AI Briefing: Generative AI and search are becoming increasingly intertwined — and top of mind

At this week’s Google I/O 2024, many marketers will be tuning in for any updates on the future of generative AI and search.

The annual developer conference, which takes place on Tuesday, has always been a place for the search giant to highlight major innovations for its various audiences. At Google I/O 2023, the company introduced its search generative experience (SGE), the generative AI search tool at the top of search results that’s been in testing for the past year.

As conversations with chatbots becomes more commonplace, generative AI will evolve SEO by challenging brands to optimize for natural language and accuracy instead of key words. Underlying all of this will be the need for credible content, but that’s a challenge with plenty of unclear strategies and solutions.

“When it comes to credible content I would say that the onus is on the chatbot,” said Craig Elimeliah, chief creative officer at Stagwell’s Code & Theory agency. “I do think that conversational and real-time responses force chatbots to have to fact-check and provide transparent citations to build user trust. You won’t talk to something you don’t trust.”

According to Elimeliah, each platform might have a different focus when it comes to credible content. While Google SGE is about authoritative fact-based content, he thinks Microsoft’s Copilot might be more tailored for B2B queries. Meanwhile, Meta’s Meta AI might focus on discoverability for social and audience engagement. OpenAI’s ChatGPT could emphasize source transparency, visuals and conversational content. 

It’s unclear how generative AI will change search — or how fast, for that matter — but agencies are already strategizing for generative AI’s impact on traditional search, generative search and answers from chat-based bots. Some are also researching and testing the current situation while also building new platforms in preparation.

Graphite, an SEO agency, released a new SEO platform earlier this month focused on topics instead of keywords. Using AI tools like natural language processing (NLP) and deep learning, the platform analyzed hundreds of millions of keywords to create a topic graph to help reverse-engineer user intent. According to Graphite co-founder Marcos Ciarrocchi, the tool might also be helpful for companies to know which topics they should focus on if they want chatbots to cite their websites more in answers to user questions.

“This is all connected to the original user intent, but it goes deeper,” Ciarrocchi said. “So we’re able to surface some of these ideas, [which] will give you kind of dynamic feedback as you’re creating content to tell you if this a good answer or not [and] what can you do to make it a better answer.”

By using the SAAS platform, clients can find topics relevant to their audiences, identify topics where they’re already an authority, create content optimized for higher ranking and measure results on a day-to-day basis. Subscribers can also analyze topics to see their areas of authority compared to competitors and look for the gaps. The platform is built around Google’s focus on topical authority rather than PageRank.

Even as humans overhaul their SEO strategies, another question arises: How will the changes compete with AI-generated content? Despite growing concern around AI-generated website content ranking higher than more authoritative publishers, Graphite says most URLs on Google aren’t for pages written by bots.

Using software from Originality.AI, Graphite analyzed nearly 5,000 URLs across 10 categories. Only 8% of URLs directed to content were either entirely or largely created by AI. Just 1.9% was AI-generated with another 6.2% a mix with high AI content. However, another 38.3% was determined to be human content mixed with AI while another 53.6% was entirely human-created. The human content also tended to show up higher in search results than entirely AI-generated content, which averaged five positions lower than human content.

Other agencies have been researching how search results vary in generative AI platforms. SOCi Inc. looked at how six platforms compared when asked about a range of local topics, scoring them based on accuracy, relevance, reliability, helpfulness, maps, and other criteria. Out of a total 100 points, it found Gemini scored best — followed by Google SGE, Perplexity, Bing and Meta. It also looked at ChatGPT, which was unable to answer most queries.

Each generative AI tool “is going to mold itself to the intent in a more direct way than search so far has been able to,” according to Damian Rollison, director of market insight at SOCi. He added that will make bots’ abilities to answer questions even more important than traditional search.

“People are going to have to start sussing out the difference between hype and usefulness and utility,” Rollison said. “We’re just starting to learn that language, and it’s important to focus on the priorities that are actionable and relevant for businesses.”

Generative AI’s impact on search has also come up in recent filings by publicly traded marketing and tech companies. Last week, Semrush’s first-quarter results mentioned recent investments in its new enterprise SEO platform, AI-powered SEO alerts, and other tools. Other companies that have mentioned generative AI — related to search or otherwise — in recent financial disclosures include Reddit, Nextdoor, Pinterest, Yext, Similarweb and Eventbrite.

Some companies, including Airbnb, said changes by major search platforms such as Google might affect their businesses. According to Airbnb, “If Google or Apple use their own mobile operating systems or app distribution channels to favor their own or other preferred travel service offerings, or impose policies that effectively disallow us to continue our full product offerings in those channels, it could materially adversely affect our ability to engage with Hosts and guests who access our platform via mobile apps or search.”

Prompts & Products: Other AI news

  • Meta announced new generative AI ad tools with more ways to create images and text based on reference images and past campaigns.
  • The Met Gala partnered with OpenAI to create a custom exhibition featuring a chatbot with the persona of the 1930s New York socialite Natalie Potter.
  • An AI-generated image of Katie Perry at the Met Gala went viral, illustrating how hard it is to identity deepfakes and stop them from spreading misinformation. 
  • OpenAI joined the Coalition for Content Provenance and Authenticity and joined C2PA’s steering committee. It also announced a new Media Manager platform to help creators protect their content from AI-training, but critics say it won’t solve issues around copyright concerns.
  • WPP was targeted in a recent phishing attack, according to The Guardian, with fraudsters attempting to scam the holding company using an AI deepfake of WPP CEO Mark Read.
  • Apple announced new AI tools for its Logic Pro software, but also faced widespread backlash for its new iPad ad that some saw as a metaphor for how tech companies are crushing the hopes of artists and creative industries.
  • TikTok said it will start automatically labeling AI content on its platforms, which will begin including Content Credentials being adopted by major tech companies. 
  • The AI audio startup ElevenLabs previewed a new tool for creating AI-generated music made with a single text prompt.
  • DotDash Meredith joined the growing group of publishers to make a deal with OpenAI and allow the startup to train with content.
  • Sprinklr announced a new “digital twins” tool to help companies make AI versions of their brands for customer service bots.
  • Starburst released a new campaign that uses generative AI to create new scenes in a range of styles.
  • Retail brands including Reebok and Adore Me debuted separate features for designing digital and physical items with generative AI.
  • ServiceNow debuted a new campaign starring Idris Elba to showcase the company’s AI capabilities.

1s and 0s: AI research, surveys and other numbers

  • A majority of Republicans, Democrats and Independents surveyed by the Artificial Intelligence Policy Institute said AI companies shouldn’t be allowed to train their models with publicly available internet data. 
  • The New York Times has spent $1 million in litigation costs in its lawsuit against Microsoft & OpenAI, according to first-quarter 2024 results released last week by the Times.

Others stories from across Digiday

https://digiday.com/?p=544676

More in Media

AI Briefing: Why WPP is adding Anthropic’s Claude models to its AI platform

Choosing which AI models to use has been a key factor for companies as they develop AI strategies for marketing and other applications. 

Inside The New York Times’ plans to correlate attention levels to other metrics

There’s a lot of buzz around attention advertising right now, but The New York Times is trying to stay grounded even as it develops its own plans.

Why publishers are preparing to federate their sites

The Verge and 404 Media are exploring the fediverse as a way to take more control over their referral traffic and onsite audience engagement.