Reactions to Frontline’s documentary “In the Age of AI”

This discussion is based off the PBS Frontline documentary, “In the Age of AI.” To watch this documentary, visit this link:

PBS Frontline: In the Age of AI

Questions for Discussion

  1. When Kai-Fu Lee says, “Data is the new oil,” and “China is the Saudi Arabia of data,” what does he mean in those statements? Can you describe at least three ways that companies or governments with data monopolies could benefit from having those resources over other companies or governments?
  2. From your perspective as a user or non-user of social media or search indexing technology, do you think social media, searching or website (cookies) tracking of US citizens is important in the race to become a leader in artificial intelligence for the United States, or is that tracking/collection an invasion of privacy and individual rights in our country? What about in other countries?
  3. Can you think of a few careers you might consider pursuing when you graduate from college or graduate school? Once you have a few ideas, visit and plug in those career ideas to see their overall risk to being replaced by automation. How vulnerable are those potential careers to automation? Can you think of 2-3 careers that existed in 1950 that do not exist today? How about 1990? How about 2010? List all the jobs from those years/decades that have become redundant and replaced by automation.

AS A REMINDER, please cite the URL of whatever sources you use to answer these questions.

21 Replies to “Reactions to Frontline’s documentary “In the Age of AI””

  1. 1. He means that data will surpass natural resources such as oil as the most valuable commodity in the world. Data is valuable to large corporations because it allows them to see consumer trends, so that they can appeal to a larger population and build customer loyalty. Hospitals use big data to better diagnosis illness, since more data means more examples for both doctors and computers to observe and base decisions on. One potentially more dangerous use of data is that it allows governments to monitor its citizens, such as by improving the ability of facial recognition software.

    2. I think that even if collecting data is good for the nation as a whole, people should be aware of what they are signing up for and be able to control it in a way that makes them comfortable. Transparency is the key here. One solution could be to offer a cheaper version of certain products for people who offer their data.

    3. Most of the career ideas that I am considering have fairly low risk of automation. Two that do have high risk are economists and computer programmers. There are a lot of other computer science professions besides computer programming that are safe, so that is not as much of a worry. Some jobs from a long time ago that have gone away include rat catchers, manual alarm clocks, bowling alley pinsetters, and lectors that entertained factor workers. Some jobs that are starting to away now are postal carriers, fast food cooks, and secretaries.

    1. Definitely agree with transparency. Cheaper products is an interesting approach to making people willing to give their data, although I think providing a better user experience is enough of a justification that people will be willing to provide it — it just has to be a real, worthwhile improvement. Like you said, in hospitals for example, providing your data has a real impact that helps others. It’s just something people have to come to terms with.

      The issue I have with making things cheaper is that then you’re basically kind of discriminating against the lower class, making them even more exploitable than they already are.

      1. I agree with Christos that making things cheaper for those who offer their data may not be the best solution. Doing so could leave those who don’t have much money to spend on tech products with no reasonable option but to give up some of their privacy.

        1. Totally agree with Shan. I think companies should just be transparent about how and why they use your data and people will voluntarily decide to share it as long as it isn’t an invasion of privacy. Like Shan said, creating monetary benefits may have unintended consequences.

    2. I definitely agree with transparency. I said it too, but adding a choice about data and benefits is interesting, but it could skew the data set.

  2. 1) Given that AI operates entirely/heavily off the data it’s given, companies and governments will constantly be in need of good data, particularly to create new products. Unless transfer of data between countries is not government-regulated as a product (which is unlikely considering privacy laws are drastically different between countries and selling data is effectively selling human private data), countries with more data will have leverage when selling it. China has tons of data simply due to the combination of tech to collect it and a massive population. This kind of data can be leveraged to:
    – sell data to other countries/companies
    – develop more advanced tech and sell the final products with built-in effective AI to other countries/companies
    – develop more advanced tech and intentionally withhold it to have a pure advantage

    2) What bothers me most about tracked user data is that it is used to entirely control what user has easy access to and siphons them off very quickly. I think that having the option to essentially view a non-tailored feed/search results/etc., along with regulation to ensure storage of the data in databases does not easily present users’ identifying information would alleviate many of my concerns about using private data for control without putting the US at a data collection disadvantage.

    3) Considering that my interest is heavily explicitly in the development of AI and/or in engineering, the careers I searched have very low chance of being replaced by AI.
    – Lost 1950s jobs -> gas station attendant, milkman
    – 1990s -> various manufacture-related jobs
    – 2010 -> we’re now in the process of losing cashiers and other retail jobs

    1. I really like the analysis you make that data isn’t just “the new oil” because it’s becoming a valuable commodity, but also because controlling it will give countries leverage in negotiations.

      I also agree that using a user’s data to control what they see is dangerous, both because it increases the addictiveness of social media and because it traps people in political echo chambers.

    2. I agree with your point on how data could become a large force of the global market. It is also possible that countries could strategically sell data to certain countries whose political beliefs they agree with. Do you think selling data to other countries in conflict like India or Pakistan would escalate their conflict and endanger more people?

    3. I think that you raise an important point about the government control of data and how different countries have different privacy laws. Given the growing importance of data and privacy, maybe it would be a good idea to create a set of international laws of privacy to ensure that data isn’t abused by certain countries.

    4. I 100% agree with your point about “having the option to essentially view a non-tailored feed/search results/etc.” The filtering done by these algorithms in places like Google results can be wrong, limiting one’s access to information.

  3. 1. Data is the new oil because it is becoming the most valuable resource for governments and companies. With enough data, humans (or AI that is trained using the data) can predict future events and human behavior. Companies can use data to better target advertising, political groups can use it to influence voters, and governments can use it to track their citizens.

    2. I think that secretly collecting data from any person in any country is an unethical invasion of privacy. Users should be able to choose which data gets collected whenever they sign up for a service.

    3. If I become a physician or surgeon, then my job will only have a 0.4% chance of being replaced by a robot. Since 1950, improving technology has rendered jobs such as milkman (refrigeration) and switchboard operator (better telephones) obsolete. Since 1990, many manufacturing jobs have disappeared due to automation. Since 2010, some retail and service jobs have been replaced by computer systems.

    1. I mean, I also think that there’s lots of arguably purely positive applications of AI that require data. Healthcare and various forms or R&D come to mind for me.

      I totally agree with what you’re saying, just wanted to point out that the data itself isn’t inherently bad, just that some groups will have unfair access to it to make products and then profit from it—even if the products are definitely societally beneficial.

    2. I agree that people should be able to decide what data they want collected from them when they sign up for a service. Also, services should have to ask users before they sell or share their data with other services because users might not be okay with any company having access to their information, maybe only a few that they can benefit sharing their data with.

    3. Letting users pick data to be collected would probably skew the dataset because they would leave out anything that doesn’t make them look good. Everyone would have the maximum social score. However, I do agree that people should know what data is being collected.

  4. 1. Data is the new oil because in the future it will surpass oil as being the most valuable resource in the world. China currently has the most efficient and expansive system for collecting data on its citizens through apps and cameras, so it will be a world leader of data collection. A food truck company that has data on traffic patterns and pedestrian routes can most strategically park their trucks where they will get the most business. Also, a police station that has data on where and when car accidents happen can help them know where and at what time to go to different intersections to prevent accidents. Lastly, having data on people social media use, like what accounts they are clicking on and what posts they are liking, can predict when a new social movement is going to take off and where it is originating from. This can help politicians learn what issues voters want to hear about and what to say in order to attract voters of a certain demographic/group.

    2. Tracking users’ actions online and on social media is not the part of data collection that makes me uncomfortable or that I think is an invasion. I think this kind of tracking has been a part of the internet for long enough that when I am on social media or browsing the web I expect that data is being collected on my searches. However, the part of data collection that I do think is an invasion of privacy and just straight up creepy is the cameras that China has on the their street and the facial recognition they are using with the cameras. With this kind of data collection there is no escaping companies or the government’s reach of what they know about you. Someone cant just leave their phone in another room and walk around and have conversations in private because there would be cameras tracking their every move. It is this kind of inescapable data tracking that is too much for my comfort level because if data collection starts to be used in a harmful way, like with the Uighurs, no one can go “off the grid”.

    3. After college or graduate school I might want to become a electrical engineer which has a 10% chance of being replaced by automation. In the 1950s people worked as pin setters in a bowling alley, ice cutters, and switch board operators. In the 1990s people would rent videos to watch, but now with streaming services the job of video rental clerks no longer exists. It is harder to find jobs from 2010 that don’t exist anymore but many jobs such as truck driver, taxi drivers, carpenters, and refuse and recyclable materials collectors have already begun to dwindle in numbers as automated system replace them.

    1. I agree with your different reasons for why data is becoming increasing more valuable. Why do you think that you find cameras uncomfortable but online tracking less so? If it’s just because you have gotten used to it, do you think it would ever be possible to get used to physical surveillance too (not saying that this should happen, just wondering)?

    2. I read the part about police using AI to determine where car crashes most occurred and I started thinking of other police applications for AI that are more similar to Chinese facial recognition. While the Uighurs are oppressed by the facial recognition/AI system China has implemented, could police use it to make racial disparities in police action/arrests less? Would it actually make it worse? Or is it a problem that can’t be solved by AI?

  5. 1) When Kai-Fu Lee says “Data is the new oil,” he means that it will play a large role in determining which countries will have power in the global space like how oil gives the countries who have it power now. There are many reasons for this, all of which come down to machine learning. Because these ML algorithms need a lot of data to be trained, it will be much easier for data-rich countries to implement them. China is already using algorithms from their surveillance data to predict who may act against the government. While I don’t necessarily agree with this policy, it does have some positives in terms of national security. Transportation is also a big one. Countries with lots of data from drivers can better implement the infrastructure for autonomous transport, saving time, money, and possibly the lives of drivers. Finally, healthcare can be vastly improved with enough data. As the documentary showed, machine learning can predict diseases like cancer far before a human doctor can.

    2) I think all data collection should be on an opt-in basis. If websites were transparent about what data they collect and collected it for a good reason, I would probably opt into most data collection to improve these sites’ algorithms. People’s data definitely has beneficial effects on the functionality of these sites. However, the current secrecy around this data collection makes me a little skeptical about what is being done with my data as of right now.

    3) Though says the field of computer science is at risk of automation (lookin’ at you Codota), I think it will be relatively safe because *someone* has to write the algorithms for automation. In the 1950s, many jobs existed that have since been replaced by computers. For example, NASA had rooms of people responsible for performing rocketry calculations before computers made this process much more efficient. Toward the end of the 20th century, many industrial and machining jobs were replaced with computer-controlled machinery that could work much faster and more accurately than humans ever could. Toward the 2010s and still to this day cashiers are being replaced by touchscreens and grocery store self-checkouts that can process customers faster and don’t need to be paid to do so.

    1. I agree with the opt-in aspect of data collection, but how far is too far? I.e. what data would be wrong for data companies to collect (if any)? And, assuming the data collection is opt-in, should it be left to people to decide what’s best for them or should it be regulated?

  6. He means China has a massive amount of data owned by the government. Three benefits to having a monopoly of data are governments will be able to make better laws and be better at policing, companies will have an economic advantage as AI and machine learning become commercialized, and companies will grow more efficient faster.

    I don’t think that data collection is an invasion in privacy because in most cases it is helpful. The government should get access because the only way to become an AI leader will be to have massive amounts of data. That said, I do think the government should have to be transparent about how our data is used. I think in other countries data collection is fine, however, they should have the same stipulation about transparency and many governments won’t put that in place.

    I think I want to go into consulting or venture capital; both these jobs are at a no worries level. I think switchboard operators were a thing in the 1950s that don’t exist today along with typists. A job from the 1990s that doesn’t exist today is disco singers (maybe 70s I don’t really know). Finally, a job from 2010 that doesn’t exist today is factory workers like the ones in the movie.

  7. 1. Saudi Arabia has been one of the largest producers of oil on the planet, so countries began to rely heavily on their exports of oil. This led them to gross massive wealth from other countries who, without the fossil fuel, couldn’t function. China has the same “monopoly” with data; they are miles ahead of many other countries on how much data they are able to collect and in which ways. Facial recognition, for example, is becoming a staple of Chinese society. While the purpose of it is stability and order, it is often used to suppress dissent and oppress minority groups. For example, this technology is being used extensively on Uighur Muslims in western China to reeducate them and increase stability (i.e. reduce the likelihood of another riot).

    2. I think that the ways in which companies collect data now is quite intrusive. While it makes sense for specific ads to be tailored to specific people – it is both beneficial to the company selling the product and the consumer looking to buy the product – I find it wrong that a person’s data can be for sale without consent. The extent to which your private life is sold to create tailored ads is also far-reaching to me. I think that this is slightly different when your data is being used to further national security. While I reject the idea of “if you have nothing to hide, you have nothing to worry about” because it insinuates a complete lack of privacy, I think that government can look at people they deem “threats” to national security with slightly more scrutiny. Of course, this could lead to discrimination and selection based on race/ethnicity (much like at airports with TSA screenings), so it’s very much a toss-up to me. I think that this applies similarly to other countries. If I apply this framework to China, I find that the extent to which they’re collecting data is far too overreaching.

    3. Most of the jobs I’m considering after I graduate fall into the political/legal sphere, and according to the website, there is little risk of these jobs being automated. When thinking about jobs from the 1950s that are likely to be automated, I immediately thought of Assembly Line and Secretary jobs, and they are both over 95% likely to be automated (I mean, they already mostly have been). This is similar to truck driving, a job that was well-documented as being automated in the movie. It has a 93% chance of being automated in the future.

Leave a Reply

Your email address will not be published. Required fields are marked *