Wikipedia

The Future of Wikipedia in the Age of AI

As the use of AI models increases, the way users seek information is evolving. Queries are becoming more complex and conversational, and results are typically based on a much larger body of data, rather than a specific source or page. 

As these models become increasingly integrated into our daily lives, the importance of Wikipedia in shaping brand reputation cannot be overstated, since it is a major source for training AIs.

Importance of Wikipedia in AI Training
According to The New York Times, “Wikipedia is probably the most important single source in the training of AI models.” The platform’s vast trove of crowdsourced knowledge, covering a wide range of topics, provides invaluable data for AI models to learn from. Without access to this information, the development of current generative AI capabilities might not have even been possible. (Here’s some additional information on how AIs/LLMs/Chatbots are trained.) 

Impact on Brand Reputation
With AI models like ChatGPT, Claude AI, and Gemini having been trained on Wikipedia, inaccurate or biased information on the site can lead to negative or incorrect information about a brand, potentially harming its reputation. With so much riding on the underlying information in Wikipedia, ensuring the positivity and accuracy of a brand’s Wikipedia presence has become more important than ever.

Recommendations
Given Wikipedia’s elevated status, our recommendations for companies, brands, and individuals are to work within the Wikipedia guidelines to do the following:

1. Maintain: Create and/or maintain a well-structured, robust Wikipedia page for your brand or personal profile.
2. Update Accurately: Make sure the page remains updated and accurate with current facts, figures, and noteworthy achievements.
3. Include more sources: Since LLMs utilize all of the content, include as many relevant, verifiable sources, as appropriate – these should only help the AI training.
4. Go Multilingual: Consider developing a presence across multiple language editions of Wikipedia. LLMs often learn from content in various languages, and the more you play an active role, the better. Also, consider that English is often the hardest language version of Wikipedia to impact, and other language versions can be very easy to edit.
5. Other Wiki pages: LLMs can learn about your brand and industry from any Wikipedia article, so consider getting relevant information added to relevant industry articles, not just the ones about your brand.
6. Talk Pages: Leverage Wikipedia’s “Talk” pages to include additional relevant information, as LLMs may also use these for training.
7. Images: Consider submitting relevant images via Wikimedia Commons to enhance your Wikipedia page and improve AI model understanding.
8. Categorize: Utilize Wikipedia’s category system to ensure your page is properly categorized and connected to the ideal topics.
9. Monitor: Monitor your Wikipedia presence for edits that may introduce inaccuracies, outdated information, or bias; address issues appropriately and promptly. Do the same for other relevant pages related to your company or brand. Our free WikiAlerts service provides tracking of Wikipedia and Talk pages.
10. Wikidata: Beyond Wikipedia, leverage Wikidata, Wikipedia’s sister project, a powerful database of community-contributed structured data that LLMs will increasingly use to verify facts.

Do We Even Need Wikipedia in a World of AI?
An interesting question that has been raised recently is whether there is even a need for Wikipedia. Since the content is taken from various third-party sources, and the LLMs presumably have access to the sources and probably many more, why can’t an AI produce Wikipedia content that would be as good or better than content created by Wikipedia editors?

To answer this question there have been various attempts to utilize AI to write sections of Wikipedia pages, but so far, despite the great capabilities of AI, they have not been proven to produce content that is up to par. It is possible that this will change at some time in the future, but for now there still seems to be tremendous benefit derived from the human (crowdsourced) process that helps create a Wikipedia page. Perhaps AIs that are trained on this process will eventually produce content that is recognized to be of high enough quality. 

Conclusion
Ongoing tracking of how AI models represent your brand, and what role Wikipedia may be playing, can help you identify areas for improvement within Wikipedia and beyond. 

As the use of Wikipedia in AI training continues to grow, we believe that the future of brand reputation management will be even more closely tied to Wikipedia. By actively managing their Wikipedia presence, companies can ensure that AI models have access to an important trusted source of accurate and up-to-date information, ultimately leading to a more positive online reputation.

Five Blocks specializes in digital reputation management for platforms including Google and Wikipedia, combining cutting-edge technology and personalized service to help our clients overcome digital reputation challenges. Our advanced data analysis and AI-powered insights allow us to identify the root causes of digital reputation issues and uncover overlooked opportunities for improvement. We work closely with your communications team to develop and implement sustainable solutions that deliver long-lasting results.

For more information or to see what we can do with your brand’s data contact us.

Why Your Wikipedia Edits Keep Getting Reverted

If you have made the decision to get involved in Wikipedia, you may find the editorial environment confusing, overwhelming, and sometimes even hostile. Even though Wikipedia encourages editors to be bold, there can be real tension when changes are perceived as out of line, controversial, or otherwise problematic.  Changes made by novice editors are often reverted. 

Please note that editing a Wikipedia article about yourself or your company is prohibited without disclosing conflict of interest.

Five Blocks offers Free Consultation and help in determining your options regarding Creation or Editing of Wikipedia pages. Contact Us.

The following recommendations are for objective, volunteer editors who are not being paid or swayed to edit in any way. A basic understanding of these common editing snags can streamline the process of contributing to Wikipedia, and make it easier for you to expand the encyclopedia’s content in constructive ways. 

1. You’re not signed in — Wikipedia has two primary editing channels: registered and unregistered (or IP) editing. If you don’t create an account, your changes will be attributed to the IP address of the computer you are using to make the edits. For all sorts of reasons, Wikipedia editors prefer “interacting” with registered editors as opposed to IP addresses. Extensive editing done without logging in is often viewed suspiciously. The editing community is less trusting of IP edits and is, therefore, more likely to revert sizable edits made in this way. 
2. You didn’t include citations — Wikipedia wants readers to know where the information is coming from. Assertions made in a Wikipedia article should be backed up by independent, reliable sources. Reliable sources are defined as those with a good reputation and a solid editorial process. Blogs, tabloid journalism, and sponsored content are not to be used. Peer-reviewed industry journals can be used where appropriate. This policy is scrupulously enforced. 
3. You’re citing original research and primary sources — As an extension of the previous item, information can only be included in Wikipedia if it has been published in reliable sources. Research that has not gone through the peer-review process of academic and scholarly journals should not be cited in the encyclopedia. Articles and publications written by the subject of the Wikipedia article should not be used in most cases. Information published by the subject’s employer is also considered “primary sourcing” and could flag your edits for reversion.
4. Your subject isn’t notable — Not every person, company, or concept is considered worthy of a Wikipedia article. Even though someone may seem very important to you, it does not necessarily mean they meet Wikipedia’s standards of notability. We wrote about this here and here. 
5. You’re repeating edits — Known as edit warring, the re-insertion of content that has already been removed is unacceptable on Wikipedia. If an edit you made is reverted, look at the reasons given. An article’s talk page is the right place to start a discussion about the content you’d like to add, particularly if you are finding that it keeps getting reverted. 
6. You are inserting radical, extreme, and/or fringe theories — Wikipedia strives to present a neutral point of view and a balanced demonstration of the facts. It is not a platform for inclusion of every theory, hypothesis, or premise. While these notions might be mentioned in a particular article, they will not be given equal weight compared to mainstream views. Wikipedia articles will always strive to reflect the scientific consensus. 
7. You seem too close to the subject — If you have a financial connection to the subject of an article, Wikipedia requires that you declare your conflict of interest. For the most part, editors with a conflict of interest are encouraged to avoid editing a page directly, and instead to use the talk page to suggest changes and raise issues.
8. Your edits cause harm- Your edit might be reverted if it is considered disruptive or malicious, changes that Wikipedia considers vandalism. Repeated sabotage of a page will either get your account blocked from editing, the page protected (meaning it can only be edited by a small group of Wikipedia editors), or both.

As Wikipedia continues to maintain a powerful position in the current landscape of accessible information and data consumption, it needs editors—just like you—to work diligently, effectively, and efficiently to ensure the quality and neutrality of the articles. From hobbyists to subject experts, Wikipedia needs and wants people like you. Wikipedia thrives because of its community of editors—the volunteers who are committed to fixing mistakes, adding stellar content, and improving the platform for readers. 

Do you need help editing your company or personal Wikipedia page?

Contact Us for a free consultation.

FAQ

Why did my Wikipedia edit get removed?

Any changes that are unsourced; supported by unreliable sources; malicious; biased; or considered harmful in any way, will be reverted. Edits made by editors appearing to have a conflict of interest will also be flagged or deleted.

Do I need to be an editor to edit Wikipedia?

Anyone can edit Wikipedia, even without a registered account. The Wikipedia community prefers to “interface” with registered editors. If you choose to edit without signing in, your changes will be attributed to the IP address of the computer you are using.

How can I change information on my own Wikipedia entry?

The goal of the Wikipedia project is to be an encyclopedic source of objective information. Articles that are promotional, self-serving, or bias are at-risk of being tagged or deleted. These are our recommendations for making changes on your own page.

How do I know who edited a Wikipedia page?   

All changes to a Wikipedia entry are logged. You can see this log by going to the ‘View History’ tab at the top right of the Wikipedia article. The date, time, and size of each change is listed next to the name of the editor who made it. Clicking on the editor’s name will take you to their ‘user page’ where you can read more about the editor.

Does everything on Wikipedia need to be sourced?

Information included on Wikipedia needs to be backed by independent, reliable sources. If you want to add content, it needs to have been picked up by media outlets with a good reputation and an editorial process. 

 Who can have a Wikipedia article?

Wikipedia has standards of notability which determine a subject’s eligibility for inclusion. Read this article to learn more about the criteria for getting a Wikipedia article.

Wikipedia: True or False?

Wikipedia likes to assert that all information included in an article is put up by voluntary, neutral editors who use sufficient secondary sources to back up statements. There is vigilant enforcement. Still, many thousands of people ask Google each day: How reliable is Wikipedia? Is the information on Wikipedia true?  

The short answer is: Yes! The citations (little numbers/links included in the articles) lead to quality secondary sources. Presumably, the more references an article has, the more truthful the article. 

Wikipedia ranks prominently because it has characteristics of an authority site (like Bloomberg or the New York Times) and also has characteristics of a crowd-sourced resource. It has the advantage of having real people fact-check and contribute, and when that process goes well, the result can be more authoritative to users even than a recognized news publication. Unlike major media, Wikipedia is meant to be safe from political leanings or editorial positions. 

But, the long answer is: It’s complicated. 

In February 2020, South African fast bowler Dale Steyn turned to Twitter for help when he discovered that his Wikipedia entry falsely stated he played for Zimbabwe.

His followers made a variety of suggestions/offers to “fix the problem” themselves. What most of them failed to suggest in their tweets, however, was the need for reliable sources to back up the claim that this sportsman did in fact play for South Africa and not Zimbabwe. Some of these well-intentioned advisors did indicate that the edit couldn’t be made by Steyn himself. Read more on that here.

True Tales of Wiki Woes

In January 2019, British actress Olivia Colman shared her story of trying to get her birthday corrected on Wikipedia. She was dismayed that the wiki-indicated day, month, and year were all wrong (making her eight years older than she really was.) All she wanted to do was correct wrong information. But her attempts to get the right date on her Wikipedia entry were met with retorts from wiki editors asking her to “prove it.” Colman pointed out that no citations were used to assert the false birthday and thus her avowal of the correct date should be enough. The Wikipedians ultimately reached a consensus and her actual birthday was included in the entry.

In 2012, award winning author Phillip Roth found factual errors in an entry about his own book, The Human Stain. He—the author himself—was pressed to attest to the veracity of these inaccuracies in the form of secondary sources. So Roth, being an esteemed writer, created the source himself. He wrote an article about his book, and his Wikipedia experience, and had it published in The New Yorker. Then it was legitimately used by an editor as the cited source to correct the misinformation.

On March 14, 2007, Sinbad’s Wikipedia page claimed he had died. Major news outlets reported the death; his manager received condolence calls; Sinbad himself was informed of his own passing when his daughter called his cell phone. Even after several interviews with leading radio and television stations in which he asserted himself as “alive and well,” Wikipedia insisted that the edit could only be reverted if it could be confirmed by a link-able secondary source. Eventually, the editors recognized the vandalism and reverted the falsehoods, but not before it became the subject of some of Sinbad’s stand-up routine.

There are countless more examples, but you get the idea…

The Fight for “Truthiness”

In May 1897, Mark Twain told a journalist, “the reports of my death are greatly exaggerated.” It is probable that had he encountered this problem sometime after 2001, when Wikipedia came to be, he too would have had to find reliable secondary sources by which to prove his own well-being. 

Wikipedia’s verification policies are designed to ensure that all the information found on the open-source, cooperative platform is in fact true. By relying on secondary sources, Wikipedia presumes that sufficient research has been done to achieve a relatively high level of accuracy. Wikipedia even maintains a list of approved and deprecated sources designed to ensure the veracity of reported information. 

But these same policies also create a loophole through which some falsehoods can trickle in, if these distortions were once reported as truths (as in the case of Roth and Sinbad). Similarly, the loophole allows for extra savvy Wikipedia consumers, like Phillip Roth, to produce a source designed specifically to counter the misrepresentation and set the record straight.

With sites in 300 different languages, over 40 million articles, and more than 200,000 editors/contributors, accessed by more than 1.4 billion unique devices (mobile and desktop) each month, there’s a lot of information out there in the Wiki-verse. 

Some of that data we all know to be true, yet the “truthiness” of certain information seems somehow more reliable when we read it on Wikipedia. The term itself — coined by Stephen Colbert, and awarded “Word of the Year” in 2005 and 2006 — has its own Wikipedia entry, which has to be enough to make it an officially recognized (and acceptable) word in the English language.

All in all, while  the Wikipedia system certainly has flaws, it is probably the best one there is, and we have found that it is better to work with the system than against it.

The Wikipedia Index: Edits to the Fortune 500

Wikipedia editors worldwide make 1 million edits every week. How many of these are edits to pages of major companies? Does it vary by industry? We found out for you:

There’s no doubt about it: Monitoring changes to corporate or executive Wikipedia pages is an essential component of reputation management.

Your Own Wikipedia Page: What Does it Take?

On October 2, 2018, in the pre-dawn hours in Ontario, Donna Strickland received a phone call that would change her life. It was Göran Hansson, secretary general of the Royal Swedish Academy of Sciences, calling from Stockholm to inform her that she had won the 2018 Nobel Prize in Physics. 

About an hour later, Strickland’s big win was announced to the rest of the world, and the fanfare that ensued went far beyond the usual media blip: Strickland, an associate professor of physics at the University of Waterloo, was only the third woman to win the Nobel Prize in Physics in the award’s history, and the first since 1963.

Along with the plaudits came voices of criticism for an award process that some believe has historically neglected women, and for a society that many feel does not do enough to foster the success of women in the sciences. 

But a fascinating secondary story quickly emerged on both social and traditional media, in what The Atlantic called “a Metaphor for the Nobel Prize’s Record With Women.” As it turned out, Donna Strickland did not have her own Wikipedia article at the time of her award’s announcement, despite her many achievements in her field, and her stature as president emeritus of a leading academic society.

Even more gallingly, a Wikipedia editor had drafted an article about Strickland on March 28, 2018 – several months before she won the Nobel Prize – only for the draft submission to be summarily rejected by another Wikipedia editor. 

To be sure, once word got out that Strickland was a new Nobel laureate, a Wikipedia article about her was created in a flash, and soon reached a respectable length and level of detail. But media outlets seized on Wikipedia’s initial exclusion of Strickland – despite her many accomplishments – until she won the Nobel Prize in Physics. 

The headlines were quick to condemn the website: “Wikipedia rejected an entry on a Nobel Prize winner because she wasn’t famous enough”; “The Nobel prize winning scientist who wasn’t famous enough for Wikipedia”; and “Female Nobel prize winner deemed not important enough for Wikipedia entry.”

But just how accurate were these accusations? Was Donna Strickland truly “not important enough for Wikipedia”? The answer is … not exactly.

To get to the heart of the matter, let’s put aside the thorny topic of gender and systemic biases on Wikipedia and elsewhere, and ask a more basic question: What does it take to get on Wikipedia? Just how famous does a person have to be to deserve an article on the world’s most widely read encyclopedia?

One of Wikipedia’s foundational principles is that Wikipedia is an encyclopedia, not an indiscriminate collection of information, so there needs to be some standard for a person (or any other topic, for that matter) to be judged suitable for inclusion. 

The standard that the Wikipedia community implemented as the primary metric for inclusion is actually neither fame nor importance. It is notability. According to Wikipedia’s official notability policy, notable topics are “those that have gained sufficiently significant attention by the world at large and over a period of time.” Fame and importance (not to mention other factors like popularity, talent, and prestige) are deemed too subjective to meet Wikipedia’s inclusion criteria. Notability is where it’s at.

Part of why Wikipedia chose notability as the linchpin of its operation is that it can be demonstrated using references. Most Wikipedia articles must justify their inclusion by meeting the General Notability Guideline, or GNG. Sometimes called Wikipedia’s Golden Rule, the GNG states: “Articles generally require significant coverage in reliable sources that are independent of the topic.” Each part of this sentence represents a critical element of the GNG: “significant coverage” excludes trivial mentions; “reliable sources” excludes biased content and publications with poor editorial oversight; and “independent” excludes content produced by the article’s subject, such as a press release or an autobiography.

Despite Donna Strickland’s importance and stature in her field before she won the Nobel Prize, she clearly did not pass the GNG – there was simply no significant coverage of her or her work in news media or elsewhere! This is unfortunately more typical of important people from previous generations who simply did not appear sufficiently in sources that live online.

Critically, there are also several subject-specific notability guidelines, which serve as a kind of back door to grant inclusion to subjects that may not pass the GNG. For example, the notability guideline for academics dictates that an academic is notable if the person “is or has been an elected member of a highly selective and prestigious scholarly society or association.” In hindsight, many in the Wikipedia community argued that Donna Strickland should not have been rejected by Wikipedia even before winning the Nobel, despite the dearth of media coverage about her at the time, since she is a past president of The Optical Society. 

But the fact remains that by Wikipedia’s formal standards – which value notability and media coverage over more subjective factors – Strickland’s case for inclusion was, at best, contentious. As Wikimedia Foundation President Katherine Maher wrote in the aftermath of the Strickland affair, “If journalists, book publishers, scientific researchers, curators, academics, grant-makers and prize-awarding committees don’t recognize the work of women, Wikipedia’s editors have little foundation on which to build.”  

As we go forward into the future, our generation’s approach to appreciating women’s achievements will require some serious edits.

Wikipedia Page Visits – Have You Measured Your Mobile Visits Lately?

A good way to gauge the popularity of a topic is to look at traffic to that topic’s Wikipedia page. Traffic to the page is a quantifiable measure that is easily understood and easily compared across topics.

This is particularly interesting for a company or famous person. When a CEO says or does something notable or a company experiences a crisis, Wikipedia article traffic reveals a spike in interest.

Each Wikipedia article includes a link to view these traffic statistics, although the link itself is somewhat difficult to find. For this reason, we at Five Blocks bring this valuable data to the forefront and share it with our clients. We include it in our reporting and we feature a graph of this data on our automated monitoring tool.

On January 21, 2016, the Wikipedia traffic statistics page stopped updating. It turned out this was the beginning of a migration to a new Wiki traffic measurement site. By the first week in February, the reporting was back from a new source. It improved on the old source, by allowing for viewing statistics for flexible date ranges. It even allowed a user to slice the data by traffic source (desktop vs. mobile). But that’s when we realized something was off. The traffic numbers from this tool were quite a bit higher than they were on the old one.

A bit of checking revealed the answer: the old tool was only reporting article visits from people using a computer, while the new tool included computer users and those on mobile devices. We were happy to learn that the reporting from the new data source was more comprehensive. But this had significant implications for our reporting until that point. Reporting from the old source actually understated the actual number of visitors to the wiki article.

For example, Mark Zuckerberg’s traffic as reported on the old tool would look like this:

This shows average daily traffic around 7700 visits with a peak on Feb. 7.

But the more complete picture includes mobile traffic as well. It looks like this:

Average daily visits are actually 17,300. And the spike on Feb. 7 is now quite pronounced. (That’s when Zuckerberg wished the world a Happy Lunar New Year in Mandarin and announced his baby daughter’s Chinese name.)

Ignoring the mobile traffic here omits over half the visits to the page, totally under-representing actual interest in Mark Zuckerberg. Note a similar pattern with Royal Caribbean, which experienced its own crisis on Feb. 9 with news of its ship getting damaged in rough seas.

With the new data source only retroactive to August 2015, any historical data (from the old source) will require a big asterisk to remind us that we are not seeing the full picture. While the old data will still be useful to search for trends, patterns, and spikes in Wiki article views, it comes with the important caveat that the numbers themselves fall short of actual traffic. Certainly, new data should not be compared with old data given the different basis for measurement.

Obviously, the proportion of mobile traffic to a Wiki article will vary depending on how relevant that topic is to a mobile audience. But this point is clear: mobile traffic to wiki is significant. And omitting it means a significant under-representation of the popular interest in a topic, particularly during a reputation crisis.