Google Gave You 10 Answers. AI Gives You 1. Nobody Is Checking.

Let's talk about permission. Not the legal kind. The quiet kind. The permission we hand over every time we ask a question and accept the first answer that comes back.

Here is what struck me this week. People are still using artificial intelligence the way they used Google. Same reflex, same posture, type a thing and take what comes out. Which honestly makes sense, that is the muscle memory we all built. But the thing on the other end of that box changed completely, and almost nobody adjusted. Most people have not wrapped their minds around where AI actually is right now, let alone where it is going.

So let me walk you through it the long way, because I lived the whole arc.

I was born in 69, so I remember waiting

There was a time when you just had to wait for things. Life had more human in it, because there was no other option.

My mom had an antique payphone in the kitchen. A real one. Took dimes, nickels, and quarters, all three. It worked like a normal phone, and the bottom of it was basically a piggy bank, so when people came over it was the whole conversation. I was a kid, so of course I got into that thing, jimmied the lock, and stole the change. Then I could not get it locked back up, because when you get something open by being fancy, you almost never re-create the fancy going the other way. So I used wood glue. I am not sure I ever told them.

On the block I grew up on in a small town in New Mexico, seven houses shared one party line. If you wanted to make a call, you waited for your neighbor to be done. Then the lines got split. Then call waiting showed up, that little beep telling you somebody else was trying to get through, and that felt like science fiction. Then answering machines, and you could actually record a human voice on a cassette tape. Then microcassette. Then digital.

Every one of those was a miracle at the time. And every one of them shaved a little more waiting out of daily life.

The card catalog made you earn it

Here is the part I want you to actually picture, because this is the whole point of today.

Say you needed to look something up. Smallpox, for a term paper. First stop was maybe the encyclopedia set your parents bought, which was always missing two volumes and nobody knew where they went, especially if you had siblings.

So you went to the library. And in the library there was this beautiful heavy box, walnut or oak, I never knew which, with drawers that slid out smooth. Inside were cards, maybe three inches by five. You went to S. You found smallpox. You found the books tied to that subject. Then you took one of those little golf pencils and a scrap of paper and you wrote down where the book lived. Then you walked into the stacks and went and found it.

And sometimes you got there and the shelf was empty. Because thirty other kids in your high school got the same assignment and you waited. In my hometown there was not a second library to go try. That was it. Good luck.

Getting an answer used to cost you time, gas, and effort. That cost was not a bug. That cost was the thinking.

You compared sources because you had to hold three books in your hands. You judged which one was better because you physically had them side by side. Nobody taught you critical thinking as a subject. The process taught it to you by force.

Then the internet showed up and gave us ten answers

Late nineties, early two thousands, the boom hits. Fiber goes in the ground. Bandwidth gets real. Computers can finally handle the traffic. Suddenly the whole thing collapses from a drive across town into a box on a screen.

And here is the detail people forget: artificial intelligence was already being worked on the entire time, quietly, going back to the fifties. Alan Turing and the people around him were imagining machine intelligence at human level or beyond while the rest of the world was still on party lines. It was never new. It was just not ready.

Meanwhile we had Ask Jeeves, Yahoo Search, AOL Search, Lycos, all of them. You typed a question and you got back resources. That is the critical word. Resources, not answers. It handed you a stack and said, here, you figure out which of these deserves your trust.

Google won that era. I am not even sure it was always the best product, but it was the most powerful and the most used search engine ever built, and Sergey Brin and Larry Page sold it better than anybody. And it is still very strong, though some of its grip is getting pulled away now that people search inside large language models instead.

The black hat era, and why it matters right now

Now, how did you get to be one of those recommended resources? That depended on a lot of variables, and plenty of people cheated.

They called it black hat SEO. The name comes from the old westerns, villain in the black hat, good guy in the white one. Black hat search engine optimization meant doing something unethical, or flat out against the rules, to trick the engine into pushing your page to the top of the SERPs, the search engine result pages.

The joke back then was that if you wanted to hide a dead body, put it on page two of Google, because nobody ever went to page two. Everything was page one. Funny enough, more people find themselves on page two now, because page one is so loaded with sponsored slots and clutter that the actual answer got pushed down.

Here is the trick that worked for a while, and I know this from being in a market where it mattered. Take a phrase like Santa Clarita real estate. In our local market that is brutally hard to rank for and expensive to buy. So back in the WordPress days you would write a normal blog post, and then underneath it you would paste that phrase five hundred or a thousand times in a row. Just stacked. Page one, position one. It worked, right up until it did not.

Because Google got smart. They refined the algorithm, which is just a fancy word for program, and they sent bots out to crawl sites and look at the architecture, the structure, and the stuff hidden behind the page that no human eye ever sees. They went hunting for exactly that gaming. The other engines learned it too, but Google figured it out best, or at least sold it best.

And that whole cleanup is what made their pay per click machine go ka-ching. That is the money. That is also how they ended up sitting on a mountain of human data, how we search, what we ask, how we phrase it, which is pure gold if you are in the business of selling things back to us. It is the same foundation the AI infrastructure got built on.

I want you to hold onto the black hat story, because we are living through the exact same chapter again right now, just with AI answers instead of blue links. People are already trying to game their way into being the source the model quotes. It will get cleaned up the same way it did before. The only question is how many people get fed junk in the meantime.

And now it does not give you resources. It gives you the answer.

Today you ask, who is the top real estate agent in Santa Clarita, and you do not get a list to sort through anymore. You get an ad block, a map three pack, and an answer composed by AI sitting right on top.

And a huge number of people are not even starting at Google. They go straight to ChatGPT, which is not Google at all, it is OpenAI. There is Anthropic with Claude. There is xAI with Grok. Google has its own. Multiple serious competitors on the field, which as a regular person I think is exactly where we want to be. Are they all quietly high fiving each other behind closed doors like attorneys comparing billable hours? I have no idea. I am not up at that level of the realm. I am still trying to get health insurance.

What I do notice is the way the danger gets marketed. They tell you how powerful and unpredictable and risky these models are, and then ask for regulation. And look, human beings do not want the church mouse product. We want the pit bull. Danger sells. But watch what regulation actually builds: a moat. Rules written at that altitude protect the companies already at the top and lock out the small shop with the better idea. The giants keep moving at full speed regardless, because good luck staffing a regulator with people who can walk into those buildings and understand what they are looking at. The people running those companies cannot fully explain why the thing does what it does either.

The doctor example that flipped my thinking

Here is something worth sitting with. The numbers people throw around for a human doctor landing a correct diagnosis are somewhere in the seventy to eighty percent range, depending on the type. Round numbers, but that is the neighborhood.

So you would assume you bolt AI onto the doctor and it jumps. It does not, or at least not the way you would expect. Because of bias. The doctor overrides. The human keeps pushing their read.

But take narrow AI, train it on cancer, feed it every image that was cancer and every image that was not, an amount of material no human could get through in a lifetime, and then hand it a scan. On its own it lands the read at a rate a generalist human is not going to match on that one task.

That is genuinely good news for people who need an accurate read. Nobody who reads scans is excited about a wrong diagnosis. Honestly that one may be less of a gut punch than the guy who drives a truck getting replaced by a robot that drives a truck. That one is going to hit harder and hit more people.

But notice what the story actually proves. A system trained hard on one narrow job can beat a human at that job. It does not prove that a confident paragraph on your screen about anything else is true.

The Johnny Walker Blue Label test

Here is the moment that made me want to record this.

When I was growing up, you are around the dinner table, it is eight thirty at night, somebody has a cocktail in their hand, and somebody says, I heard a term today, Blue Label, what does that even mean? And if nobody at that table knew that Johnny Walker Blue Label is their premium bottle, then it just sat there. Unanswered. There was nowhere to go. You wondered about it. Sometimes for years.

That problem is solved now. Completely. And that is good. I am not romanticizing ignorance.

But ask the second question. What did solving it cost?

Look at what already went. Most people driving around are running a map app, and that is fine, in Los Angeles that is the difference between a thirty minute drive and a three hour one. I mostly refuse to use it, because I worked all four bureaus with LAPD and I know where things are. Give me an address, I will find it. Give me an intersection, Sherman Way and Van Nuys, Sepulveda and Ventura, I will get you there from anywhere. But I am the exception, and honestly the tool is better at traffic than my memory is.

Phone numbers, same story. Do you have any memorized? Yours, to hand out. Maybe. That ability is just gone, and nobody mourned it.

Here is why answers are different from navigation. A map has a destination you either arrive at or you do not, so a wrong turn corrects itself. An answer does not self correct. An answer can be wrong, biased, incomplete, or shaped by whoever trained the thing, and if you never compare it to a second source, nobody ever finds out. Not you. Not the machine.

What we actually gave up

I know somebody is going to say Google was telling you what to think too, and fair enough, position one versus position four is a judgment call somebody made for you. But there is a real difference between a list and a verdict. A list still asks you to do something. It puts ten doors in front of you and makes you pick one. That tiny act of picking is where the thinking lived.

Now it is one door and a voice telling you it is the right one. And the people who treat AI as all knowing, all correct, and always current are going to accept whatever comes out of it. That is a problem. That is the whole problem.

And look, maybe the answer is me. Maybe you type in best Santa Clarita real estate agent and it says Connor MacIvor. Great. But maybe you were looking for something else, for a use case only you understand, and you deserved to actually go look. The machine is formulated to produce an answer for you. How that answer got tabulated, weighted, and dressed up in clean text is the part nobody sees. Is there bias in there? Almost certainly. There is bias in every human system it learned from.

So what do you actually do

I am not telling you to go back to the card catalog. I use these tools all day long, every day, in real businesses. I am not running out onto the freeway lighting myself on fire about it. I try to be the internal optimist. And I genuinely do not know if this ends in super abundance or the other thing. That pendulum can swing either way, and it is going to swing fast.

What I am telling you is smaller and completely doable:

• Ask it where it got that. Make it show sources. Then actually open one. That single habit puts the comparing step back into your day.
• Ask the same question two different ways, or in two different models, when the answer matters. Watch what changes. That is where the shaping shows up.
• Notice when you stopped wondering. Wondering is the muscle. If nothing has puzzled you in a week, you are being handed too much.
• If it involves your money, your health, or your house, verify it with a human being who is accountable for being wrong. Machines are not accountable. That is not a knock, that is a fact about what they are.
• If you own a business, understand that you now have to be findable by machines, not just by people. Real content, real specifics, clean structure, consistent identity everywhere your name appears. Same game as the honest sites played twenty years ago, just faster.

That is it. Use the tool. Keep the judgment. Those two things are allowed to coexist and most people are quietly dropping the second one.

We went from waiting days for an answer, to ten links and a choice, to one voice and no choice. Somewhere in that trade we handed over a permission slip nobody read. I think we should pull it back a little. Maybe a lot.

Be safe out there. I do appreciate you watching. I am Connor. Be well.