The search bar, as you know it, is dead, or at least the version of it that required you to think.
Google, Bing, Perplexity, OpenAI, Yahoo! are all racing to replace the traditional search box with something that simply gives you the answer. It is a blurring of the lines between search engines and AI that promises convenience and quick answers to everyday questions. From the best local restaurant to healthcare, we all have questions. However, as a researcher, the human making judgement in those moments, it also replaces you.
AI Search
Notably, Google introduced a new AI powered intelligent search box, but they are not alone. A significant change to the 25-year-old style search that we all know. It extends AI mode, which on its own surpassed one-billion monthly users, proving AI is popular, but it might not also be what we need, or what you want.
There seems to be dwindling choice. With Google’s lead others seem to follow. Users who might have found themselves toggling between search tools and GPTs are more often just using AI as a substitute for traditional search.
Maybe this is the point. As Google states, sometimes “…curiosity doesn’t always fit into keywords” and it needs AI to extend the conversation. The question is will AI help us be more curious or simply more dependent on the answers given? Will our curiosity end with a convenient answer?
Mental Offloading
AI is easy, and we all like a little convenience which is fine when the stakes are low. Finding a nearby grocery store is not an issue. Using AI to diagnose an ailment or make a major financial decision is another matter. When AI scans the internet, webpages, and news sources, it is doing the heavy-lifting we are accustomed to doing with traditional search. We input keywords for a topic, and we get a list of hyperlinked sources. Follow a few. Read the pages. Scrutinize the sources. Maybe search a few more before making a decision. These are the pillars of research that we often do not recognize we are doing. A pattern of search we have been trained as a society to follow for decades. What happens when this changes?
The dynamic shift to AI answers is another form of mental offloading, where we substituted the easy for the more difficult tasks. AI is doing the work for you, but it does not do it the same way that you would yourself. It does not know your personal preferences (yet) and it has proven to be unable at times to access accurate information. Having been trained on the internet, it is vulnerable to misinformation and a variety of diverse views on contested topics.
Most of the popular AI models still have a hallucination rate of 3.1% to 19.1% depending on the model and topics. Reliability remains an issue with limits.
Recently BBC journalist Thomas Germain demonstrated that in addition to hallucinations, AI could also be injected with misinformation. He created a false news story on a personal website about the hot eating champion among tech journalists. Shortly afterwards, he asked Gemini and ChatGPT, who parroted the information as fact. Obscure topics and deliberately created alternative facts that AI models consume make their answers vulnerable. And when we cannot see behind the curtain where and how AI gets its information, then we cannot verify and are left to blindly trust.
By the time they view just five consistent messages their beliefs are solidified
Belief Formation – 5 Times Charmed
Relying on AI answers alone can have an impact on how we form our beliefs. A recent study revealed that it is a surprisingly quick process, especially for unfamiliar topics. When people encounter the first few social media posts they begin to build a scaffold for their beliefs. By the time they view just five consistent messages their beliefs are solidified. A highlight of the research also shows as those beliefs are formed it becomes increasingly difficult to unanchored them. Individuals resist information that contradicts what they have already accepted.
This type of information fluency works against us online. We rely on mental heuristics, shortcuts our mind takes to be efficient, when we see new content. Unlike structured information settings, social media presents users with rapid snippets of pictures, video, and text, that are fragmented across a range of topics. It is a whiplash of ideas all at once. Those heuristics help us make quick decisions of it all. Unfortunately, they also defer our more rational thinking and evaluation. It is those quick decisions during the scroll that make us vulnerable to misinformation because as we take it all in, we are looking for things that simply ‘fit’ our own narratives. The more it easily fits, the more we believe.
AI has its place, but deciding what is true probably isn’t it. Traditional search made you a researcher. The new one makes you an answer recipient. We are moving from using our judgement, the human-in-the-loop process, to one that quickly gives us information without the inconvenient nuisance of rational thinking. In our hurried lives, that convenience has a premium. The real cost is not just bad information, it is forgetting we were supposed to be looking for it.
Discover more from Derek W Gibson
Subscribe to get the latest posts sent to your email.