On the frontier of artificial intelligence journalism

In an era when venerable news publications such as the New York Times and the Washington Post are being maligned by key societal figures accusing them of “leftist” bias, when publications such as Breitbart or similar media sources often deemed “conservative” unabashedly present stories slanted to reflect political views far right of the majority of public sentiment in the country and when a significant portion of the general public acquires its “facts” about the nature of world events from Twitter, Facebook and a network that used to fly a banner emblazoned with the words “fair and balanced”, bias both inadvertent and intentional has become an insurmountable impediment to attaining the “truth” about what is going on around us. In the face of such a large, sprawling situation, we sometimes get lost, failing to understand the meaning of our daily encounters and  the most appropriate intent we should seek regarding any of these events. Enter knowherenews.com.  This website proposes that computer software can filter reports of world events and effectively sanitize them of ideological bias.

From it’s About page, knowhere.com describes its intent this way:

Our technology scours the internet, discovering stories and identifying narratives, evaluating the factual claims and bias in reporting. This analysis informs a set of algorithms that write our three perspectives on every controversial story. Once articles have been drafted, our journalists review and edit the story before publication. This helps teach our algorithms to gain an ever-deeper understanding of quality and bias in the news.

Bias in EveryDayLife

We as a nation have always struggled with reality vs. fantasy, often succumbing to what Kurt Andersen, author of Fantasyland calls our “promiscuous devotion to the untrue”.  Take a look at the following infographic to get an idea how knowherenews.com is attempting to ameliorate this fantastic promiscuity (click on the image).

According to a Buzzfeed survey:

• 75% of people are fooled by headlines
• people who rely on Facebook as a “major” news source believed fake news 83% of the time
• people who rely on Facebook as a “minor” news source believed fake news 76% of the time
• people who didn’t rely on Facebook as a news source believed fake news 65% of the time

From another source, “Americans have such an appetite for fake news that creating it is becoming its own cottage industry.” Sounds a bit like a “promiscuous devotion to the untrue”, yes?

Where Do You Stand?

Think about the content presented in the infographic above for a bit, then consider this question.  Do you believe the stats presented within it?  Do you believe the content the linked infographic presents?  Why? Do you believe the follow-up text (presented here on this site) after the infographic?  Why or why not? Where do you “stand”? The perspective from which you assume truth will shade your reception as to whether the “facts” presented by any of the above sources or from anyone else are considered by you to represent “truth”. More specifically, take another look at the following knowherenews.com headlines (originally presented for comparison on futurism.com).

---

---

Comparing the above statements, can you separate the statement which seems to reflect more “factual reporting” from the more biased versions? Which is which? Why? How can you tell?  If you did not have all three for comparison, could you label the biased reporting as such and temper your responses to any single version based on your knowledge of what “bias” looks like?  If you only read one of these, would you believe it represented biased reporting?

To parse the above headlines a bit, identify the apparent content of the news story—a certain type of question in the 2020 census. Now compare the two slanted takes to your understanding of what the actual story is supposed to be about.  The fact is that the 2020 census is to contain a citizenship question. To comment on the combative actions some group might launch against another or the reactions of one group to the citizenship question itself is not the point of the story.  Each of these is commentary on the fact—not news.  This is not to say that the actions of Californians or the reactions of so-called liberals might not, in themselves, represent worthy, standalone “news” stories.  However, neither of these descriptions is about the content of the 2020 census story.  [Please read through the article yourself to verify this assessment.]

To split hairs a bit further, compare these statements.  “U.S. to add citizenship question to 2020 census”; “A citizenship question will appear on the 2020 census”.  Staunch adherents of contemporary writing style (who clearly play a negligible role on this blog) will decry the presence of passive voice in the latter statement.  And yet, to “accuse” the “U.S.” of including a politically loaded question on the 2020 census is a different statement than to refer to the more placid (non-editorialized) “appearance” of such a question.  Active voice requires an actor, in this case the “U.S.”.  And yet, given these political times, does such wording constitute bias?  Perhaps such wording cannot be avoided (which is to say perhaps writers cannot avoid such wording).  Splitting linguistic hairs in this way is not meant to pronounce either of these statements as good, bad, better or worse.  It is presented (passively?) here merely to illustrate that word choice matters and triggers differing responses in various readers—particularly unthinking readers.  The conclusion should be that we, as readers, need to be working harder at understanding the “information” presented to us.

knowherenews.com in the Flesh

If you have not yet perused the knowherenews.com site, the user interface employed on the site works like this.

The three words in the graphic above are displayed above a body of text—the news story.  When you select a word, that word becomes bold and an associated body text is displayed.  As depicted in the above graphic, the larger, bold heading announces a version of the news story which, according to knowherenews.com, reflects the heading you select. When a reader clicks on the word “left”, a left-leaning story is displayed on the screen below it; when “right” is selected, a right-leaning story appears.  Below is a set of categories and associated headlines which appeared on the September 23 issue of knowherenews.com.

What do you think? Does the center headline really focus on the facts in the story? From the headline, do you think the actual story is about (1) denial of access to green cards or (2) the reaction of “US rights groups”, or (3) relatively punitive actions expressed toward immigrants by the person in the White House?  In the “unbiased” version of the story appearing on the knowherenews.com website, the text describes the actions of the current administration in a fairly straight forward manner, but not the reaction of rights groups. The writing in the so-called “unbiased” story does appear fairly straight forward, relatively balanced and factual.  The headline, however, is misleading.  So do you think this process works? Perhaps that depends on what you might like to get out of interacting with this AI-driven site.

Reviews Are In

To help you decide how/if you might want to respond to this idea of artificial intelligence filtering news reporting, consider the reviews from the following sources:

• Futurism
• Motherboard
• Techcrunch
• Masters of Media
• Digital Trends

Responsibility

Can and should we appropriately interject a surrogate adjudicator of literary bias as a translator of truth? Even if we do, can we trust that this translator (as a creation of humans) is not intentionally or inadvertently (and perhaps inevitably) influenced to skew results in some biased manner? One of the clear benefits of the knowherenews.com site is it gives us access to common situations described from apparently different perspectives.  This AI-generated journalism might be quite informative in helping us understand what bias sounds like, how we can spot it in ourselves and how we might resist it.

Perhaps the question is not so much can this AI-based journalism successfully present unbiased stories.  Perhaps the real question is whether we should rely on or even desire such a situation.  Do we truly serve ourselves by abdicating such a responsibility?  As the reviews seem to suggest, maybe this idea will work.  Establishing the ability to read unbiased news can certainly help us understand the world.  But is this really the solution toward which we should strive?