Google Search’s future is a large green box.
This is exactly what Google showcased this month at Google I/O. The company’s annual developer conference. AI was the theme of 2023, and it was mentioned over 140 times in the keynote presentation. Google announced AI products which will be made available to the public. This is a dramatic change for the internet giant, who was frightened by the growing competition.
OpenAI launched ChatGPT late last year to almost universal adulation. Everyone had instant access to a generative engine that can answer seemingly any question in a novel way. It is powered by a large-scale language model (or LLM), which allows it to act like “autocomplete on steroids”, using huge amounts of text data in order to determine what the next word should be.
ChatGPT’s power and simplicity helped it to become the fastest-growing consumer web platform in history. Microsoft increased its investment in OpenAI, and integrated ChatGPT’s technology directly into Bing earlier this year. This helped the company to see a 16% rise in traffic. Google launched its own AI engine, Bard the day before Microsoft announced Bing AI. However, it botched the launch, and lost $100 billion of stock market value. Since then, the stock has recovered to its highest level this year.
Google I/O in many ways was a referendum about the company’s bumbling entry into consumer AI, and a message to investors that they are willing to make radical changes to remain at the forefront of Internet search, even if it means changing their core product. Google Search is the main engine that we use to find product information, the latest news, and interact with the web. It’s also the way many businesses make their money.
SGE is an experimental version that prioritizes 10 blue links, which have been the hallmark of Google’s search for the last quarter century. The answer to any question, no matter how specific, is displayed in a mushrooming box of green that grows as the person’s response fills up the screen.
“Now, search will do the heavy lifting,” said Cathy Edwards during Google’s I/O. She explained that, in the Search of today, complex questions have to be broken into smaller ones, where the user has to go through information and form the answer themselves. SGE does all that automatically and even allows you to ask further questions.
Google Search Generative Experience (also known as Google Search Generative Experience) is an experimental version that integrates AI generated results similar to Bing or ChatGPT.
CNET
It also means that you don’t have to visit multiple websites – the clicks on which webpages depend, and could potentially upset the internet business model based on ads.
Statcounter reports that Google has a 93% share of the online search market. According to a Brightedge Research report from 2019, online search engines are the biggest drivers of website traffic. 68% of all online experiences begin with a search engine. Google’s dominance of search helped it reach a valuation as high as $2 trillion at one time.
Google’s SGE could be a game changer for internet users, businesses, and the AI industry. It will require a rethink of how information can be distributed while incentivizing content creators to create valuable content.
SGE has just begun to accept signups, so there are no data available yet on the user experience. Microsoft has been collecting feedback on Bing AI for the past three months. This could give a good idea of how users will react to AI-driven Google search.
Microsoft spokesperson said that feedback on answers generated by Bing was mostly positive. 71% of people in the preview gave the AI-powered responses a thumbs up. We’re seeing healthy engagement with the chat feature. Multiple questions are asked to find out new information.
Uncertainty surrounds how AI-generated stories of news will be incorporated into Google or Bing AI results. CNET and other publications are already experimenting with AI-written articles. Unfortunately, AI is not always accurate. It can also have “hallucinations”, where it says that something is right when it’s not.
If the problem of hallucinations is finally solved, generative AI could make search faster and better for consumers. It’s not yet clear how it will affect the digital publishing sector, particularly if people stop clicking on links in mass.
Google spokesperson: “As Google experiments with new LLM powered capabilities in Search we will continue to prioritize approaches which send valuable traffic for a variety of creators, and support a health, open web.” Although it is true that Google links to sources prominently within SGE, there’s no guarantee the SGE will result in increased or better quality traffic.
Microsoft did not answer any questions about traffic to sources using AI search. Google has said that it does not have any plans to disclose information about publisher compensation, but will “continue to collaborate with the wider ecosystem.”
Monica Ho, Chief Marketing Officer at SOCi – a digital marketing firm – said, “I believe [generative AI] will reduce the amount of traffic that goes out, because that’s its purpose.” She believes that the traffic to sites is higher quality because people are searching for specific information rather than bouncing around.
Google may not have any viable alternatives if it loses its ability to deliver traffic. Rasmus Kleis Nielson, professor of political communication and director of the Reuters institute for the study of journalism at Oxford, said that social media platforms like Facebook are unreliable partners, lowering news rankings on a whim. He said that platforms such as Instagram or TikTok, “drive relatively few referrals and do not feature links in the same way that search and social does.”
Search engines crawl websites every day to gather new information. They then index this into their results. The conversion of traffic allows websites to allow crawling for free. If AI-search results in fewer clicks then the search economy could need a complete rethink.
Don White, CEO at Satisfi Labs – a conversational AI firm – said: “I expect original content to be behind paywalls that require LLM models pay to read.” White envisions a world where sites will be paid-per view in a “Spotify style compensation model”.
Google will need to figure out a way to get revenue to creators and publications, so there is still an incentive for them to produce quality content.
Ho said that “Google doesn’t create all their original, unique content, but they do feed it with quality data.” It has to be created by creators. They understand that they will have to feed the engine in some way and make it worth while for content to continue to come.”
Editor’s note: CNET uses an AI engine to create personal finance explainers, which are then edited and fact checked by our editors. See this post for more information.
