Recently people refuted the accuracy of ChatGPT, the new AI machine that has destabilized many people. Now more than ever, we are beginning to critically examine the capability of AI to replace humans in every vocation because the threat is getting more real. While several case studies have been used to demonstrate that humans are irreplaceable, a simple image search for “happy white woman” is still making the rounds.
A Quora user asked why there are unexpected results when “Happy White Women” is Googled. The responses were as follows:
“Google has a strong anti-White agenda. People try to claim it is just the “algorithm.” An algorithm is not pure artificial intelligence. It is basically just a crawler, and humans with their own agenda input what to look for. Most search engines are actually powered by Bing (such as DuckDuckGo), which does not have this strong anti-White agenda.”
Tom Thomson
Tom further added;
“They are trying to push an agenda. They hate White people. It’s OK to be White”.
While I couldn’t outrightly disagree with Tom on this, I decided to find out for myself what to believe and to offer evidence-based ideas about why the search results are so.
Could Google be racist, promoting black population groups, or trying to facilitate inter-racial unity? I hopped into my Google search and typed in the keyword “Happy White Woman”. These results (shown below) were largely consistent with that assumption (I marked some of those with a green tick shown below).
But what do you think about this? Why does Google display unexpected results when the phrase “happy white woman” is searched for on the search engine?
If you have not already seen this before, open a new browser tab and type “happy white woman” in the Google search bar. After it opens, click on “Images” to display only image results for that phrase. These image results for “happy white woman” have become a topic of controversy for a while now. But we now examine the possibilities here.
On Google search, the result should look like this.
Image results for “happy white woman” on Google seem to be quite unexpected. We see several images of happy white women standing with or hugging black men. Is Google trying to tell us something we don’t know? Or does the “happiness” of “white women” depends on black men?
Maybe this is another AI glitch that might be resolved soon, or Google is intentionally promoting BLM propaganda. However, this observation has persisted for a while now, and all these claims are only based on assumptions. To find out exactly why the results are so, we need to try out several things.
First, I decided to search for the related phrase “happy black woman” to see if Google would provide mixed results as well. Here was what I found.
As you can see, the results were consistent with “happy black woman”. Google didn’t mix image results for black women as it did for “happy white women”. This sort of raises a different set of debates around why this is so.
Just to understand it more clearly, I compared the search results for the same keyword on two other search engines – Bing and DuckDuckGo.
Bing is Google’s closest search engine rival. The same search keyword on Bing Search gave the image result below.
DuckDuckGo is another search engine that claims full privacy for the searcher. That means it doesn’t track users to offer results influenced by user bias or location. An image search for “happy white woman” on DuckDuckGo gave the following result.
Now that we have almost everything we need to solve this case, can we begin to answer the question of why search results for “happy white woman” are quite inappropriate?
Here are some possible explanations for this anomaly;
- Search intent manipulation.
- Google may be biased, racist, and not to be trusted.
- AI algorithms, in general, may not be as intelligent as we think.
Search intent manipulation by BLM groups
Google is the foremost search engine used by almost everyone in the world today who has access to the internet. It has a large database of information organized by its search algorithm. Unfortunately, Google’s search algorithm is not fool’s proof.
When the question of “why these unexpected search results show up” was asked on Google Search Forum, a Platinum Product Expert on the Google Forum answered as follows;
Usually there are simple explanations such as the search engine has interpreted the word white to mean the background color, not the person’s color; but interpreted the word black to mean the person.
Regarding images, many of the results are from stock photo sites, and they categorize their content using themes (including sex, color, background color etc) that skew what you see in Google’s Image Search. Note that your question used the word “happy”, a concept often used as a category by stock photo sites, making it more likely the results will be full of stock photos.
Usually it’s about the language used, the specific wording of the query. Google indexes the wording of other websites, so its results tend to reflect societal prejudices, or just mirror the way words are commonly used in real life.
Bear in mind that some of the people sharing this question on social media, are not doing so with the best of intentions, and know full well it’s a trick of language. This question has repeatedly been asked by people who enjoy race baiting, but also by others who were tricked into being honestly outraged by a perceived racial bias.
One further thought: Google’s automated algorithms are not immune from the same prejudices as the rest of society.
This shows several things but it also demonstrates that special interest groups on the Radical Right could search for “Happy White Women” and repeatedly click on interracial pictures to prime Google search bias and use the search results as evidence against Google for BLM propaganda. From the explanation above, this could actually work on Google’s algorithm.
This proposition asserts that Google might be pushing BLM propaganda against destructive efforts by groups on the Radical Right, where it is programmed to display more racially friendly results when users search for keywords like “Happy White Woman” which may have been seen to have racial intents. And now, there is a counter-rise of race-baiters who now claim Google is racist in favor of blacks.
David Junge, after thorough testing, found that there is no racial-based search intent associated with “happy white woman”. And if Google considers “happy white women” a means of race-baiting, we’d expect other search engines like Bing, DuckDuckGo, and Yahoo to pick it up as well, which apparently isn’t the case.
Google, itself, may be secretly pro-BLM
This is of course the most interesting argument of the Radical Right. They claim that since this occurrence of mixed race results for a search for “Happy White Women” is seen only with Google and not with other Search engines, Google may be directly involved in BLM propaganda. They are not arguing that BLM groups have found a way to bias the search algorithm, but that Google is openly supporting BLM and minority groups.
However, this is very unlikely as Google would not intentionally alter its search results to promote any group. That would raise suspicion, queries, and litigations, and to think that Google would go ahead is extremely unlikely.
Here is another way to look at it. If Google is deliberately promoting BLM or minority groups, it should also reflect in other related search keywords. For example, a search for “Greatest Scientist” should show more individuals in the minority groups if indeed Google is promoting such groups but it clearly does not, as shown in the image below.
Bear in mind that another similar allegation had been raised against Google in 2019, claiming that Google’s algorithm discriminates against black women and girls. In the publication, it was noted that a search for “girl” or “women” shows only white girls and women respectively.
The inconsistencies in the claims against Google show that there is no factual evidence to prove that Google is biased and intentionally promoting BLM and minority groups.
AI algorithms may not be as intelligent as we think
As depicted by the Google Forum expert above, the Google search algorithm, and by extension, all search engine algorithms, are not immune to society’s prejudices. More importantly, search algorithms can misunderstand what is being searched for based on how the images are categorized.
Most images used by many websites are obtained from stock image sites which may categorize their images inappropriately. The alternate “alt” text in images also describes what an image should be, and this also can be wrongly written.
Google, like other search engines, relies on this information to pull out results from its large database. The original categorization of a “mixed-race white women” image in a “white women” image category could influence how Google sees those images.
In addition to the search intent that I talked about earlier, results may reflect the signal received by Google as to what is the real reason behind the search, and hence display results based on what it perceives that people want to see more.
Lastly, Google’s AI algorithm may misinterpret the word “white” to mean background color and display happy women on a white background.
But whatever the case is, we now see that Google and other search algorithms may not be as intelligent as we thought them to be.
My thoughts on this matter
Most global brands use their influence to promote inter-racial unity, speak against political violence, child and women abuse, etc. Google is a global brand that might not want to be associated with any sociocultural violations like racism and others. This might lend credibility to the claim that Google could tweak its algorithm to be more socially inclusive, by gathering search data and understanding the intents for which searches are made relating to a particular topic.
Searching for “happy white woman” after hearing about the conspiracy that the results are showing some unexpected results, already influences your search intent – you want to confirm or refute the claims. This, in some ways, affects your interactions with the search results, a behavior Google can detect and adapt to modify future results about the search keyword.
Final words
Google search for ” Happy white woman” shows results with happy white women and a few black men. This behavior is not seen using other search engines like Bing and DuckDuckGo. This has raised controversy on the internet concerning how these search behavior on Google may have a racial or BLM undertone.
Evidence examined so far has not shown any such bias. The most probable explanation is how search intents affect search results, and the role of image description in stock photo sites can determine what shows up for a given keyword phrase like “Happy white woman”.