The Publisher Lab: Core update misconceptions, Bing gaining on Google’s search lead, & Elon Musk urges an AI pause

We’re back with another episode of the Publisher Lab!

This week, Tyler Bishop and Whitney Wright discuss the completion of the Google March 2023 broad core update and misconceptions about algorithm updates, Bing gaining on Google’s search lead, and Elon Musk (along with 1,000 others) calling for a pause on the development of AI.

If you would prefer to watch this episode, head to our YouTube channel. You can also listen to it wherever you listen to podcasts.

Also, make sure if you have comments or suggestions to leave us something on YouTube or use the suggestion box at the top of the Publisher Lab’s homepage.

The Google March 2023 broad core update has finished rolling out

The Google March 2023 broad core update completed its rollout on March 28 after kicking off on March 15. The biggest volatility was on March 16, the day after it started, and then again on March 23-24.

As far as volatility goes compared to other algorithm updates, it was a little more volatile than the September 2022 update (the last update Google rolled out) but ultimately was the same. When looking at the top 3, 5, and 10 positions, data showed that it had lower rank volatility than the September 2022 update in the top 5 positions.

What SEO experts have gathered from what they’ve seen online and from talking to Google, the update:

• looked at all kinds of content
• was not a penalty-based update, but rewards great web pages
• is a global update affecting all regions and languages
• seems to be a typical update, reaching wide and impacting fast
• affects Google Discover, feature snippets, and more

What we typically see with core updates is that publishers who see traffic fluctuations around an update typically attribute it to being “hit” by the update, but this is often not the case. If most publishers look at their data from a broader point of view, rather than honing in on data surrounding the update, they would see that there is probably not that much out of the ordinary. Or, if there was some fluctuation in traffic (though it may have nothing to do with the update) during that time period, if it returns to “normal,” one might assume it’s because the update was rolled back.

In fact, after pulling data, we found that there is only one core update in that time period more publishers didn’t actually get more traffic from Google (on average). It’s in Google’s best interest to distribute traffic amongst as many sites as possible because it makes it so no one has significant negotiation power.

If you do want to look at your data around the update, we suggest looking at Google Search Console, not Google Analytics, look at “pages,” select “clicks,” then select whatever time period you would like to look at and compare it against another time period. It will show you changes amongst your whole site or certain keywords; you can look at changes in ranking spot, the number of clicks, and the number of times it showed up in search results.

Bing gaining on Google’s search lead

Recent data by SimilarWeb shows that OpenAI integration into Microsoft Bing was followed by an increase in traffic and pageviews by 15.8% for Bing and a nearly 1% drop for Google. It is expected that Microsoft has the opportunity to gain over $120 billion in the search market with its gain in popularity.

Additionally, Data.ai shows there was an 8x increase in global Bing downloads after its chatbot integration while Google search app downloads decreased by 2%. Bing has, historically, had less than a tenth of Google’s market share, and so even 1-2% of a gain is a big deal.

Comscore shows that when we look at social media mentions, “Bing” and “search” on Facebook, Twitter, Instagram, and TikTok are up 48% YoY while “Google” and “search” on those same platforms have decreased 71% MoM when compared last year. Additionally, prior to the day Bing went live (February 7), it held just 1.22% of organic search traffic; as of March 23, it now holds 2.6%, while Google declined 1.5%

When we consider the journey Microsoft has been on through the years, these gains are significant. For example, Microsoft Office Suite used to be the dominating workspace and pretty much every computer had the software. However, then you had Google Suite come in and build just as good of a product, and now you are basically either a Microsoft Suite company or Google Suite company.

Now, it is likely Microsoft feels a little vindicated, and Microsoft has a real opportunity here to develop something that serves a very different, but important, function than search, and really own the market.

There is conclusive evidence to show that Microsoft Chat-GPT is currently better than Google Bard, and what is a major differentiator is how Chat-GPT’s natural language is able to communicate with humans.

What’s maybe even more devastating for Google is that it had 10 years of natural language processing (NLP) under its belt with LaMDA and it took its sweet time with creating something like this with it. It’s likely that Google thought it had all of the time in the world to get it “just right” and that others would wait and see what Google did before acting.

Many companies tried something similar with their voice assistants—Siri, Cortana, Alexa, etc.—but these tools haven’t really improved that much over the last 5-6 years, so it’s incredible that in three months, we have gotten these strides in chat AI.

It would be in AI’s, and people’s, best interest if Google did not come back swinging with Bard and become the dominant figure in AI as it has in so many other realms; the best-case scenario would be that there would be hundreds of other models, engines, and platforms that provide all sorts of AI; the more powerful, yet specialized, AI becomes, the better products that will be developed, rather than one company trying to cover all bases.

If we look back on how Big Tech has operated in the past, however, this likely won’t happen; these companies tend to see what the other companies are doing and try to replicate it rather than carving their own path. This typically just leads to “under-achieving together” rather than innovation in any specific direction but it is likely that these Big Tech companies see it as too much of a risk to branch out and try to be really good at something no one else is doing rather than following the crowd.

Elon Musk and 1,000 others urge AI development to pause

According to Reuters, Elon Musk and a group of artificial intelligence and industry experts are calling for a six-month pause in developing systems more powerful than OpenAI’s most recent rollout of GPT-4 due to the risk to society and humanity it holds. The letter, which was signed by more than 1,000 people, was issued by the non-profit Future of Life Institute, which is primarily funded by the Musk Foundation.

The letter says the pause is necessary in order to develop and implement shared safety protocols, and cites that the risks include economic and political disruptions.

On Monday, the EU police force, Europol, specifically pointed out the potential misuse of AI for phishing attempts, disinformation, and cybercrime, while the UK government recently announced proposals for an adaptable regulatory framework regarding AI.

Critics of the letter say it is feeding into “AI hype” and that the harm that AI can cause has been greatly exaggerated and that instead of pausing research, AI researchers should have greater transparency requirements.

However, what people really want is not necessarily possible; they want to know why a machine made a certain choice but it’s not like a machine’s choices and outputs are kept in a tidy spreadsheet—its choices are based on 0’s and 1’s and it makes choices based on the path of least resistance that has been developed over time (much like how the human brain works and develops neural pathways). More pointedly, there are no receipts that spell out exactly why a machine made a choice much like we can’t look at a human brain and determine why someone might have committed a crime (that’s what the justice system is for). This especially makes sense when we consider that we are literally trying to build a machine that mimics human behavior and processes.

The likelihood of some sort of “pause” on AI development though is small because AI has already infiltrated so much of our lives at this point and trying to “put the cat back in the bag” (as Tyler suggests) is impossible.

The one concern that is worth addressing in this regard, however, is cybersecurity. It is likely we are only months away from someone being able to deep fake their voice and visual likeness, which could have potentially devastating consequences, even just for the average person (consider if a voice that sounds exactly like your spouse calls you and asks for your social security number, or the ability to visually replicate IDs). What is in place right now as security measures are not going to be able to hold up against AI that can process and run all sorts of mathematical tests to find an answer or way around.

There is no singular solution that is going to work across all industries; instead, each industry is going to have to create its own processes and safeguards to protect against fraud and security issues.

Publishers can use AI to their benefit by beginning to add options at logins or user data captures to make it easier for users to join in simply with one click. As these sorts of login measures grow, and one becomes the more dominant figure, it has the ability to endear your audience to you because you are making things easier on them. It will increase your odds of a conversion.

A rapidly-evolving space

Digital publishing is changing very quickly with the implementation of AI and there is no end to the use cases of how you can make your life easier and improve your digital publishing business. There are extensions for Chat-GPT where you can basically add it to almost any application.

Have something to say? Visit us on PublisherLab.org and drop us a line!