Information Consumption
Three weeks ago, on our way to Bob's Donuts (a must-visit if you're ever in SF), my friends and I found ourselves debating the ongoing war in Ukraine. Somehow, conversations about restaurant choice evolve into high-decibel debates about foreign affairs.
At some point, one of us made a profound yet seemingly unreliable claim about Ukrainian elections. What ensued was classic debate tradition; I'm sure you can relate. First came the "no way that's true." Then, everyone simultaneously reaching for their phones and searching the web. All in an effort to prove each other wrong.
Recently, I've been seeing those three bolded words more frequently and maybe you have too. Not from friends or peers but from AI tools: Claude, ChatGPT, Gemini, and Grok. When my question needs the context of recent events, a glowing, beautifully animated "Searching the Web" text appears. As that's shown, these tools crawl relevant websites, add them to context, and synthesize this information into a succinct answer.
It's awesome— incredibly powerful, productive, and undeniably convenient.
In the 1990s, traditional research, even something as simple as checking the Knicks vs. Pacers game (it was 108-125), shifted from books and articles to web searches. Similarly, today, manual browsing is gradually being replaced by AI-powered information retrieval.
Put simply, manual human searches will decrease over time. Instead, large language models and autonomous agents will increasingly handle information-seeking tasks, summarizing and presenting the results on our behalf.
This shift is profound for two main reasons.
Because the current search paradigm relies heavily on advertising. Google, YouTube, The New York Times, Washington Post, Quora all operate by attracting human attention and selling that attention to advertisers. But as human interaction on these platforms gets replaced by AI, ad revenue will inevitably decline. After all, GPT-4o isn't exactly tempted by the new MCrispy Strips at McDonalds (as opposed to me).
Instead of writing and publishing content primarily for human readers, people (alongside their AI assistants) will increasingly create web content in formats and styles optimized for easy understanding and retrieval by LLMs and automated agents.
We're on the brink of a paradigm shift in how information is consumed (and produced).