Published on November 28, 2023
Google, Bing and other search engines will offer AI super-snippets.
Many public relations professionals profess not to use AI, based on its potential to deliver truthy but inaccurate results. Generative AI products introduce errors even in summarizing a short set of facts, according to tests led by a former Google executive. Yet search engines have been using AI for years to analyze user behavior and display more personalized and relevant results. Google’s increased reliance on machine learning has made search snippets increasingly common and verbose.
Now, search is taking the next step by adding a chatbot interface. AI content creation in search raises a host of questions for the PR pro: Should a chatbot be the publicist’s sherpa through a mountain of market research? Will chatbot-led searches shrink the share of voice for corporate websites, social media and other PR platforms? Do marketers need a chatbot optimization strategy? ChatGPT won’t have the answers, so let’s work through a few of the questions.
Which Chatbots Are Best for Search?
Search engines have rushed to give their search products a chatbot front end. In November 2023, Google rolled out AI-generated search results in 120 new countries and territories. The Google Search Generative Experience gives users who opt in from Chrome browsers or the Google mobile search app get a chatbot-generated summary of top-ranked results.
Google’s parent Alphabet is following the playbook of Microsoft, which introduced Bing Chat as a sidebar feature in the Edge browser before integrating AI in its main search product. DuckDuckGo launched and quickly dropped its beta product, DuckAssist. Yahoo!, which had a similar experience in 2016 with its short-lived Radar travel assistant, has stayed on the sidelines.
Search chat draws from the new generation of AI models. ChatGPT collects a huge library of internet text but do not index or update it. Google SGE and Bard Chat draw text directly from indexed search pages, using generative AI as a content creation crutch.
Search bots also show their work: They link to their source material, lending a level of transparency to a product that still keeps search engine optimizers guessing about the logic of the underlying algorithm.
Google SGE: A Glimpse at SEO’s Future
SGE follows Google’s familiar format but with a brief outline composed on the fly while the browser downloads a search results list. In a challenge to e-commerce marketers, a search for “kids sleeping bags for camping” produces a list of important product features (insulation, shape, storage web sack) from product review sites, then describes top-rated products alongside thumbnail images.
A down-arrow button reveals a gallery of sources, ending with Google search. The display ends with follow-up questions, seemingly automated but still in the vein of “People also ask.” To optimize for chat results, marketers likely will have to write snippet-worthy catalog copy and court authoritative reviews.
The same query in Bard Chat is closer to the chatbot experience, composing a text-only summary of product reviews in the voice of a personal shopper. The text is chatty in Creative mode (“I’m glad you asked”) and matter-of-fact in Precise mode (“Here are some options”) but in any case the text is linked to its sources, with footnotes calling out their top-level domains.
In other SGE searches, not every page presents AI results; featured snippets are still the more common enhancement. Results for “recent stories on AI in public relations” seem routine; there are even two off-topic sponsored links about business uses of AI. The main difference is an “Ask a follow up” search bar at the bottom of the page. Type “How should PR professionals use AI in search?” to read four tactics from a Forbes Business Council post.
In Bing Chat, a search for “recent stories on AI in public relations” launches a chatbot essay on AI use cases, with linked sources from 2020 to two weeks ago.
Startups see generative AI as a business opportunity to develop a breakthrough search product. Perplexity generates a gallery of results and a footnoted summary. The results in one search session were impressively current, all published within eight or 10 weeks and one just the day before.
Do Search Chatbots Hallucinate?
Chatbots seem to speak with authority even when they are winging it. ChatGPT and its early competitors were notorious for misreading metaphors or jokes as factual. With search engines’ growing adoption of AI, marketers now must watch for such “hallucinations” in search results too.
In our testing, chatbots seem to get individual facts straight. But they tend to overgeneralize emerging topics that are thinly reported. Search chatbots seem to be fact-checking in real time: Users will see the chatbot add a few lines of text, then delete them. Content strategists must click through to the source material and ask follow-up questions.
Yet the model also picks up fake news for snippets and lets the Googlebot BS its way through a search query. As a result, PR pros should question the veracity of both snippets and summaries; while easily taken for granted, their authority can be easily sized up on clickthrough.
How Do Chatbots Change SEO Strategy?
Think of search chat as a snippet on steroids. Featured snippets use AI to find and present an authoritative answer to a simple question. Using generative AI, search engines now can string together key points from multiple articles and compose a useful overview. Search engineers will mix techniques in the future to present a mix of canned and on-the-fly results. Ultimately, AI soon might acquire enough speed and specificity to replace snippets entirely.
Artificial intelligence now seems to be everywhere, all at once, but search engines were among the early adopters of machine learning technology to study user behavior. Google took some of the juice out of keyword matching by using AI to find phrases that are semantically related. A query about “strategies for attracting business growth” will suggest articles on stimulating the economy just as well as typing “economic development.”
Get Ready for AI Super-Snippets
Google’s search analysis now puts more emphasis on experience, expertise, authoritativeness and trustworthiness. AI-enhanced super-snippets apply Google’s Multitask Unified Model (MUM) to screen featured snippets for alignment with high-quality sources. This applies to both standard and AI-enhanced results. It’s no secret that MUM’s the word for revealing user intent, compensating for the misspellings and unique phrasings in user queries.
AI also adjusts search rankings based on browser search history. This becomes obvious on road trips, when users ask for nearby retailers and get home-town results. More subtly, media relations managers may find that AI-shuffled search results are biased toward client sites they already frequent.
PR pros must keep extreme AI activity in mind as they prepare content to serve their target audiences. To see plain-vanilla Google search results, turn off Chrome’s user customizations: Navigate to “Data & privacy” and pause the History settings for Data & App Activity and Location History.
An enduring content marketing trend has been to recast clients’ copy to be more direct and snippet-worthy. As search engines adopt more generative AI features, marketers will have to redouble their efforts to produce customer-responsive, chatbot-ready content.
Technology PR Case Studies
- Aviation Services Branding Takes Flight
- Birthing an International Life Sciences Center
- Establishing a New Retail Technology Leader
- Social Media Framework Sets Stage for Project Management Success