Judgment Call

[Judgment Call] Google’s SGE – and why it matters to you 

Gemma B. Mendoza

This is AI generated summarization, which may have errors. For context, always refer to the full article.

[Judgment Call] Google’s SGE – and why it matters to you 
Dear readers, your favorite news websites could literally die

How many “T’s” are there in the word Rappler? A friend used this question a few days ago to explore AI Overview, Google’s new feature powered by Search Generative Experience (SGE). This was after Google announced that it was rolling out the feature to all US users.

The AI’s answer? Rappler had one letter T.  

Obviously Rappler does not have a letter T. I opened the Google app and tried it myself. When I asked how many T’s (without the quotation marks) there are in the word Rappler, Google gave a safe answer (see right image below). But when I put the T’s within quotation marks, AI Overview repeated its error, that Rappler had one letter T (see left image below).

The machine’s two responses indicate that Google’s team of engineers appears to have patched whatever bug was causing the erroneous response. I tried running the same question again today, as I was writing this, and – with or without the quotation marks – got the results that just linked to pages in Rappler. 

This, however, does not mean that AI Overview will not get it wrong again. 

But it also means it does get things right, either. And in a strange way, that is an even bigger concern to many publishers.

This is why I am writing to you about this. Hello, I’m Gemma Mendoza, Rappler’s head of disinformation and platform research. 

Our CEO, Maria Ressa, has been talking about the bleak prognosis for the future of news given trends in Big Tech.  

The roll out of Google’s SGE poses an almost existential threat to newsrooms. 

And we need your help to address this threat. 

Packaged by Google as a helpful service to avid Google users, SGE does not immediately serve links to articles from authoritative news sources the way Google was originally designed. Instead, it tries to answer questions on its own. 

The UI purposely hides the source of each piece of information. 

For instance, if you ask Google for the age of embattled Bamban, Tarlac, Mayor Alice Guo, Google responds up top that she is “about 38 years old.” The generated response does not immediately include any link to the information source. 

You have to go further down the search results page to the part where the links to news sites are located. In this case, the articles that came out (which include those by Rappler and Vera Files) mention that she is roughly 37 years old. 

Granted that there are inconsistencies in Alice Guo’s declarations, how did Google come up with its own number? 

This is a problem not just for media and journalists whose research could have been used to generate the AI’s responses without attribution. 

Further, like other platforms, it prefers to hide behind the disclaimer that “Generative AI is experimental,” rather than take responsibility for content they produce. This makes it also a problem for the general information consumers who may have been served wrong information. 

Beyond the risk of circulating false information, SGE prevents users from accessing the original information sources by introducing friction. To get to the links to the original news and content sources, users have to take additional steps, such as clicking on the pulldown button or scrolling down the page. 

This challenge also makes SGE a threat to the sustainability of newsrooms.

To illustrate the dynamic between news websites and Google’s search engine, I tried running this question: “What is Mayon Volcano’s most destructive eruption?” on the current Google search engine results pages (SERPs). The following results came up in sequence: (1) a Wikipedia entry, (2) a list of related questions under “People also ask,” (3) websites which have articles on the topic. 

Since we published it in 2015, Rappler’s feature story entitled, Looking back: Mayon Volcano’s most destructive eruption usually captured this spot. On a mobile phone, the story immediately becomes immediately visible on the second frame when you scroll down, if not on the first frame. 

If there is a developing news coverage related to a search query, the most recent and most relevant articles from news websites about the subject matter usually takes the top-most part of the search results pages. 

Because of the way its SERPs were structured, organic search results from Google typically deliver a substantial chunk of the traffic of news websites. More traffic results in more revenue – essential to keep the operations going.

For the ordinary users, Google’s original algorithm, which put a premium on ​​expertise, authoritativeness, and trustworthiness, also meant they get the best content with every query.  

All of this could go away with the full roll out of SGE. 

Note that this feature is not yet fully available to Philippine-based users. I have been testing it using Google App, where it still bears the logo of Google Labs. 

The result? Even when I used the exact same query as the earlier one (“What is Mayon Volcano’s most destructive eruption?”), I was only able to get to the related Rappler story on Mayon after scrolling down five times on my mobile. (See screenshots below)

Think: If a user already saw what he or she was looking for in the first frame, why would the user scroll down to where the links to source websites are, much less click on those links?

AI Overview does provide a way to check the original source of the information it generated. But this is hidden under hardly noticeable pulldown buttons. Any web designer worth his salt knows that every additional click introduces friction that prevents users from getting to another site.

SGE, in effect, changes the relationship between news publishers and Google. From primarily being a curator and distributor of content, Google becomes almost like a competitor that ingests content it scraped from websites, and uses it to train its large language model which then generates and “publishes” its own personalized derivative content. 

Of course, Google is not the only one doing this. A number of newsrooms, for instance, have already sued OpenAI for copyright infringement. In a statement, the company behind popular AI chatbot ChatGPT, argued that using publicly available internet materials for training is “fair use.” 

But clearly a lot more is at stake in Google’s case. 

Because its powerful search function has been driving traffic to news sites, publishers, for decades now, have been voluntarily submitting their sitemaps to Google for its crawlers to index and curate. 

The rollout of SGE puts publishers in a bind: allow their sites to be indexed to get search traffic, or disallow the Google crawler and risk getting de-indexed? In the Philippines, for instance, SGE could in effect put at risk around 50% of existing traffic to news websites that are coming from Google search results pages.  

To begin with, news websites have been losing not just traffic but revenue due to fickle Big Tech algorithms for years. Most recently, Facebook, still the top social media platform in the Philippines, has deprioritized news sites on its algorithms. The result: total pageviews of news websites in the Philippines have been consistently compared poorly with the previous year’s numbers, according to data web analytics platform SimilarWeb.   

From being the top source of traffic, social media now accounts for only around 13 % of website traffic of Philippine news and media sites for the past 12 months. 

Things could get worse with the full SGE rollout.

Unless Google changes its plans, the company’s current SGE rollout plans could be disastrous to news sites already suffering from traffic losses due to changes in social media algorithms. 

This should be a concern for ordinary news consumers because further losses in traffic could affect revenue and, ultimately, newsroom sustainability. In short, dear readers, your favorite news websites could literally die. 

Rappler has been studying these trends over the past months and have been working on solutions that could help Rappler and other news organizations survive. This is why we decided to roll out Rappler Communities – a news app and community hub rolled into one. 

Besides the goal of creating a digital town square, a shared space for conversations centered on facts and free from Big Tech’s insidiously manipulative algorithms, Rappler Communities is a lifeline. It is a way for newsroom to stay connected with you, our audience, our community. 

Big Tech has already effectively given up on news. But we believe you have not.    

Help us keep independent news alive. 

Please download the Rappler App and ask your friends and family to do the same. Rappler.com

Add a comment

Sort by

There are no comments yet. Add your comment to start the conversation.

Summarize this article with AI

How does this make you feel?

Download the Rappler App!
Person, Human, Sleeve


Gemma B. Mendoza

Gemma Mendoza leads Rappler’s multi-pronged efforts to address disinformation in digital media, harnessing big data research, fact-checking, and community workshops. As one of Rappler's pioneers who launched its Facebook page Move.PH in 2011, Gemma initiated strategic projects that connect journalism and data with citizen action, particularly in relation to elections, disasters, and other social concerns.