The most obvious development in search over the past few years is the implementation of AI. First, it was rank-brain (AI sorting algorithm). Now, it seems to be AI-generated ‘featured snippets’. AI was launched for search in October 2024, and Google documents and tools are embedded with the company’s rival to ChatGPT, Gemini (which, is terrible, but that’s another post).
While news outlets may hail this as the most recent step in a marathon of groundbreaking innovation and progress, we look at it with the eyes of a suspicious ( hopefully, wise) elder; these new ways of serving consumers appear fantastic on the surface, but, even a small amount of picking through the top-soil reveals a greater number of problems.
If you’re interested, we’re going to share what we think about search engines using AI.
1. Search engines are becoming the arbiter of what is correct, incorrect, truth or mistruth.
AI in search is not new, but it is continuously creeping into significance for users.
Previously, Search engines, like Google, leveraged the galaxy of data indexed onto their database to determine which websites will serve the best interests of a searcher (whether they are asking a question, or looking for the best camera for their budget). It has always been the case that search engines – almost by accident – are the arbiters of creating what becomes “true”. Politically or otherwise.
For instance, the algorithm may favour a huge authoritative site, like the BBC, over a less authoritative website of a newly award-winning independent journalist. Therefore, the BBC gets more visibility, and monopolies the narrative.
Which is best for the searcher? A cooperation with years of authority or the rising star? This is a conundrum that all search engines face; and a problem we all face in life. Do we trust what is already established source, or take a risk on a new discovery or point of view?
How do search engines get around this? They democratised the offering of a search results page (SERP), so, searchers have access number of sources at once. Google would use their data (like links, page content, user behaviour etc…) to determine which result should appear highest in the rankings. They’d (appear to…) offer a range of different viewpoints on contentious issues.
It’s true to say that the SERP results are never “perfect”, but, at least a searcher had access to a number of independent source to make informed (as much as is possible) decisions, with relative freedom to pick which is best for themselves.
AI is now being deployed, to replace much of the SERP, to make it “easier for users to get the answers to the questions they are asking”.
As you can see above, in a search for the word “arbiter”, Google has taken a significant amount of space in the search results. It takes a number of seconds to scroll past the Google suggested content, to the actual organic results.
Why would Google swallow more space?
There is a serious argument to be made that this is a further attempt by Google (and other search engines) to monopolise search results and have ultimate control over what users are exposed to. This provides more space for Google to place sponsored content (which they get paid for promoting) and provides an opportunity to control what users are exposed to.
Google is deploying AI, and in doing so, they are taking away agency from a searcher, and asserting what it thinks is best.
Does that improve the searcher experience or does it open the potential for abuse? Is AI really sophisticated enough to be able to accurately – and fairly – assess whether information is correct, incorrect, good, bad or indifferent?
Should Google not remain a Library of sources and retract from it’s current position of an arbiter? Who, or what, decides what is the best outcome for a searcher? And, what if AI provides an answer that is simply wrong?
2. AI, although very clever, is sometimes wrong.
It didn’t take long before the early AI models to be outed as a Nazi’s. They are fairly prone to making stuff up too. Why? Because humans programmed them.
Google rolled out AI-featured snippets in October 2024 which came with some unintended consequences; namely suggesting glue was a great topping on pizza. Google makes up to 90% of global search traffic (albeit dropping), so it is bewildering that they decided to roll out such an ‘experimental’ service. Especially considering the technology is anything but reliable.
AI models absorb data given to them to train their behaviour – it is why Microsoft’s early AI model went so far off the deep end. Google is training its AI using information gathered from the search results, which could be incorrect.
The Google AI is reporting that the body contains 60,000 miles of blood vessels, which has recently been debunked.
There is a twist in this, however. Just below this featured snippet, the same search results from the British Heart Foundation are also reporting the same ‘misinformation’, so, even before AI, it is certain that searchers would have taken on this falsehood forward (as it has been shared for generations).
Using AI also removes a human from the decision-making process. AI may not know the difference between a joke or a serious answer to a question: a possible reason for Google suggesting users put glue onto a pizza is likely a Reddit comment (which Google uses to train it’s AI) . Most, humans understand it is a joke, but AI reads this as the most valuable (signalled by upvotes, links etc.. ) answer to the question.
This is not a post-finger-wagging about search engines providing results that are incorrect, it is more concern for the move into being an “arbiter” of information, or an authority on correctness, rather than simply a Library of sources. To cross that threshold is potentially dangerous.
Is it not the job of a librarian to point you in the general direction of an answer, rather than give you a lecture on what they believe to be true themselves? This move towards outsourcing correctness to AI could have catastrophic consequences.
Imagine, if somebody put glue on their pizza!?
3. There is simply too much information to handle.
To finish this post, it can’t be easy trying to control a social media platform or search engine. With pesky SEO’s hacking the algorithm to push their clients up to the top of search results, it’s understandable to see why search engines are looking towards technology to help provide the best answers to users.
AI seems like the way forward, but it may be an easy way to outsource an impossible problem. Technology, it seems, may be causing all of us more trouble than we ever imagined. It also still has a fatal flaw: us, humans.
Why should we be worried about search engines using AI? Well, AI is being used to fix an impossible problem; to determine what information is valuable in an endlessly growing galaxy of information and attempting to monopolise the information it already services to users. AI removes the human, and the technology takes it further away from human comprehension (ie – Skynet and the end of the world).
Google is already coming under increasing pressure to limit its monopoly across multiple sectors, and AI may accelerate the monopolisation of search, tech and the information we believe to be true.
What does this mean for the future? We suspect that search engines will need to roll back the way their platforms make decisions. Old school ways of assessing value are surely to come back in fashion!
How to we move forward?
The answer to this, we think, is simple. Be more human. Just be a person, living in the real world, and see these new technologies as tools to improve our lives in the real world. Let’s all walk together into this Brave New World.
