Machine Learning and Media Bias
How does artificial intelligence influence our political landscape?
The last decade has been a technological whirlwind; those of us born at the turn of the century or before have experienced several technological innovations at breakneck speed. The internet became more of a fixture in our lives (rather than merely a utility), smartphones have made it possible for a streamlined internet experience in our pockets, and new online media platforms have nearly replaced many traditional forms of news and entertainment consumption once limited to only broadcast radio and television. This has changed almost every aspect of our society; commerce, social media, media distribution; more importantly, it has changed the way people find information on current events. News has become less centralized; no longer are the news networks a primary source that people use to understand current events. The internet is now where most of this information finding takes place. What does this mean for us, our society, our politics?
Curating Our Online World
What is it that determines what we see on our Facebook feed? What trends do we see on twitter? What kind of headlines are we being exposed to while browsing the internet? The answer may seem simple, but it is really determined by our behaviors and interactions with content. The process of this learning of behaviors is made possible through a concept in artificial intelligence known as machine learning.
What is machine learning? According to the top result on google, machine learning is defined as “an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. Machine learning focuses on the development of computer programs that can access data and use it to learn for themselves”. (E. 2020)
Without getting too complicated, computers store information and curate content for us based on our online interactions with different content, search engine searches, as well as our purchase histories. Have you ever seen ads so relevant to you that it is almost uncanny?
It is important to distinguish machine learning from AI; machine learning is artificial intelligence that learns patterns without being programmed.
“Artificial Intelligence is the overarching concept of machines performing intelligent tasks, and machine learning is the idea of giving those machines the tools to learn on their own — AI does play a large role in how machine learning operates.”
(Lawrence 2020)
Knowing this, how does this concept affect what news we are exposed to? How does that, in turn, affect our political biases?
Digital Echo Chambers
A Buzzfeed journalist did an experiment in which he only interacted with right-wing content online, and found that his news feed was being bombarded with conspiracy theory articles, as well as suggestions for meme pages that echoed the sentiments of the pages that he was engaging with. (Broderick 2017) This in turn created an echo-chamber of (often false, misleading, offensive) information and news being displayed on his feed.
These findings can also be applied to left-leaning content as well; if you tend to only interact with pages that are left on the political spectrum, you will have a personalized feed that contains a lens that is adjusted for your sentiments.
This is important to think about, because many people now rely on Facebook as their main source of news. Luckily, most people surveyed do not trust it as a news platform. (Lunden 2019)
But what about those that do? Twitter has been accused of ‘tweaking algorithms’ and manipulating information to advance political interests. Similarly, Facebook was criticized for not fact-checking political posts, to which they (mostly) reversed this decision following scrutiny. Many Republicans believe that “the media” in general has a left-wing bias, and therefore turn to more right-leaning news outlets such as Fox News or Breitbart.
Knowing this, how can we differentiate fact from fiction? Who is a trusted source of information? Interestingly, machine learning is also being used to combat “fake news”, or, more bluntly, propaganda.
Machine Learning to combat Fake News
Experts at MIT have developed a machine learning algorithm that can detect fake news with stunning accuracy. (O’Brien 2019)
This makes machine learning both a blessing and a curse; it can cause us to have a “digital echo chamber” where we are only exposed to fake news. On the other hand, it can also be used to combat the spread of false information.
With this, though, is a catch 22. We need the information from human interaction to inform machine learning algorithms. (Shu 2017)
“First, fake news is intentionally written to mislead readers to believe false information, which makes it difficult and nontrivial to detect based on news content; therefore, we need to include auxiliary information, such as user social engagements on social media, to help make a determination.”
excerpt from Fake News Detection on Social Media: A Data Mining Perspective
On that note…
So…what do we do with all of this information about machine learning and its role in affecting our political biases? My answer to this question is that people need to not rely on just one media outlet or a patterned algorithm to determine what is best for them to know. We have the capability to watch press briefings, follow things on twitter as they happen, watch livestreams, all before the information gets filtered through a media outlet; there are ways in which we should look at information in order to not be manipulated. Machine learning has created a lot of conveniences for society, but with it has come many complications.
References
Broderick, R. (2017, March 08). I Made A Facebook Profile, Started Liking Right-Wing Pages, And Radicalized My News Feed In Four Days. Retrieved October 21, 2020, from https://www.buzzfeednews.com/article/ryanhatesthis/i-made-a-facebook-profile-started-liking-right-wing-pages-an
E. (2020, May 29). What is Machine Learning? A definition. Retrieved October 21, 2020, from https://expertsystem.com/machine-learning-definition/
F. (n.d.). Working to Stop Misinformation and False News. Retrieved October 15, 2020, from https://www.facebook.com/formedia/blog/working-to-stop-misinformation-and-false-news
Lawrence, A., & Lawrence, A. (2020, August 21). How Machine Learning Affects Our Daily Lives. Retrieved October 15, 2020, from https://www.illinoisscience.org/2019/11/how-machine-learning-affects-our-daily-lives/
O’brien, N. J. (2018). Machine Learning for Detection of Fake News (Unpublished master’s thesis). Massachusetts Institute of Technology. Retrieved October 15th, 2020, from https://dspace.mit.edu/bitstream/handle/1721.1/119727/1078649610-MIT.pdf?sequence=1&isAllowed=y
Selyukh, A. (2016, May 13). Amid Allegations Of Bias, Facebook Explains How ‘Trending Topics’ Works. Retrieved October 21, 2020, from https://www.npr.org/sections/thetwo-way/2016/05/12/477874985/amid-allegations-of-bias-facebook-explains-how-trending-topics-works
Shu, K., Sliva, A., Wang, S., Tang, J., & Liu, H. (2017). Fake News Detection on Social Media. ACM SIGKDD Explorations Newsletter, 19(1), 22-36. doi:10.1145/3137597.3137600
This was an excellent, prescient article. It’s eye opening to realize just how many people accept an algorithm filter on the information they view. Have you researched into movements to eliminate machine learning? I expect it would be small but would certainly expect that the push is there (albeit in remote pockets of society.)
Hi, your post was very informational. You are right, Nowadays, whatever news pops up in our feed we tend to believe it. Because of social media, we are constantly checking facebook and twitter for news. I totally agree with you, that we should just rely on one media outlet, which we can betrusted. I think sometimes it does get hard to differentiate if the source of news should be trusted. Because they manipulate the news in a way that our mind could not accept the fact that everything should not be trusted on social media. I think MIT has done an excellent job to developed a machine learning algorithm which can detect fake news.
A good read on a fascinating and pertinent topic. I’d like to see a follow up/continuation on the flaws of AI learning/software in government/law enforcement uses. Tons of people have heard of the Twitter bot that people taught to be racist, but few are aware of some of the biased, flawed systems that are already in place and being used.
This article talks about a few of these issues: https://www.theguardian.com/inequality/2017/aug/08/rise-of-the-racist-robots-how-ai-is-learning-all-our-worst-impulses
This article particularly focuses on biases in facial recognition AI: https://www.washingtonpost.com/technology/2019/12/19/federal-study-confirms-racial-bias-many-facial-recognition-systems-casts-doubt-their-expanding-use/
People live under the false assumption that technology/algorithms are without prejudice/bias, but when people are the ones teaching the tech, stuff is bound to slip through the cracks.
Excellent post. The internet is full of echo chambers even on the smallest of sites. Not only should we be aware of this but also research information we review and not take it at face value. Even I have seen posts online and assumed they were true just because the information sounded plausible. The first step to finding the truth is accepting you might be being lied too.
Machine learning! I should have read your blog before writing my own to connect even better what I have to say about “outside forces” that drive us, and I plan to link to this really helpful blog that will expand on an idea I really didn’t address as well. This is a really important concept for anyone to understand and you wrote it in a way various users may understand.