Monday, May 25, 2026

AI SAVED US OR AI KILLED US — EITHER WAY, YOU'RE LIVING THROUGH THE DECIDING CHAPTER

 

AI SAVED US OR AI KILLED US 

EITHER WAY, YOU'RE LIVING THROUGH THE DECIDING CHAPTER

A dispatch from the front lines of the information revolution

In the Beginning, There Was Chaos (And It Was Glorious)

Cast your mind back, if you dare, to the early 1990s. The internet existed, technically. Finding anything on it was roughly equivalent to being dropped blindfolded into the Library of Congress, handed a flashlight with dying batteries, and told to locate a specific sticky note. Good luck. Godspeed.

The first "search engine" was called Archie — built in 1990 by a McGill University grad student who, bless his heart, just wanted to find files faster. Archie didn't read pages. It didn't understand questions. It catalogued file names. That's it. You had to know the exact name of the thing you were looking for, which is a bit like calling a librarian and saying, "Yes, hello, I'd like the book. The one with words in it."

Then came Veronica and Jughead — yes, named after the Archie comics, because apparently 1993 was a time when computer scientists had both a sense of humor and absolutely zero marketing budget. These searched something called the "Gopher protocol," which sounds like a rejected Disney sidekick but was actually a menu-based file system. Revolutionary. Briefly.

The point is: we were all fumbling around in a digital wilderness with hand-drawn maps, and we thought it was perfectly normal.

The Phone Book Phase: Yahoo! and the Human Editors

Then the World Wide Web arrived, and everything exploded — in the best and most chaotic way possible.

Jerry Yang and David Filo, two Stanford students with too much time and too much ambition, decided the solution to internet chaos was... a phone book. Their creation, originally called "Jerry and David's Guide to the World Wide Web" (a name that absolutely screams billion-dollar valuation), became Yahoo!

Here's the beautiful absurdity of it: actual human beings sat in actual rooms and manually clicked on websites, then filed them into folders. Computers > Software > Games. Entertainment > Music > Bands. It was the internet organized by a very enthusiastic team of librarians with carpal tunnel syndrome.

It worked — until the web grew faster than any army of humans could possibly categorize it. You can't hire enough people to manually sort the entire internet. The internet laughs at your staffing budget.

The Crawler Wars and the Dark Age of Keyword Stuffing

Automation took over. Engines like WebCrawler, AltaVista, and Lycos sent digital "spiders" crawling across the web, indexing every word on every page. This was genuinely revolutionary. It was also immediately, catastrophically abused.

The ranking logic was simple: the more times a word appears on your page, the more relevant your page must be. Simple. Logical. Completely, hilariously wrong.

Within months, the internet was flooded with pages that looked like this:

"Welcome to our vintage cars vintage cars vintage cars website about vintage cars. Are you looking for vintage cars? We have vintage cars. Vintage. Cars."

Written in white text on a white background, invisible to humans but irresistible to search crawlers. The mid-1990s search experience was essentially a spam festival dressed up as an information tool. You'd search for a recipe and get seventeen pages about casinos. You'd look up a historical event and find a suspiciously enthusiastic site about mortgage refinancing.

It was, in a word, a mess. In two words: a delightful mess. We didn't know any better, and we loved it anyway.

Two Guys in a Dorm Room Change Everything

In 1996, two Stanford PhD students named Larry Page and Sergey Brin had a thought that seems obvious in retrospect but was genuinely brilliant at the time: what if we ranked pages the way academics rank research papers?

The most important academic papers aren't the ones that repeat their title the most. They're the ones that other respected papers cite. Credibility flows through references.

Their algorithm, PageRank, treated every hyperlink as a vote. If trusted, authoritative websites linked to your page, your page climbed the rankings. Spam sites linking to each other in dark corners of the web? Those votes counted for far less.

Paired with a homepage so clean it looked like someone had accidentally deleted everything, Google launched in 1998 and felt, to everyone who used it, like someone had finally turned the lights on in that dark library. Results were relevant. Pages loaded fast. The little search box just worked.

By 2000, the search wars were essentially over. Google won. Everyone else went home.

The Golden Age of Blue Links (And Its Hidden Flaw)

For roughly two decades, the model was settled and comfortable:

Copy
You type keywords → Google finds pages → You get a list of blue links
→ You click → You read → You synthesize → You (eventually) find your answer

It was magnificent. It was also, if you think about it, slightly insane.

You, the human, were doing a enormous amount of the intellectual work. Google was the world's greatest librarian — phenomenal at finding the shelf, completely uninterested in actually reading the book for you. You still had to click through five articles, cross-reference three of them, ignore the one that was clearly written by someone's conspiracy-theory uncle, and eventually piece together your own answer.

And then there was the volume problem. Ask Google something even mildly complex and it returns 437,000,000 results in 0.43 seconds. Congratulations. You now have more information than any human being in history has ever had access to, and absolutely no idea what to do with it. The information age gave us a firehose when most of us just wanted a glass of water.

Enter the AI: The Chef Has Arrived

Here's where things get genuinely interesting — and genuinely strange.

AI-powered search doesn't hand you ingredients. It cooks the meal.

Ask an old search engine, "What are the pros and cons of three different school funding models across four states?" and you get forty-seven links. Enjoy your afternoon.

Ask a modern AI search tool the same question, and it reads those forty-seven links in real time, extracts the relevant data, builds a comparative analysis, formats it clearly, and hands it back to you in thirty seconds. It understands that when you ask "that movie with the guy and the thing in the 80s," you probably mean Back to the Future. It holds context across a conversation, remembers what you asked three questions ago, and adapts to your specific angle of curiosity.

It's less like a card catalog and more like having a very well-read research assistant who never sleeps, never gets annoyed, and never judges you for asking the same question four different ways.

The evolution, in clean terms:

EraPrimary MethodRepresentative ToolsThe Catch
1990–1993File name indexingArchie, VeronicaCouldn't look inside anything
1994–1995Human directoriesYahoo! DirectoryCouldn't scale with internet growth
1995–1997Keyword matchingAltaVista, LycosDestroyed by spam and stuffing
1998–2000sLink-analysis algorithmsGoogleStill made you do the synthesis
2012–2020Knowledge graphs + NLPGoogle Knowledge GraphBetter, but still mostly retrieval
2022–PresentLLM synthesis + live webAI-powered searchThe paradox problem (see below)

The Paradox Nobody Has Solved Yet

Here's the plot twist that nobody in Silicon Valley has cleanly resolved, and it's a genuinely fascinating problem:

AI gives you the answer. So you never visit the website. So the website loses traffic. So the writer stops getting paid. So the writer stops writing. So the AI runs out of new information to learn from. So the AI gets dumber. So you get worse answers.

It's an ouroboros — the snake eating its own tail, dressed in a hoodie and a $50 billion valuation.

The entire web was built on a beautiful symbiotic deal: "Give us your content to index, and we'll send humans to your door." AI search quietly tears up that contract. It's not malicious. It's just the logical conclusion of making search too good.

The industry knows this. Nobody has fixed it yet. It sits comfortably alongside other unresolved tech questions like "who owns your data" and "will AI kill us all" — the latter of which, we should note, has been the subject of serious academic debate and approximately ten thousand Reddit threads, with no consensus in sight. 😄

Not Everyone Searches the Same Way — And That's the Whole Point

Here's something worth sitting with: we never all used Google the same way.

Your grandmother used it to type full sentences in the search bar like she was writing a letter. "Dear Google, I am looking for a good recipe for apple pie, please and thank you." Your teenager uses it to type three words and somehow finds exactly what they want in four seconds. A researcher uses Boolean operators and date filters. A journalist uses it to verify sources. A small business owner uses it to spy on competitors.

Same tool. Completely different experiences. Wildly different results.

AI search will be no different. Some people will use it as a glorified autocomplete. Others will use it to run research that would have taken a team of analysts a week. Some will trust it completely. Others will treat every output with the healthy skepticism of someone who has been burned by a confident-sounding wrong answer before — and they'll be right to do so.

The technology doesn't determine the outcome. The human using it does. It always has.

A Note for the Younger Souls in the Room

If you're young enough that you've never had to use a card catalog, never had to call a library to ask a reference question, never had to go to an encyclopedia to look something up — pay attention, because you are living through something genuinely historic.

You will be able to tell people, with complete accuracy: "I was there when it changed."

You were there when search went from file names to full sentences. You were there when a clean white homepage dethroned an entire industry. You were there when AI stopped pointing at the library and started reading it for you.

Whether that story ends with "AI saved us" or "AI made things really complicated for a while before we figured it out" or "AI killed us" — well, that last one seems a bit dramatic, but the jury is technically still out — you get to be part of writing that ending.

And that is genuinely, thrillingly, wonderfully cool.

The Most Important Search You Can Do Right Now

Here's the thing about transformative technology: it doesn't wait for policy to catch up. The internet didn't. Social media didn't. AI certainly isn't.

But that doesn't mean policy is irrelevant. It means timing matters enormously.

The most important search you can run right now isn't on Google or any AI tool. It's finding out who represents you — your school board, your state legislature, your congressional representative — and making sure they hear from you. Loudly. Clearly. Repeatedly.

You want AI in education, not AI instead of education. You want tools that expand human capability, not replace human judgment. You want the next generation learning with these tools, not being left behind because the adults in charge were too busy arguing about whether to be afraid of them.

The people making those decisions right now are, statistically speaking, not the heaviest users of AI search tools. They need to hear from the people who are.

So go ahead. Look up your representatives. Use whatever search tool you like — Google, AI, a very well-organized phone book if that's your preference.

Just make sure you actually send the message.

Great times to be alive. Greater times to be paying attention.


Sources & Citations

🔍 History of Search Engines

These foundational sources trace the full arc from Archie to the AI era.

🎓 PageRank & The Google Algorithm

🤖 AI Search vs. Traditional Search — The Paradox

📖 Recommended Future Reading

These articles and resources will take you deeper into every dimension of the search evolution story.

🏛️ Deep History

TitleWhy Read ItWhere to Find It
"As We May Think" — Vannevar Bush (1945)The visionary essay that predicted hyperlinks, search, and the internet — written 50 years before the web existedThe Atlantic Archives
The Innovators — Walter IsaacsonThe definitive narrative history of the digital revolution, including the people behind early searchMajor bookstores / libraries
How Google Works — Eric Schmidt & Jonathan RosenbergAn insider account of Google's rise and the philosophy behind its search dominanceMajor bookstores / libraries

🔬 AI & The Future of Search

TitleWhy Read ItWhere to Find It
"The End of Search As We Know It"Explores how LLMs are fundamentally restructuring the information retrieval modelSearch: Wired Magazine
"Perplexity, ChatGPT Search, and the New Search Wars"Covers the competitive landscape of AI-native search tools challenging GoogleSearch: The Verge / TechCrunch
"Retrieval-Augmented Generation (RAG) Explained"The technical backbone of how AI search tools combine live web data with language modelsSearch: Towards Data Science
"Google's Search Generative Experience (SGE)"Google's own documentation and analysis of its AI search integrationGoogle Search Labs blog

⚖️ Policy, Education & AI Governance

TitleWhy Read ItWhere to Find It
"AI in Education: A Policy Roadmap" — UNESCOThe global framework for integrating AI into schools responsiblyUNESCO.org
"The AI Act Explained" — European ParliamentHow the EU is legislating AI — the most comprehensive regulatory framework currently in forceeuroparl.europa.eu
"Teaching AI Literacy in K-12" — MIT Media LabPractical frameworks for how schools can build AI understanding from the ground upmedia.mit.edu
"Who Regulates AI in America?" — Brookings InstitutionA clear-eyed analysis of the current U.S. regulatory landscape and its gapsbrookings.edu

🌐 Content, Creators & The Ecosystem

TitleWhy Read ItWhere to Find It
"The Death of the Click: How AI is Killing Web Traffic"Tracks the measurable decline in referral traffic from search as AI answers replace link clicksSearch: Nieman Lab
"Who Pays for the Internet?"A philosophical and economic examination of what happens when the ad-supported web model breaks downSearch: Columbia Journalism Review

The best time to start reading was yesterday. The second best time is right now — preferably using whichever search tool you trust most. 📡




at