There's a certain way I like researching topics. A lot of the time, when I watch somebody trying to figure out a problem by searching it up on the web, I get really annoyed. They get this tunnel vision, only looking for results with their exact problem, in that exact wording, disregarding anything else. They seem to have this expectation of the problem already being solved for them and stop once they don't find that.
Now that the era of LLMs has befallen us, this has gotten exponentially worse, with people expecting to get problems of any complexity and obscurity answered by "AI" and growing more and more confident in its findings. The idea of looking through several resources to find an answer seems to be a burden, when there's this all knowing machine that will answer the question in mere seconds.
I really dislike this mindset and hate how it's shaping the search engines people use with it. I don't want to be given just one result. The chance of it being wrong is rather high, but it also just doesn't give me enough context on the problem at all. It's evolving into this death spiral, taking any understanding of topics away from people. The search engine is starting to keep people from learning about things.
When I research a topic, I want to see anything that matches my search terms. Especially the results that don't quite describe or solve my exact problem. I want the wrong results as well.
Now in the best case, I just find a solution to my problem right away. I might try to read some more to actually understand it, but for the most part I'll just be happy, getting my problem solved, my question answered.
Does that however not happen, I try to read the other results as well. If I can't find an answer to a question, the solution isn't to search again until the 5th search page. The next step should be to search for something else. Read related problems, read up on the fundamentals of the area the problem lies in. Find a way to the solution, not through the brick wall in front but through the door described on another map.
The solutions to a different problem, sharing the same base might give me just the bits I actually need to solve my own. In some cases there may just not be anything on the particular problem at all out there yet. In those cases the key is to understand the core of whatever you're working with first. Once you actually understand your environment, finding a solution will come a lot more naturally. The reason you can't find anything on it online might not be because it's so obscure, but because it's so simple.
There are also many cases where an article or forum thread doesn't help with the problem altogether. But I don't see these as a loss, I like learning about things and each piece of extra info helps building my understanding. There have been various cases in which such prior "random trivia" gave me more understanding in or outright solved a later problem.
I often find myself solving problems others researched for hours with a simple search and some digging in the results within minutes. This of course also requires some understanding on the given topics, which is exactly what you build by not ignoring the more vague and seemingly non-matching results.
Experiencing such cases just makes me sad, because I don't have more or better resources, maybe I'm not even more educated on the topic at all. They just refuse to actually look into it and learn about the topic. In the end they are unable to find solutions to their problems. Given how often I learn about it, I can only imagine how many things they encounter, that they just end up not being able to solve, because of their primitive researching skills.
Regarding LLM usage to answer questions, it looks even more grim to me. I certainly know the feeling of just wanting a problem solved quickly. After also having gone that route for a few months, I realized two big problems though.
Every search you just ask like that, you essentially miss out on knowledge in. As I've explained in the prior paragraphs, there's a lot of knowledge and understanding you can build up by reading through articles on your problem, rather than just getting the solution right away.
But it's also a terrible slippery slope. At least as I've seen it with myself. The usage for only simple unimportant questions, turned into solutions for IT problems I faced, before turning into actual debugging of code, as I became lazier and lazier.
I fear there will be so much knowledge and understanding lost if people actually continue to use LLMs to answer their questions and solve their problems.