Google is its own worst enemy

Google is its own worst enemy

In its bid to push AI, it’s destroying its products and trying to kill the web too

Don't buy the AI hype

by<br>Paris Marx

May 20, 2026 ∙ 5 min read

Screenshot: YouTube/Google

Written by

Paris Marx

Paris Marx is a Canadian tech critic and host of the award-winning Tech Won’t Save Us podcast. He writes the Disconnect newsletter and is the author of Road to Nowhere: What Silicon Valley Gets Wrong about the Future of Transportation.

Editor’s Choice

Getting off US tech: a guide

Social media must be reined in

How I became a tech critic

Newsletter<br>Get all the latest posts delivered straight to your inbox.

Subscribe

Processing your application<br>Please check your inbox and click the link to confirm your subscription.<br>There was an error sending the email

When OpenAI launched ChatGPT in November 2022, the media and talking heads quickly adopted the company’s mantra that the world had changed. Generative AI was about to revolutionize everything, and that included presenting a challenge to some of the juggernauts that had previously defined the digital economy. None was more at risk than Google, or so they suggested.<br>Google’s power came from its dominance of search, and now that OpenAI had its own interface for people to find information, Google was supposedly in trouble. Microsoft’s Bing couldn’t steal significant market share from the king of search in the past, but expert commentators were certain that was about to change with ChatGPT. How wrong they were.<br>In November 2022, Google controlled 92% of the search market. By last month, it hovered around 90%. Not the decimation all those smart tech watchers predicted. That should have been obvious. As I argued at the time, those claiming to foresee a decline of Google had it all wrong: ChatGPT wasn’t a threat to Google; the release of the chatbot by a company outside the top echelons of the tech industry set it free.<br>ChatGPT sets Google’s AI ambitions free<br>AI tools will increase the power of tech’s incumbents<br>DisconnectParis Marx

Google had already been experimenting with chatbots, but was hesitant to release them to the world for fear of the public relations disaster that could follow from the responses they might output, and the scrutiny they would attract from regulators. Microsoft had seen that first hand when it released Tay in 2016, only for it to begin echoing the Nazi and racist sentiments of those engaging with it in less than 24 hours. But once OpenAI deployed its “hallucination” machine, the guardrails were off.

Disconnect relies on the support of readers like you. Become a paid subscriber today!

Become a subscriber

For years, Google had been working to make itself less of a directory and more of a destination in its own right. It didn’t just want to send people to different parts of the web; it wanted them to find the information they were looking for on Google itself. The goal was to get users to spend more time on Google’s website and look at more of its ads, even if that meant they were less likely to visit the websites and news publishers that had come to rely on Google for significant shares of their traffic.<br>Long before Google introduced AI Overviews, it had added Featured Snippets to its search engine. Those snippets took sections from external websites and displayed them for the user, meaning they might get the answer they were looking for without having to leave Google. It was already a pain point for many publishers, but Overviews took that to the next level.<br>Instead of just pulling from one source, Google began pulling from many and using its AI tools to output a few paragraphs bringing together tidbits from each of them. Just ignore how the prevalence for the AI system to output false answers — or hallucinations, in industry speak — meant Google was generating millions of inaccurate Overviews every hour. Its ambition was clear: it wanted more users to spend more time on Google and far less visiting other parts of the web.<br>Generative AI is a societal disaster<br>Governments are deluding themselves into believing investment justifies allowing AI to upend society<br>DisconnectParis Marx

Publishers and website owners began warning of Google Zero as they saw their traffic plummet, but Google didn’t care. It was free to take the next step in destroying their businesses all to serve its own business goals. The company also didn’t care that it was further decimating the kind of information environment a healthy democratic society needs, especially one flooded with AI-generated false information. It’s also one you would imagine a search engine or chatbot that’s supposed to be relaying accurate, up-to-date information relies on.<br>Google made its ambitions quite obvious in its I/O announcements in 2023. On Tuesday, it laid out the next stage of its plan. The company made it clear there will be no escape from its generative AI products as AI hype continues to grip Silicon Valley, despite...