Why Is Google AI So Bad and Hurting the Web

Google's AI has been making waves in the tech world, but not for the right reasons. Its algorithms have been criticized for being overly aggressive in promoting its own services, such as Google Search and Google Maps, over others.

This biased approach has led to concerns about the health of the web as a whole, as smaller websites and businesses struggle to compete with Google's dominance. For example, a study found that Google's algorithm change in 2019 led to a 30% decline in traffic for small websites.

The impact is real, and many businesses have reported losses due to Google's biased search results. A survey found that 75% of small business owners felt that Google's algorithm changes had negatively affected their business.

If this caught your attention, see: Go Back to Google Images Search Instead of Google Lens

Google's AI Overviews have been giving incorrect, misleading, and even dangerous answers, sparking negative mainstream media coverage.

The issue is not limited to a few isolated examples, as Google claims. A few examples of Google's AI Overviews include advising to drink urine to pass kidney stones quickly.

Google's excuse that these are "extremely rare queries" is odd, considering 15% of queries are new, according to Google Marketing Live. This is a significant number, and it's not an excuse for ridiculous, dangerous, or inaccurate answers.

Google is blaming the query rather than taking responsibility for the mistakes. A Google spokesperson said the examples are "extremely rare queries and aren't representative of most people's experiences", but this doesn't address the issue.

Google's AI Overviews have severe, fundamental issues. Trust in Google Search is eroding, which could lead to people searching elsewhere and hurting performance for advertisers and organic traffic to websites.

Google AI's lack of contextual understanding is a major issue. It can't grasp deeper context, especially in nuanced queries, and relies heavily on keyword matching rather than nuanced reading.

This shallow understanding leads to confused or irrelevant results, where Google might mix up results about the fruit "apple" with information about the tech company. People have to click through several links before finding what they need, causing user frustration.

Here are some serious consequences of this issue:

To work around this issue, focus on clear, direct writing and use explicit headings and FAQs to signal relevance clearly.

Google AI's lack of contextual understanding is a major issue. It can't grasp deeper context, especially in nuanced queries, and relies heavily on keyword matching rather than nuanced reading.

This shallow understanding creates serious consequences, including content misinterpretation, user frustration, and SEO confusion. People have to click through several links before finding what they need, and marketers find it harder to align content with search intent.

The problem is that Google's systems don't truly read and understand meaning, just words. This distorts search experiences and makes it difficult for users to find relevant information.

To work around this issue, focus on clear, direct writing and use explicit headings and FAQs to signal relevance clearly. This can help users quickly find what they need and reduce the number of clicks required.

Additional reading: Search in Google Drive

Here are some specific actions you can take:

Google's AI has become increasingly complex, but it's also become less transparent. In the past, SEO experts could predict how to optimize content with clear signals like backlinks, on-page SEO, and mobile friendliness.

This complexity without communication hurts everyone, especially when traffic drops and diagnosing the problem becomes nearly impossible without clear guidelines.

Difficult troubleshooting is a major issue: if you're trying to figure out why your website's traffic is down, it's hard to know where to start without clear guidelines.

Misinformation is also a big problem: vague documentation leads to speculation, myths, and ineffective SEO practices circulating widely.

As a result, people are growing distrustful of investing time and money into SEO, because results seem governed by invisible forces.

The March 2024 Google Core Update is a case in point: it introduced major ranking volatility, but official statements remained cryptic, leaving marketers guessing about whether E-E-A-T principles had been devalued or simply reweighted.

In a world where search engines are essential infrastructure, the lack of accountability from Google's AI creates an uneven playing field.

Here are some of the specific challenges caused by Google's complex but opaque AI:

Google's AI has some major quality and performance issues.

Inconsistent application of quality standards is a big problem. Google claims to promote helpful content, but its AI-driven evaluation tools are still primitive and apply the same standards to vastly different types of websites.

Several factors contribute to this, including overgeneralized guidelines that treat a blog about recipes the same as a technical manual. This leads to genuine helpful websites losing traffic while thin, affiliate-heavy pages remain untouched.

The impact on SEO is severe, with recovery from penalties being unpredictable and creating a climate of uncertainty for anyone relying on organic traffic.

Here are some of the most ridiculous errors Google's AI has made:

These aren't just bugs – they're a glimpse into the core issue: AI doesn't understand context, sarcasm, or truth.

Publishers are being hurt by Google's AI Overviews, which use content taken directly from independent publishers, blogs, and websites without proper credit or compensation. This is a problem because it means that Google is essentially using the web's content to replace the people who created it.

Imagine starting a business, building a website, and working hard to create high-quality content, only to have Google come along and use that content to train its AI models. And then, when people search for answers, they get them directly in the search results without ever visiting your website.

Google's response to this criticism is that AI Overviews "sends traffic to a greater diversity of sites", but there's been no data to back this up. Site owners say their traffic tells a different story.

Here are the consequences of Google's actions:

This is a growing concern for anyone running a website, blog, or news outlet.

Google's AI-driven evaluation tools are still primitive, and it shows in their inconsistent application of quality standards. This is evident in the Helpful Content Update, where some genuinely helpful websites lost significant traffic while thin, affiliate-heavy pages remained untouched.

Several factors contribute to this inconsistency. Overgeneralized guidelines are one major issue, as AI applies the same standards to vastly different types of content, such as a blog about recipes and a technical manual, even though audience expectations differ.

Lagging adaptation is another problem, where Google's systems are slow to recognize improvements on websites, meaning recovery after an update can take months – even if problems were immediately fixed. This creates a climate of uncertainty for anyone relying on organic traffic.

Flawed detection mechanisms are also a major issue, where machine learning models often misinterpret long-form, detailed articles as keyword-stuffed spam if the structure isn’t “perfect” by AI standards.

Here are the key factors contributing to inconsistent quality standards:

These issues have severe consequences for SEO, making recovery from penalties unpredictable and creating a climate of uncertainty for anyone relying on organic traffic.

Google's AI has some major limitations that make it less reliable than you might think. For instance, it's not great with cultural nuances, often misfiring when search terms have different meanings in different regions.

A search for "football" in the US will give you American football results, but in the UK, you'll get soccer results. This leads to global SEO challenges, content localization issues, and bias against underrepresented languages.

Google's AI relies too heavily on patterns, often promoting outdated information and struggling to recognize when human circumstances change. This means old data dominates search results, and core algorithm updates rarely fix foundational misunderstandings.

Here are some examples of Google AI's limitations:

Google's AI also has a "black box" problem, where it's hard to understand how the AI came up with its answer or what sources it relied on. This makes it difficult to catch errors before they go live.

Google AI's performance takes a hit when it comes to emerging languages and cultures. It struggles to deliver accurate results, even in well-established languages like English.

For instance, searching for "football" in the US yields American football results, but in the UK, it's soccer. This cultural nuance is often lost on Google AI, leading to irrelevant results.

Companies expanding internationally face significant challenges due to Google AI's cultural blind spots. They must put in extra manual effort to ensure their content resonates with diverse audiences.

Here are some of the consequences of Google AI's poor performance in emerging languages and cultures:

Google's AI Overviews often rely too heavily on patterns in data, rather than using judgment to make decisions. This can lead to outdated information being promoted, even if it's no longer accurate.

Old data dominates search results because the AI assumes popular older data must still be relevant. This means that inaccurate content can stick around for a long time.

Google's machine learning models sometimes prioritize engagement metrics like clicks over accuracy or freshness. This can lead to a focus on sensational or attention-grabbing content, rather than providing the most accurate information.

Here are some common issues that can arise when Google's AI Overviews rely too heavily on data patterns:

As a result, SEO professionals have to account for Google's heavy dependence on historical data, even when it no longer matches reality. This requires a more nuanced approach to search engine optimization.

Google's AI tech is being held back by its product culture, which prioritizes building products without user input. Google's Gemini is a prime example of this, being described as a barebones AI with fewer capabilities than competitors and an overstuffed AI Studio that mixes consumer and developer use cases.

The lack of user involvement is evident in the way Google approaches product development. Raiza and the core NotebookLM team left Google after trying to engage users, only to be met with resistance from Google employees who didn't understand popular platforms like Discord.

To succeed, AI product development needs to be done in public with users, listening to and responding to user feedback on social media. This approach is not being followed at Google, where meetings are spent building products rather than discussing their development.

Google's use of "labs" and "experimental" labels is a sign of a lack of conviction to release new features to everyone. This approach is not only holding back Google's AI tech but also making it harder for users to get the products they need.

Here are the key takeaways from the article sections:

The recent launch of Gemini 2.5 was encouraging, with Sundar Pichai highlighting a desire to move quicker. However, it remains to be seen whether this will translate into meaningful changes in Google's product culture.

Google's Gemini and product culture are closely tied, and it's worth examining how they intersect. Google's product culture has been criticized for prioritizing internal processes over user needs.

Raiza and the core NotebookLM team left Google shortly after implementing innovative product culture changes. These changes included building with users, not for them, and focusing on shipping quickly.

Meetings at Google were spent talking about building, rather than actually building, which hindered progress. This is a stark contrast to the approach of building in public with users, listening to and responding to feedback on social media.

To succeed in AI product development, it's essential to build in public with users and listen to their feedback. This approach was exemplified by Raiza's persistence in creating a Discord community for NotebookLM despite initial pushback from Google colleagues.

Here are some key takeaways for building successful AI products:

Google's recent Gemini 2.5 launch was encouraging, as Sundar highlighted a desire to move quicker. This shift in focus may indicate a willingness to adopt more innovative product culture practices.

Google AI Overviews raises serious ethical concerns.

The system lacks transparency, making it impossible for users to verify the accuracy of the information provided.

Critics question who decides what gets shown as "the answer" and how users can trust a system they can't see into.

Researchers call this the "black box" problem, where there's no clear way to understand how the AI came up with its answer or what sources it relied on.

This makes it almost impossible to catch errors before they go live.

Google itself admits the tech needs more safeguards, but critics say these fixes are surface-level and don't solve the root problem.

Here are some of the key concerns:

These questions highlight the risks of relying on AI-generated content without proper oversight or accountability.

The tech just isn't ready for prime time. AI Overviews gives you information that's sometimes 60% or more inaccurate.

Google, the go-to source for factual information, has been trusted for over 20 years. People relied on Google search because it worked, it was consistent, dependable, and accurate.

Google could've spent a few billion running ads or campaigns explaining why large language models like ChatGPT aren’t reliable for fact-based search. But instead, Google chased the AI hype, hoping to keep shareholders happy.

These systems don’t “know” anything, they don’t verify facts, and they generate text based on patterns from the web. Even when those patterns are wrong or misleading, they'll still give you bad answers.

The “black box” problem makes it almost impossible to catch errors before they go live. There’s no clear way to understand how the AI came up with its answer, or what sources it relied on.

Even Google admits the tech needs more safeguards. The company has since limited AI Overviews on “high-risk” queries and made changes to avoid pulling content from Reddit and other unreliable sources.

Google's AI Overviews raises important questions about who decides what gets shown as "the answer." This creates a system where Google becomes a gatekeeper for information, deciding what users see first, how they see it, and who gets traffic.

The lack of transparency in AI-generated content is a major concern. Users can't see into or verify the system, making it difficult to trust the results.

There are three key issues that arise from this: who decides what gets shown, what happens when biased training data leads to biased results, and how can users trust a system they can't verify.