Have you ever wondered why Google News surfaces some articles higher than others? If you answered “no” to that question, you probably don’t work in marketing or journalism.
Google News is among the largest traffic drivers for online news sources and it plays a huge role in our day-to-day lives.
Think about it. Let’s say you’re a reporter who writes about barns. You travel the countryside looking for the most interesting barns and you write barn stories.
When people search Google News for the word “barn,” you’d probably hope it surfaced your work.
And if the barn beat was as competitive as, say, the artificial intelligence beat, you’d probably expect space in the Google News feed to be equally competitive. Or maybe you’d figure some big time barn news outlet was gaming the algorithm or buying priority.
Folks. I don’t think that’s how any of it works.
A mystery?
As of the time of this article’s publishing, if you navigate to Google News and type “artificial intelligence queer” into the search box and then hit enter, you should be introduced to a lot of my work.
Using Incognito Mode, Google News returns 120 results for “artificial intelligence queer,” of which 37 are stories that I wrote.
Of those 37 stories I wrote, only nine have even the slightest connection to LGBTQPIA+ issues.
Here’s what that means: my journalism represents approximately 30% of the Google News search results for “artificial intelligence queer,” even though only 24% of my listed articles are about anything queer-related.
An easy answer?
My colleague here at Neural, Thomas Macaulay, obtained the same results when he replicated my search, so we started looking a bit deeper.
It turns out, when he searches for “artificial intelligence Arsenal,” more of his stories surfaced than when he just searched for “artificial intelligence.” But only a few of Tom’s pieces made the list either way.
When I search for my pieces there’s a huge difference between just searching for “artificial intelligence” (only one result) and “artificial intelligence queer” (37 results).
So why does it matter? Because Google News is obviously pulling the words “Arsenal” and “queer” from our author profiles.
It seems really basic to think that Google News chooses which articles to surface based on author profiles, but it sort of makes sense and adds up. Puzzle solved right?
Wrong. The puzzle is not solved
My author profile (which is hosted on TNW’s website and has nothing to do with Google) also says I cover cannabis, politics, and gaming.
But when I search for “artificial intelligence cannabis” I get no hits. If I search for “artificial intelligence politics” I get one and changing it to “artificial intelligence gaming” nets me a whopping three pieces.
Here’s where things get interesting. If I search for “AI queer” the number of my articles that surfaces drops to 21, but if I search for “artificial intelligence pronouns” it leaps to 44 out of 120.
The Google News algorithm apparently pulls information from my profile, but it associates me with AI and queerness more so than any other combination of keywords or any singular search term.
In fact, if I substitute “machine learning” for “artificial intelligence,” a term I don’t use at all in my profile, it surfaces zero results on its own and 27 results with “queer” added.
The problem
I’m not sure how many queer AI journalists there are, but I have to believe I’m not the only one. I’m certainly not the only person writing about AI and queerness during Pride month.
Yet, I don’t think I could purchase a 30-40% share of the Google News search results if I wanted to. But here I sit: apparently the most popular queer AI journalist in the world if you go by Google’s algorithms.
That might sound cool, but it really sucks that my own work on queer issues in the artificial intelligence and STEM communities is being driven down in the search results by my unrelated work.
Worse, other journalists are obviously being slighted. I don’t like the way that feels. Slapping some pronouns in my bio and having the privilege of working for a company that enthusiastically supports my reporting on queer issues shouldn’t give me a lion’s share of the results for something as broadly reported on as the search phrase “artificial intelligence queer.”
The solution?
I’m still not sure exactly why “machine learning queer” and “artificial intelligence queer” give similar results. One of the stories I wrote about AI in the cannabis sector shows up if I search for “AI queer,” but it doesn’t show up if I search for “AI cannabis.”
It’s clear that my TNW author profile plays a large role in how Google News treats my work, but it’s also evident that the words “queer” and “pronouns” play a much larger role than any others when it comes to surfacing my work.
At the end of the day, none of this makes me feel good. We all want our work judged for its content. Having an algorithm apparently associate everything I do with queerness is weird and off-putting.
There’s more to me as a journalist than my queerness. And it does a disservice to both the queer and artificial intelligence communities when the algorithm surfaces my unrelated writings on, for example, quantum physics and Facebook when someone searches for “AI queer,” instead of important stories from reporters about LGBTQPIA+ issues in the STEM sectors.