Google AI Overviews (AIO) is an AI-powered generative feature in Google Search that provides summaries of search results rather than the traditional list of sources. While the jury is still out on whether Google Previews are a good or bad thing for the technology industrypublishers and SEO practitioners call it a double-edged sword that could end up drawing everyone’s blood, including Google.
“Google’s strategy may seem very current: take advantage of AI, keep users on their site longer, and maintain their iron grip on the search market,” says Lars Nyman, CMO of CUDO Compute and former Googler. “The question is, what are the implications for search revenue and downstream user behavior? If the impact is negative, then the insights will be a thing of the past.”
Clearly, Google is trying to stay relevant in an era where search competitors like startups Perplexity are having lunch. But it’s harder to do when you’ve established revenue streams to protect and a never-ending appetite for AI to feed.
“Google itself may need to help support certain news sources, perhaps paying for the use of those sources,” says HP Newquist. author and executive director of The Relay groupconsultant and exhibition organizer. “This probably won’t be enough to keep all the affected new outlets afloat, but there may be no alternative. Google needs to feed the beast, so to speak, and so it may need to preserve the trough from which it feeds.
Meanwhile, Google isn’t sure what its plan is for Google Previews. A Google spokesperson declined to comment for this story. Google can hardly be blamed for taking a “deer in the headlights” stance when asked to make a public comment, as there are clearly no easy answers to a problem created by a monster that they helped build: AI.
“Google is in a tough spot because OpenAI and other big AI companies have trained their models on most of the Internet and are cutting into search engine volume,” says David Smooke, founder and CEO of Hacker noona popular and community blog for hackers. “YouTube, owned by Google, is a particularly familiar dataset because it’s not just about how people write, but billions of hours of training data on how people speak in the In another world, YouTube could have trained only Google models, but that’s not the case today.
Mixed news for Google, searchers and publishers
Following the money is generally good advice when you’re trying to solve a business problem or investigate a mystery.
“Google catching up with the AI search trend alone won’t kill digital publishers. But combined with feature updates to search engine results pages and updates to their search algorithms intended to combat AI-generated spam, they are now killing digital publishers,” says Yoyao Hsueh, a digital publisher (Digital Surfer) and a member of the board of directors of the Association of Web Publishersa non-profit trade association that begins to advocate for the interests of small and medium-sized digital publishers.
While killing publishers’ businesses, whether intentionally or not, may be the result and one that will cost Google dearly after running out of data sources to power its AI models, there may be a method to the company’s madness . At least that’s what some observers think.
“Google appears to be changing its strategy from pushing users off its platform to keeping them trapped within its own digital walls. This pivot is quite evident with their AI previews, where the search engine now offers spoon-fed summaries without a whisper about their origins,” says Nyman.
But the problems are not as severe in the case of Google Previews. Google is likely to gain search traffic, but can still easily lose revenue. It’s quite a difficult situation. The duality of these results seems to come from two types of researchers.
“Some people look online for a quick answer to an immediate question,” says Michelle Symondsfounder and CEO of Ditto digitala UK-based digital marketing and SEO agency. “These people will be satisfied with a glimpse of AI. They are not interested in detailed expert opinion or an answer to a complicated or nuanced question.
This looks like a victory for Google in retaining and growing its search business in the face of challengers like Bewilderment aisince it is Google’s dominance in search that attracts advertisers. But this type of search can negatively impact the advertising side of Google’s business, as they are unlikely to consume ads on the search page or from network partners. This is potentially bad news given that Google makes most of its money from ads served alongside search results. Google ad placement brings in extra money through things like bid amounts. He also earns considerable sums from advertisements on partner sites in the Google network. Finding a balance between increasing search traffic and increasing search ad revenue is hard work. But a second time searching helps.
“The other type of researcher is one who is looking for reliable content from experts to help them make an informed decision that cannot be reliably found in an AI overview. So there’s room for Google’s AI insights and well-researched and attributed content from respected digital publishers,” says Symonds.
“These two types of searchers serve different needs at different times: rapid response to queries of relatively low importance and in-depth responses to searches of high importance such as typical YMYL (Your Money or Your Life) queries,” adds Symonds.
But the question remains as to which content publishers (many of which are retailers or other businesses) will fall sufficiently into the YMYL category to warrant a visit from this second type of searcher.
“As for profits, Google generates revenue from search ads only because users cannot find what they need on the search results page and have to click through to a site. So I don’t think Google will show AI summaries from commercial and transactional searches unless the sponsored AI previews or search ad results are superior to regular AI previews,” says Onajite Omare, Founder of Konvartan AI SEO tool provider.
“Data shows that AI is commonly present in information searches, so search marketing will evolve to be purely commerce-based for businesses while publishers look to use other platforms to more visibility,” adds Omare.
And what about all the investments made in top-of-funnel content – which is typically informational in nature – and designed to engage buyers at the earliest stages of the buying process? Will these investments become worthless when Google Previews take their place? So what will be used to direct traffic to specific websites?
“The problem is that AIO will remove clicks to websites,” says Vince Nero, director of content marketing at BuzzStreama digital PR and link building provider. “But most likely it will only remove the most important questions and terms from the sites. So instead of learning something like “what is content marketing” from a website, users will get the answer from AIO instead of going to the site. But this type of traffic is not very valuable for most websites anyway.
Summaries without sources, attributions or bias checks
Then there is the not insignificant question of the reliability of the information contained in Google previews. It’s no secret that generative AI models sometimes get it wrong, and the Google brand is no exception.
“To the public, AI insights are absolutely presented as fact or truth, which is a gross misrepresentation,” says Joe Karasin, SEO specialist at Karasin PPCa digital marketing company.
“As more consumers are exposed to the AI Overview, they are lulled into a false sense of security. Google’s AI told people to put glue on pizzaso it’s far from sure. As more and more misinformation is posted on the web and consumed by LLMs without any oversight, the public will be exposed to these hallucinations with no real way to distinguish fact from fiction, at least for those seeking ” quick responses.” Overall, these previews are bad for everyone, publishers and consumers alike. And until Google finds a way to present accurate information with attribution, this will remain a problem. »
Given the serious questions surrounding AI summaries – whether on Google, Bing or Perplexity – it’s difficult to see any value beyond user convenience, and even that is dubious. But other problems also loom.
“Additionally, when using large data sets of information, search engines such as Google may prioritize quantity over quality, which can have a significant impact on people’s access. individuals to information. For example, after the outbreak of war against Hamas in Israel, search engines slowly began to change what showed up when users tried to search for anti-Semitism and Israeli history. This shows that AI and algorithms prioritize large sets of information over potentially credible, historical or reputable sources,” says Kashman Starfounding partner of Cyber law firm.
“If this is the case and algorithms potentially exhibit bias on many occasions, people may be fed hallucinations that may also be racist and dangerous to society or minority groups,” adds Kashman.
The legal liabilities surrounding all of these issues put Google in a precarious situation with AI previews, as do its competitors who are adopting a similar tactic. Providing sources for research summaries is probably not enough to resolve or mitigate the risks.
