AI is everywhere these days – we’re constantly reading about what the technology is capable of, how it can improve our lives, or what human tasks it can replicate (or threatens to replace altogether). One area that is poised to get a big AI upgrade is something most of us use every day – search engines.
There are no prizes for guessing who the front runners are for AI-assisted search engines. Google announced in May 2023 that they would be starting an experiment to enhance the search experience with generative AI, but as of yet, there is no definitive word on when or even if the project will become anything more than just an ‘experiment’.
What is Search Generative Experience?
Search Generative Experience (SGE) is Google’s new search experience that takes your search query, feeds it into an AI engine and returns the response.
The idea is that the use of AI provides a much more intuitive response to your search query. For example, if you were to type “best headphones for the gym” into Google, rather than just a list of responses, you would get a rich snippet advising you on the factors that make a good pair of gym headphones, along with some product recommendations, and maybe a few comparison blogs or other useful resources to help you guide your decision.
One of the standout features of SGE is the conversational ‘ask a follow up’ box. Instead of a series of searches that the engine makes pretty much in isolation, a search can be informed by previous ones. This allows the AI to use your previous search queries and results to formulate a more accurate answer to your follow-up question, because it now knows the two are related.
So going back to the headphone example, if your previous search gave the consensus that over-ear headphones are best for gym workouts, and you follow this up with ‘where to buy headphones’, you’re likely to get results based on shops and websites that sell over-ear headphones.
What factors affect the search results you’ll see in SGE?
The response SGE gives is determined by its preexisting data model, but it will presumably have a level of personalisation based on the user profile, previous searches and noted interests, creating search results much more tailored to the individual user.
In the same way that their existing search engine works, SGE will probably also tailor results based on what others have been using and whether the provided answer is ‘accepted’, and what users do after receiving it.
If the answer is way off and the user ends up clicking a link, then it is a good bet the language model will then visit that site and use the content to ‘train’ itself (ie. it will copy the content, paraphrase it and add it to its growing data model).
In addition to providing answers to straightforward factual questions, SGE will also provide rich snippets as it does now in its shopping results, again with the extra context of the individual, the results of previous searches to the same query, the specifics of the request and of course the level of ‘sponsorship’ of each option.
Accuracy is still important, so in cases where there are obvious risks to health, wealth or security it will tend to stay away from direct advice (such as medical or financial advice from unverified sources), pushing the responsibility back onto the website owners. Getting sued is expensive, after all.
What are the downsides of SGE?
So this all still sounds pretty good right? Why wouldn’t we want search results that are more accurately related to our queries, more comprehensive, and that allow us to ask follow up questions without starting from scratch? Well…
Google claims that SGE will improve the search experience even more, will drive better insights and further improve the structure and consistency of websites and on the web, and it will probably do so.
However, It will also improve the ability to target individuals for certain types of advertising and (paid-for) advice by drilling further into their intentions and needs – all very profitable.
More than this, it means that search results can be more easily manipulated by those providing them, and that you are more likely to see what the advertisers want you to see, rather than what you necessarily what you are looking for.
Most importantly, it seems that this is another way for Google to keep you on their platform. And the longer you stay on their platform, the more opportunities there are to show you ads and gather data.
What does this mean for SEO as we know it?
Search rankings are a simple, measurable metric. SEO is carried out by best practices and educated guesses, and by putting in the work and optimising a site, businesses attempt to rank as highly as possible for their key search terms to increase their chance of being seen (and clicked on) by their target audience.
However, SGE theoretically has the power to decimate organic SEO as we know it. If people are getting the result they want without having to search through a bunch of links then it doesn’t matter how highly you rank. Sites will still have to employ good SEO practices to rank high on the SERPs, but even the coveted position 1 might be difficult to see if it needs to wait in line behind the SGE results.
If this is the case, paid SEO will become the only way to get noticed in the screaming match of results, and there is only ever one winner there. SEO will go from being a dark art to a lottery where the only way to win is to buy copious numbers of tickets.
The final word
SGE has the potential to be a game-changer, whether you are in support of it or not. The functional aspects seem to be a natural evolution of what Google was trying to achieve with rich snippets – answering user queries without having to click through and find the relevant section of a web page. ‘Ask a follow-up’ also has the potential to provide much more case-specific, well-rounded advice for users.
But this all becomes murkier if the results and advice given are heavily influenced by who is paying the most, and users are being shown the results most likely to result in a sale, rather than the results that provide honest and impartial answers to their query.
At the time of writing, there is no launch date for SGE. In fact, it is still being branded as an ‘experiment’ by Google, with rumours it will never launch at all. However, it seems unlikely that even if SGE doesn’t see the light of day in its current form, that something very similar won’t take its place once the kinks have been ironed out and it has been given the obligatory facelift.
For now, we just have to keep watching Google’s ‘experiment’ and see how it evolves.