Google Search Is Broken

Google Search is Broken, Not in the obvious sense. You can still check the weather, find a restaurant, look up a football score or discover the name of an actor you recognise from somewhere.

The trouble starts when you actually need Google to search.

Try finding an old forum post. A photograph of something slightly unusual. An honest review written by somebody who has owned a product for more than a week. The exact sentence you vaguely remember reading three years ago.

Suddenly you are changing the wording, adding quotation marks, excluding terms, opening five tabs and eventually typing the word Reddit at the end.

For a product that became famous by making the internet easy to search, that is a strange place to end up.

Search for a Baby Peacock

A wonderfully stupid example explains the problem better than almost anything else.

Search Google Images for a baby peacock.

In 2024, people noticed that prominent results for the search included a remarkable number of AI-generated birds. Some looked like tiny adult peacocks, complete with extravagant feathers that real peachicks simply do not have. One widely shared screenshot showed AI images dominating the results.

This was not somebody asking Google for fantasy art.

They were asking what an animal looks like.

One workaround became popular: search for baby peacock before:2022.

Google officially supports the before: operator, so the trick pushes the search back to material published before AI image generators became widespread. It can remove much of the synthetic material, although it also removes every legitimate photograph published after that date.

It works.

But think about how ridiculous that is.

We are learning little commands to make one of the most advanced search engines ever built show us a reliable photograph of a bird.

The Internet Learned How to Please Google

AI did not create this problem.

Long before ChatGPT, websites had learned that appearing on Google could be enormously valuable. So an entire industry developed around understanding what Google wanted.

Sometimes that improved websites.

Sometimes it produced articles that feel as though nobody expected a human being to read them.

You know the page.

You search:

How long can boiled eggs stay in the fridge?

The article begins by explaining that eggs have been eaten by humans for thousands of years.

Then comes a stock photograph.

Then an advertisement.

Then a paragraph about the nutritional benefits of eggs.

Another advertisement.

A heading called Understanding Boiled Eggs.

Eventually, somewhere beneath all this, you find the sentence you wanted.

Google itself has acknowledged the scale of this problem. In 2024, it changed its ranking systems specifically to reduce low-quality and unoriginal pages, including content that seemed designed primarily for search engines rather than people. After the rollout, Google said those changes had reduced low-quality, unoriginal results by about 45 percent.

That is a substantial improvement.

It is also an extraordinary admission.

Search became important enough that large parts of the internet started writing for the machine that was supposed to help us find the internet.

Then We Started Adding Reddit

This may be the clearest sign that something has changed.

People now search things like:

best office chair reddit

is this hotel actually good reddit

washing machine making clicking sound reddit

best neighbourhood to stay in Tokyo reddit

Why add Reddit?

Not because strangers on Reddit are inherently reliable. Spend five minutes there and that theory disappears quickly.

The appeal is that the answer often sounds like it came from somebody who actually did the thing.

A polished article might tell you:

The 12 Best Office Chairs for Ultimate Comfort in 2026

A Reddit user might say:

“I’ve owned this chair for four years. The back is great. The armrests drive me insane.”

The second answer is messy, incomplete and possibly wrong.

It also sounds human.

That increasingly matters.

When every product review might contain affiliate links, every page might have been written to rank, and enormous amounts of content can now be produced automatically, the question is no longer simply:

Is this the answer?

It is:

Why was this answer created?

A badly written comment from a stranger can sometimes feel more trustworthy than a beautifully formatted article because the stranger appears to have less reason to manipulate you.

That is not necessarily rational.

But it explains why adding Reddit has become its own search technique.

Google Search is broken in a different way

Google itself is also changing what search is supposed to do.

AI Overviews now place generated answers above traditional links for many searches. Google describes them as a way to give users a quick summary while still providing links to explore the web. In 2026, Google pushed this further, adding conversational follow-up questions and deeper AI-driven search features.

Sometimes this is excellent.

Ask a simple question and you may get what you need without opening anything.

But there is an important difference between these two ideas:

Help me find the answer.

and

Give me the answer.

Google became Google by becoming extraordinarily good at the first one.

It is increasingly interested in the second.

That creates a strange situation. The search engine is sitting between the reader and the websites it is searching, summarising the information before the reader ever reaches the source.

For quick questions, that may be exactly what people want.

For anything complicated, obscure or debatable, I am not sure that more distance between us and the original information is necessarily an improvement.

The Web Underneath Google Is Broken Too

It would be easy to blame Google for all of this.

That would be too convenient.

Google is searching a very different internet from the one it indexed twenty years ago.

Useful conversations moved into Facebook groups, Discord servers and other semi-closed spaces. Affiliate marketing created thousands of near-identical recommendation pages. SEO encouraged websites to chase the same valuable searches. Generative AI made producing vast quantities of passable text and imagery extremely cheap.

Google now has to search an internet full of material created specifically to influence Google.

That is a nasty loop.

And Google clearly knows it. Its own search policies now explicitly target “scaled content abuse,” including large amounts of low-value material produced mainly to manipulate rankings, regardless of whether humans or automation created it.

So perhaps Google Search did not break by itself.

Perhaps the web and the search engine slowly trained each other into this.

Google Search Still Works. That Is Almost the Problem.

Google remains extraordinarily useful.

That is why the deterioration is frustrating.

We remember what it feels like when it works perfectly.

You type half a thought into a box and somehow the exact obscure page appears.

No tutorial required.

No site: command.

No before:2022.

No adding Reddit.

No mentally separating real photographs from generated ones.

No scrolling past a machine-written answer to find the source underneath.

Google still has all the machinery of the world’s great search engine.

What feels broken is the simple relationship we used to have with it.

You asked.

It found.

Now, increasingly, you ask, filter, inspect, doubt, rewrite and try again.

Google Search is broken not because it can no longer find the internet.

It is broken because finding something you can actually trust has become part of the search.

Read Next: Artificial Neurons Brain Cells Can Now Communicate.

The 5 Best AI Chatbots for Building a Website for Free

What the iPod Nano Reveals About How We Listen Now