Google AI Search: The Humanless Internet
When AI systems generate content that is consumed by other AI systems, humans become second-class entities and will have to deal with the negative implications | Edition #207
👋 Hi, Luiza Jarovsky here. Welcome to our 207th edition, now reaching 62,600+ subscribers in 168 countries. To upskill and advance your career:
AI Governance Training: Join my 4-week, 15-hour program in July
Learning Center: Receive free AI governance resources
Job Board: Find open roles in AI governance and privacy
Subscriber Forum: Participate in our daily discussions on AI
👉 A special thanks to hoggo, this edition's sponsor:
Tired of drowning in manual assessments? hoggo's AI-powered platform gives legal and privacy teams their time back with one-click vendor assessments, automatic monitoring, and seamless RoPA generation. Eliminate repetitive tasks, focus on strategic work, and reclaim 3,000+ hours. Get started for free and unlock 20% off all plans with code LUIZA20.
Google AI Search: The Humanless Internet
Last week, among the 100 things Google announced during its annual developer conference, one was the expansion of AI search capabilities.
In today's edition, I discuss some of the individual and social implications of the new technological era we are entering: the humanless internet, powered by AI search, large language models, and AI agents.
-
It's not news that Google has been aiming to merge search and AI.
As I've covered in the last few months, since the end of 2022, millions of people have been using AI chatbots as if they were search engines. Google had no other option if it wanted to remain relevant. Merging AI chatbots and search was an expected development.
When Google launched AI Overviews last year as an eventual “AI spotlight” at the top, adjacent to some search results, many realized that they were testing the waters before expanding it and fully merging its most lucrative product into AI.
Remember that the company has almost 90% of the global search market, which translates into around 8.5 billion searches per day. Search corresponds to more than 50% of the company's $348 billion+ annual revenue; a wrong move could cost the company its dominance.
-
Last week, during its developer conference, it announced that it was expanding AI search capabilities. The title of the article covering the topic is “AI in Search: Going beyond information to intelligence.”
Words matter, and Google is being clear about what's actually happening: search is not about information anymore. But what is it about then? What does intelligence really mean in this context?
A reminder that before AI features, a search engine was basically an indexer. It was constantly scanning the internet and organizing websites. After a user query, algorithms decided which websites would be shown and their ranking. The user had to navigate through the results and decide which one to click.
Something that most people have so far ignored, but which we'll soon start to miss, is the fact that the title, snippet, blue link, and content of each page indexed by the search engine are created by the provider of the website. And until recently (before the generative AI avalanche), they were mostly human-made.
The list of blue links and their descriptions, as impersonal as they might seem today, actually gave the internet a fresh human touch. These were diversified, heterogeneous results, written by humans (content providers), for humans (people looking for information).
As Google writes, this era has ended.
When a user prompts the ‘AI search engine,’ the result is a uniform (same language), centralized (same company), standardized (same model specs), monolithic (a single result), and ultra-processed AI-generated output.
It's the junk food of the internet: the organic human content eaten and regurgitated by an AI system.
It's probably faster and more convenient to have only one answer than a list of links to choose from. However, what is fast and convenient is often not the healthy option (such as fast food).
-
Google has also expanded its agentic capabilities in search by launching a new agentic checkout feature that buys on the user's behalf with Google Pay.
A reminder that a few weeks ago, I wrote about how PayPal and credit card companies are rushing to launch new agent AI features, so that shopping can be automated through agents.
As agentic capabilities grow and replace users in navigating the internet, buying, communicating, deciding, and so on, the content being served on the internet will be more and more tailored for and targeted at agents.
When AI systems generate the content that is consumed by other AI systems, humans become second-class entities and will have to deal with the negative implications. Why?