designed to assign different learners two personalities at random, so you might see slight variances than your peers.
You are being assigned the following two personalities:
Donald Trump and Hillary Clinton
You can click on each of their names to access their speeches or you can find them as a resource under “Course Resources” in this course.
Are you ready to get started?
- Yes, I have obtained both speeches for my two personalities and am ready to get started.
- No, I am not ready to do this assignment, but I know my two personalities and have either already downloaded their speeches through the links above or will access them later under “Course Resources”.
Module 2 Quiz
Q.1 In lecture we saw several drawbacks and limitations of ‘dig data’. Which of these was NOT one of them?
- Representativeness ≠ Generalizability
- Data ≠ Reality
- Discrimination ≠ Personalization
- Correlation ≠ Causation
Q.2 consultant gets hired by a company to study how their newest product is received by consumers. After studying Twitter, the consultant presents the conclusions. What is an immediate caveat you raise during that meeting?
- People do not mention consumer products on Twitter
- The consultant must have hacked Twitter, which is unethical
- Doing this is illegal
- This data may not be representative
Q.3 In lecture, we have seen an example of a big data study that used mobile phone trace data. It seemed to suggest that in an African country, there are many more young women than in Latin America. What did we take away from this?
- Data-fusion led to the end of theory, which can be tackled with artificial intelligence
- Many young men left Africa as economic refugees to Europe, while women are better treated in Latin America
- Skewed mobile phone penetrations introduced a sampling bias
- Mobile phones are the best source to identify population characteristics in developing countries
Q.4 Once everybody has access to some form digital technology, the digital footprint will be representative of society.
- True
- False
Q.5. What is “predictive policing”?
- Machine learning applied to determining parole or pardons for convicted criminals
- Analytical techniques in law enforcement to predict potential criminal activity
- The use of body cams to monitor the activity of police offers and predict offenses
- Digital big data footprints to predict how well police officer do their job
Q.6. What does Prof. Hilbert mean by “data is not equal to reality” in the case of algorithm predicting if homicide parole candidates will commit homicide again at an accuracy of 60-70 % accuracy?
- The big data footprint is not representative of all people
- Algorithm prediction doesn’t mean it happens or will happen in reality
- It is immoral to use algorithm to make such decisions on human life in reality
- Algorithm prediction is not higher than human psychologists’ prediction
Q.7. Why is data mining, per definition, “always statistical (and therefore seemingly rational) discrimination”? Because the point of data mining is to:
- Statistically make sense of social hierarchical structure
- Conclude the general rules that applies to everyone
- Use machine learning, which uses code manipulated by computer scientists
- Provide a rational basis upon which to distinguish between individuals
Q.8. According to Washington Post, Trump’s administration alleged that manufacturing decline increases abortion, infertility, and spousal. What is the mistake in this allegation?
- It gives rise to discrimination due to the personalization
- The result of the data has meaning but not meaningful
- It assumes causation from correlation
- It is based on biased data
Q.9. A new big data study that works with digital footprints from baseball stadium cameras claims that watching baseball causes people to eat ice cream. Immediately, many things come to your mind of what could be wrong with this study. Which of the following are your concerns?
- The sample might not be representative
- Summer might be a confounding variable
- Eating ice could cause people to watch baseball
- The study might not be replicated and might be a victim of the replication crisis
Q.10 The original ‘Google Flu Trend’ algorithm predicted the outbreak of the flu from Google searches. Why did the algorithm from 2009 not work anymore when it was applied again in 2013?
- Replication studies always fail (“replication crisis”)
- It was not Google itself, but some academics who did the 2013 study
- There was less flu variation in 2012/2013, so less data to work with
- Reality changed: people Googled different things a few years later
Q.11What does Lucas’ critique point out?
- Making the world a better place will require help from artificial intelligence, but it is dangerous to empower machines
- Unethical big data work has to be regulated, especially automated weapons like slaughterbots
- Changes in policy will systematically alter econometric results if the used computer simulations are agent-based
- Using data to design interventions, one cannot assume that the same predictions will still hold after the intervention
Week 03: Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Module 3 Quiz
Q.1. The newest aspect of today’s digital revolution is that for the first time in history, we have started to discuss and aiming to build artificially intelligent machines over the recent decade.
- True
- False
Q.2. A machine passes the Turing test for intelligence if:
- A human cannot distinguish the answers of a machine from the answers of a human
- A computer outperforms the best performing human in a given field (chess, image recognition, driving, etc.)
- A Turing machine (a universal computer) has become a general purpose intelligence
- A data scientist is able to turn a testing set into a validation set, while fine-tuning the size of model to prevent overfitting
Q.3. What was the important new ingredient that allowed AI to start its ultimate victory run across the global economy and society during recent years?
- Availability of data
- Supply of computer science trained millennials
- Public attention, such as through chess and Go tournaments
- Commercial support
Q.4. Why do some machine learning researchers NOT consider themselves in the field of AI?
- Because digital footprint cannot explain individual behavior
- Because artificial intelligence is already solved, and the new frontier becomes machine learning
- They believe that artificial intelligence is not possible, which is proven by machine learning
- They come from other disciplines (like statistics) and are interested in the questions about data instead of mechanisms of AI
Q.5. Graph search is probably the first method taught in an AI class. What is the graph search approach to solve the goat, wolf and cabbage river-crossing problem?
- Assign rules to each agent (goat, wolf, and cabbage) and simulate the process to find the optimal choice
- Write down the cost and benefit of each choice, model the process, and calculate the equilibrium state
- Create a graph of logical statement and calculate the simplest combinations logical operators
- Create branches for all possible combinations at each state and search for paths that satisfy the final condition
Q.6. A traditional programming approach to face recognition would feed a computer with a program to recognize faces and with data to detect where in an image a faces. What does a machine learning approach do instead?
- Feed the computer with images with blacked out faces, having it discover the missing part in future images
- Feed the computer with knowledge about biological characteristics of faces and face images and have the computer produce simulated data on idealized faces
- Feed the computer with image data and known face-presence output and let the computer discover the program with the relationship between them
Q.7. What do we call the part of the dataset that is being reserved for a final evaluation of how well a machine learning algorithm captures patterns in the data?
- Evaluation data
- Possible data
- Testing set
- Training set
Q.8. What was the key that helped industry to make speech-to-text solutions sufficiently robust for applications in the commercial market?
- Deep neural nets for machine learning
- Big digital trace data
- Natural language processing
- Real-time recording by dialogue systems
Q.9.In modern dialogue systems, what is the difference between a ‘call flow paradigm’ and the ‘query paradigm’?
- A call flow paradigm approach describes answering only questions asked and the query paradigm approach describes asking contextual questions to carry out conversation
- A call flow paradigm approach describes asking contextual questions to carry out conversation and the query paradigm approach describes answering only questions asked
- A call flow paradigm approach describes answering questions without feedback and the query paradigm approach describes querying possible feedback and the second response
Q.10. What was the so-called ‘AI winter’?
- A period of disappointed expectations for artificial intelligence during the 1970s, leading to less interest in the field
- A period during the Cold War when Russia and the US were fighting to establish themselves a leader in the field
- An artificially induced winter, used as a counter-reaction to global warming, controlled by AI
- Artificial intelligence setting of a chain reaction resulting in a nuclear winter across the globe
Week 04: Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Module 4 Quiz
Q.1.The United States Public Health Service conducted a syphilis study in Guatemala from 1946 to 1948. At least 83 people died from it. What did the U.S. government do in this study?
- It searched for volunteers to participate in the study, assuring that no one would get seriously ill
- It purposefully infected people with the deadly syphilis disease, without their consent or knowledge, often purely for reasons of documentation
- It searched for volunteers who would give their consent to participate in the study, without assuring that any could get seriously ill
Q.2. Why was Laud Humpheys’ famous tearoom study on homosexuality so controversial?
- He did not explain in detail what his study was about, what he was testing for and why he was testing for this
- He deceived the study subjects, and was not clear that his activities were part of a systematic study
- He discovered homosexual activities while during a time they were illegal and didn’t report it to the police
- Because the results he obtained confirmed existing stereotypes and endangered people with homosexual orientations
Q.3.Which of the following is NOT included in the principles of respect for person in the Belmont Report?
- Privacy, subjects need to be assured that their privacy will not be compromised with a study
- Information, subjects need to be informed to a given degree
- Comprehension, subjects need to be able to understand the given information
- Voluntariness, subjects need to be able to decide freely if to participate, given what they know
Q.4. You gave consent to answers some survey questions about personal preferences for a research study. Most of them were about political opinions, so you thought the study was about that. Later you found out that the study was about transsexual orientations, which you found offensive. Was this obvious purposeful deception unethical?
- Yes
- No
Q.5. Your employer gives you the approved task to webscrape social media data on profile pictures and likes. While you are at it, you might as well scrape their age, usernames, location and some other existing data from the accounts. You never know if your employer might need it later on. Would this be unethical?
- No
- Yes
Q.6. Which of the following is NOT part of the mandate of an Institutional Review Board (IRB)?
- Managing conflicts of interest
- Minimizing the risks of research involving human subject
- Maximizing the benefit of research involving human subject
- Making sure the collected data is not biased
Q.7. For ads-driven social media platforms, those who buy ads are the customers. What is the “product” being sold?
- The content featured on the platform
- The behavioral change of the user
Q.8.What is “nudge” in behavioral economics?
- Nudge proposes positive reinforcement and indirect suggestions as ways to influence the behavior and decision making of groups or individuals.
- Nudge is a strike with the point of the elbow, the part of the forearm nearest to the elbow, or the part of the upper arm nearest to the elbow.
Q.9. What is the confirmation bias?
- The active formation of beliefs based on what might be pleasing to imagine.
- The tendency to favor information in a way that supports one’s prior beliefs or values.
- The bias that confirms that election outcomes nowadays are typically fake news.
Q.10. What is fiduciary duty?
- An obligation to act in the best interest of another party.
- The duty to reveal when an obnoxious person (‘douche’) only exists in fiction (fi).
- A suggestion to consider the interest of another party.
Q.11. A company decides to create a campaign on an adult social media site to promote the use of condoms. It creates two versions, one tailored to elderly with homosexual orientations, and another general one. They randomly show each version to 10,000 users. Is this unethical?
- No, this is the kind of usual A/B testing that happens millions of times every day at all social media sites.
- Yes, this will reveal both age and sexual orientation, which are both legally protected against discrimination
Get All Course Quiz Answers of Computational Social Science Specialization
Computational Social Science Methods Coursera Quiz Answers
Big Data, Artificial Intelligence, and Ethics Coursera Quiz Answers
Social Network Analysis Coursera Quiz Answers
Computer Simulations Coursera Quiz
