The Question May Contain an Answer That Isn't True
I Googled Garth Brooks. I wasn't asking whether people still liked him. I saw a Google ad that he was coming in concert to Salt Lake City and was curious. I used Google's normal search, not AI Mode. As I scrolled through the search results, Google showed me the familiar section: People also ask
One of the questions was: Why did people stop liking Garth Brooks? That caught my attention. People stopped liking Garth Brooks? That's making the assumption people stopped liking him as well as saying here's why.
I clicked it.
Google displayed an AI Overview, explaining that people hadn't “entirely” stopped liking Brooks. That his “cultural standing shifted” because of overexposure, changes in the music industry and controversial public positions. It went on to offer explanations. Several of the sources displayed alongside these explanations were Reddit discussions.
There was just one problem.
Before explaining why people stopped liking Garth Brooks, shouldn't we first establish that people did stop liking Garth Brooks?
Did They?
The reason I had goggled Garth was an ad pooped on my screen promoting his arena tour in Salt Lake City. Before his Salt Lake City dates had even announced, more than 200,000 tickets had been sold for his 2026 arena tour, following eight sold-out shows. According to the Delta Center, the queues for this Friday and Saturday performances each exceeded 50,000 fans. Demand was so high another performance was added, brining it to three nights.
That doesn't prove that Brooks has never lost fans. It gives me a reason to question the Google AI premise displayed in the People Also Ask section: People stopped liking Garth Brooks.
The assumption built into the question
Consider the question: Why did the bridge collapse? There is already an assertion contained within it: The bridge collapsed. The question isn't asking whether the bridge collapsed. It is asking you to accept that it did and move on to determining why.
The same thing happened with: Why did people stop liking Garth Brooks? The question presupposes: People stopped liking Garth Brooks. I didn't make that claim. Google introduced it to me. Had I searched for, “Why did people stop liking Garth Brooks?”, I would have owned the assumption and brought it into the search myself. I didn't. I searched: Garth Brooks. Google presented the loaded question.
What “People Also Ask” means
This is where I discovered something about how Google Search operates. I had interpreted, People Also Ask as: Here are useful questions about this subject based on information Google has established.
That's not what Google means when it says “People Also Ask.” Google describes People Also Ask as questions related to what some originally searched for. That's different from: Google has verified the premise contained within each of these questions.
People search for things that aren't true.
People discuss things that aren't true.
People repeat rumors, exaggerate, misunderstand history, express opinions as facts, and ask questions based on wrong assumptions.
A search system can recognize those questions as being related, without establishing that the premise contained within the question is true. When Google shows, Why did people stop liking Garth Brooks? we should not interpret that to mean, Google has determined that people stopped liking Garth Brooks. Would you like to know why?
AI makes the assumption even more convincing
In my search, something else happened. I wasn't using Google's separate AI Mode. I was in Google's ordinary All search results.
When I expanded the People Also Ask question, Google displayed an answer labeled, AI Overview.
AI Overviews are part of ordinary Google Search. You don't have to deliberately enter AI Mode to encounter a generative-AI answer. Google says AI Overviews appear in search results when its systems determine that generative AI may be useful.
This is dangerous for someone genuinely trying to learn more about something.
Once an AI provides five or six plausible explanations for something, the explanations make the original claim seem more credible.
Google gave me another example on the same page
There was another People Also Ask question in my search results:
What ended Garth Brooks' career?
Think about the assumption inside that question! Garth Brooks' career ended. Ironically, in the add next to the People Also Ask, Google was showing me an ad for the upcoming Garth Brooks arena concerts. The page itself, a Google Ad, contained evidence that should have made Google AI question the question, and made us question the premise of one of the other questions Google was inviting us to explore.
Better questions would be:
Did Garth Brooks' career decline at some point?
Why did Garth Brooks temporarily retire from performing?
Reddit wasn't the problem
Reddit can be extremely useful. If I want to know, why do some country music fans dislike Garth Brooks, Reddit may be exactly where I should look. People can explain their opinions directly. Someone might say, “I thought he was overexposed.” Another might say. “I didn't like Chris Gaines.”
Those are legitimate answers to the question:
Why don't you like Garth Brooks?
I tried the question myself
I tried something else.
Instead of searching for Garth Brooks and clicking Google's suggested People Also Ask questions, I typed in, "Why did people stop liking Garth Brooks?", directly into Google's search box.
This time Google's AI Overview gave me a different answer.
It began: “Garth Brooks faces significant public backlash and legal scrutiny primarily due to a sexual assault and battery lawsuit filed against him by a former hair and makeup artist.”
Something was still missing. None of that established that people had stopped liking Garth Brooks.
A lawsuit exists. Allegations were made. Brooks denied them. Some people criticize him. Those are facts and opinions relevant to Garth Brooks. They don't answer the question: Do fewer people like Garth Brooks?
Google had answered why without establishing whether. Tens of thousands of people were waiting in online queues for tickets to his Salt Lake City shows, and a third show was added because of the demand.
There's a larger problem here with AI-generated answers. The information AI gives you can be true, while the conclusion it leads you toward is false.
This becomes a bigger problem when the subject matters
Garth Brooks is as perhaps a harmless example. Imagine instead that I ask Google:
Why did the United States bomb Iran?
Suppose Google gives me an AI answer based partly on Reddit comments and partly on blog posts saying we bombed Iran because of XYZ. I may walk away, believing XYZ is the reason we bombed Iran. And for that matter, that we did in fact bomb Iran. Google may have summarized what people on the internet are saying.
I may mistake people's opinions for facts. AI makes those opinions sound like an answer. I leave believing something that may be biased, incomplete, or wrong.
That's the bigger problem.
If I want to know why a country went to war, why a medication was withdrawn, whether immigration increased crime, or whether a politician did something they're accused of, the quality of the source matters.
A Reddit discussion may be useful for answering, What do people on Reddit think caused this? A blogger may be useful for answering, What is this person's interpretation?
Neither answers: What actually happened? or: Why did it happen?
Those require evidence.
AI can blur that distinction as it doesn't present ten Reddit comments, three blog posts, and two news articles - It gives you one answer. The different sources have been synthesized by AI into fluent paragraphs. The uncertainty, disagreement and differences becomes less obvious.
That creates a danger for someone using Google if they genuinely don't know the answer. If I already know a great deal about Garth Brooks, I might see: Why did people stop liking Garth Brooks? and think: Wait. Did they?
Someone who knows very little about him may instead come away with learning: Apparently people stopped liking Garth Brooks. AI supplies the reasons. The person doing a search may never realize that the first thing they “learned” was never established.
The question changes what you look for
That's why the difference between these two questions matters so much:
Why did people stop liking Garth Brooks?
and: Did people stop liking Garth Brooks?
The same applies to:
Why did we bomb Iran?
versus: Did we bomb Iran? What evidence do we have about why?
Why did immigration increase crime?
versus: Did immigration increase crime?
Why is this medication dangerous?
versus: What does the evidence say about the risks of this medication?
The wording of the question can determine what evidence gets looked for. Once AI gives us a coherent explanation, we forget to ask whether the thing being explained is true.
What to do when you see “People Also Ask”
When Google shows you a People Also Ask question, don't assume Google has established those questions and statements as true. Take the question apart.
If it says: Why did people stop liking Garth Brooks? remove the why, and ask, Did people stop liking Garth Brooks? Then look at the evidence.
If five Reddit users say Garth Brooks became unpopular, that doesn't establish that Garth Brooks became unpopular. If a newspaper reports that a lawsuit was filed against him, that establishes that a lawsuit was filed.
Check the premise first
Only then ask why.
I thought that solution might be as simple as changing the question. Instead of asking: Why did people stop liking Garth Brooks? I searched: Did people stop liking Garth Brooks? That didn't work. Google gave me the same answer, the same AI Overview about lawsuits, backlash and criticism.
I became more explicit:
Is there evidence that Garth Brooks has become less popular? Don't give me reasons why people might dislike him. First determine whether his popularity has actually declined.
This time the answer changed completely and in a good and useful manner: “No, there is no definitive commercial evidence that Garth Brooks' overall popularity has declined.” Google looked for evidence. and presented as facts, current ticket demand, tour performance, his successful Las Vegas residency and other measures of his current popularity. That's what I had wanted to know in the first place.
This taught me something very important about using AI. Don't assume AI will figure out the question you should have asked. If you ask why something happened, AI will start looking for an explanation. Even asking whether it happened may not be good enough. If what you want to know is whether a claim is true, say so explicitly.
Ask:
What evidence establishes this?
Is there evidence this happened?
Don't explain why until you have established whether it's true.
We're increasingly asking AI to think for us. We shouldn't assume it will stop us when we start by asking a questions loaded with bad assumptions. We shouldn't assume it will challenge the validity of a question, or distinguish between evidence that something happened, and people's opinions about why it happened.
AI doesn't have to invent a fact to mislead us. It can give us real facts, real controversies and real opinions—and arrange them into a convincing answer to a question whose premise was never established.
With AI, asking the right question is now more important than ever. Asking the right question means explicitly telling AI:
First establish if it's true. Only then tell me why.
Even one word can change the question
Google's better answer revealed something else. It said that while there was no definitive commercial evidence that Brooks' overall popularity had declined, his cultural visibility among younger audiences had declined. One possible reason is Brooks has his catalog off Spotify and Apple Music.
Did Google mean “liking” or “listening”?
Someone who doesn't encounter Garth Brooks on Spotify playlists isn't necessarily someone who stopped liking him. They may never have listened to him in the first place. If fewer young people listen to Garth Brooks because his music isn't available where they normally discover and consume music, that's different from saying people stopped liking Garth Brooks. Someone who doesn't encounter his music on Spotify hasn't necessarily rejected him. They may simply not encounter him.
Ironically, Brooks' decision to stay off the major streaming services wasn't about rejecting younger listeners. His long-standing objections to the economics of digital music have included concerns about protecting the value of albums, artists and songwriters. Whether that strategy was wise is another question.
Maybe Google really did mean liking. Maybe its systems detected enough searches and discussions about people disliking Brooks that this was the question generated. Or perhaps some of the underlying signal was really about declining listening, visibility or relevance and it became expressed as “Why did people stop liking Garth Brooks?”
We don't know. That's part of the problem. A single word changed the claim from something measurable—Are fewer people listening —into a claim about what millions of people supposedly think and feel about him. Then AI tried to explain why.
Before asking AI to explain something, make sure the words in the question describe what you actually want to know. “Listening,” “liking,” and “popular,” sound related.
Those are asking three different questions.