Search Engine Land was kind enough to publish my piece titled Surviving the search revolution: 7 lessons from losing weight with AI.

I’d suggested the title you see on this post, but for obvious reasons they didn’t go with it. 🙂 They also had to edit my original piece down for length. There were also some things (like the YouTube video) that couldn’t be put into that piece.

For the 3 1/2 people who read this blog, I figured I’d post my original, unedited version. Unlike the polished piece on SEL, this one has more of my Steve Liu long-windedness, but a few more details that may help some of you out there.

How I Lost 20 Pounds in a Month

No, this isn’t clickbait. It’s a story of how everything is about to change.

Lately in SEO circles, there’s been a debate: will AI “search” overtake Google search?

Some insist yes, Google’s days are over. This is usually followed by a pitch for them to do AIO for you (or GEO or AEO or whatever it’s being called this week).

And then there are some who confidently say news of Google’s demise is greatly exaggerated. Most of the talk about AI is just hype, and Google’s not going anywhere. Just do SEO the way you’ve always done and you’ll be fine.

I hear a lot of opinions on both sides, but there’s one thing I’ve never heard anyone articulate:

What exactly WILL “search” look like in a few years?

Déjà vu all over again

In many ways, these days remind me of 1995.

When I was an undergraduate at Rutgers in the early 1990s, “the Internet” was mainly e-mail, Usenet groups, and chat with fellow students and nerds across the country.

In April 1995, I did a research project for my MBA class called Business Uses of the Internet. I sent a questionnaire to the Webmaster of every domain whose brand I recognized (it was short enough to fit on less than one sheet of paper). I suppose technically that makes me one of history’s first e-mail spammers. But nearly everyone replied and was super gracious.

While their responses were fascinating, very few “called” what was coming.

During my final presentation I pulled up pizzahut.com which at the time looked like this:



I explained how one day in the not-too-distant future we’d all be ordering pizza from our computers. My professor and the entire class stared in disbelief (remember, we were still three months away from the launch of amazon.com). 

I feel we’re in the same boat with LLMs. Everyone knows AI will change the world but nobody quite knows how. It’s fun to watch babies giving press conferences and Will Smith eating spaghetti, but none of those are exactly practical. AI is still in its “parlor tricks” stage, much like the Pizza Hut site of 1995.

But I’m going to tell you a recent story from my own life of how AI is poised to disrupt a $300 billion industry. Just like with Pizza Hut in 1995, your jaw might drop today. But in 30 years, you’ll look back and laugh at how mundane it was.

Saddle up

Like many Americans, I’ve struggled with weight loss my whole life. In 2009 I ran a blog on the topic (some of you might remember; it ranked #1 for “wii exercise games”). It was about using a combination of Wii games and a popular diet plan to lose weight. I lost 40 pounds in 2009 (and gained it back), then 25 pounds in 2012 (and gained it back again).

This year I found myself at 238, my highest weight ever.

My daughter started fourth grade this year. Every fourth grader in the US can get a free Every Kid Outdoors pass from the federal government for free entry to any National Park.

We planned two trips: one to Utah and Arizona in February, and another to Montana and Wyoming in July (both were planned with the help of ChatGPT and ended up being among the best trips of my life, but that’s for another article).

One of our must-do activities: horseback riding in Glacier National Park. The concessionaire was very clear when I booked it in January: you must be 225 pounds or less, or you will NOT ride, and there will be NO refunds. No biggie, I thought. I had seven months to lose 13 pounds.

For the next few months, I watched what I ate and went to the gym a few times a week. By mid-June, my heart sank: I was still 231. Only 7 pounds lost in 7 months.

At a doctor’s appointment that week, I asked if he knew a secret to losing 6 pounds in 20 days. “Cut out the carbs,” he said. That was it.

When I got home, I decided to ask ChatGPT.

Here’s a copy of the actual chat so you can read it for yourself.

https://chatgpt.com/share/688b8819-3604-8002-b5de-7e0e89e79008 

Good basic information from AI, but nothing ground-breaking

ChatGPT gave me advice I’d heard all my life: create a calorie deficit, cut carbs, hydrate, walk, get good sleep.

But here’s where things changed.

I gave ChatGPT my weight, and it told me my exact maintenance calories and how big a deficit I’d need daily to start losing weight. I’d developed a rough idea of what “maintenance calories” meant for years, but I’d never had anyone calculate mine so precisely.

Intrigued, I started asking questions, including some questions I’d be hesitant to ask a human for fear it was a “dumb question”. Would working out add muscle weight? How many steps is 20 minutes on a treadmill? Why can’t I get rid of belly fat? ChatGPT answered each one clearly in a way that was tailored to me.

Finally, I asked it to put together a 20 day plan for me. It came through with a plan telling me the exact calories to eat and calories to burn each day. It threw in reminders to focus on high protein, high fiber, low carbs, and no sugar, alcohol, or refined carbs. It even made me a spreadsheet to track weight loss.

Helpful. But so far, nothing a good diet coach couldn’t do.

The real game changer

One morning at breakfast, I told it what I ate and asked it to calculate the calories. It did. Then I asked if it could track them daily rather than having me fill out a spreadsheet. It agreed to.

From then on, after every meal and even after every little impromptu snack, I’d pull out my phone and tell ChatGPT exactly what I ate. When I exercised, I’d report that too. No “points,” no “diet,” no coaches, just me opening my phone (which I had with me at all times anyway). 

Over time I started getting more creative in what I’d ask. 

While I was standing in the middle of Costco one day, I asked it for a shopping list of healthy snack ideas and for “permission” to eat a $1.50 hot dog. (It gave both to me).



There were a few times that my daughter wanted to eat dessert with me. If I’d been managing on my own I would have said absolutely not. But ChatGPT gave me advice about portion size that literally let me have my cake and eat it too.



There were a few times my weight shot up 3-4 pounds, causing me to panic. When I told ChatGPT it reassured me that it was just water retention from salty food.


There was one time I had an overripe banana and some strawberries. I asked it for a simple recipe for a banana smoothie that I could make with my portable blender. It came through.

At one point I got tired of typing in every piece of food I was eating, so I started sending it photos of my plate. It analyzed the photos and gave me an accurate breakdown of the calorie count.

In short, ChatGPT was the embodiment of the “butler” I mentioned in my article back in April 2024.

At the end of each day, it gave me a calorie rundown. Each morning, I weighed in.

I hit 225 in four days.

By day eight, I was 219.

On day 15 I was down to 217.

Travel test

On day 17 we flew to Montana. Travel would be dangerous for me. Unlimited plane snacks, airport lounges, and restaurant food were a recipe for disaster. Worse, I had no scale for the next 3 days so I’d be flying blind. So I asked it to tailor my diet for the rest of the trip.

Finally, the moment of truth. The time came for me to step on the scale.



Final weigh-in: 209.

Here’s what I looked like on the horse.



And here’s what I would have missed out on (in addition to $150) if I didn’t make it.

Welcome to the future

Why did AI succeed where my own attempts at dieting failed?

Like I said, this moment in history reminds me of 1995. At the time the Pizza Hut site was considered cutting edge. No one could imagine how the Web was going to change in just a few years.

Most people I observe using AI today are still in the “dabble” phase. Some use it like a search engine, others use it once in a while for a quick task or two.

But it’s when people start to incorporate AI into their daily lives that everything will change.

In this case, here’s why I think AI worked for me when so many diets, programs, and books didn’t:

It gave me solid, basic information, which is all I really needed. For something like dieting most of us don’t need cutting-edge research; we just need the fundamentals applied to our situation. AI happens to be good at the basics because the Web is saturated with basic content. We have Google to thank for that. For the last 27 years it’s conditioned us to create endless content for head terms while largely ignoring the vast long tail. 

It didn’t judge me. I was able to ask ChatGPT “dumb” questions, sometimes over and over again, without feeling embarrassed or self-conscious. That encouraged me to ask more questions. 

It talked directly to me. There wasn’t a lot that ChatGPT told me that I couldn’t find myself on a blog or forum. The difference is that I didn’t have to dig for it; it distilled exactly what I needed and filtered out the noise. 

It encouraged me. We’ve all experienced where ChatGPT can get a little too flattering and unctuous. But for something like losing weight, the encouragement and motivation completely worked.

It had endless patience. A human coach would’ve gotten sick of me by day two. ChatGPT never tired. 

I wasn’t programming the AI. The AI was programming me. In a good way. I now have experience and knowledge and tools I can use the next time I find my weight ballooning out of control. And I find myself instinctively being “good” in my day to day after this experience. 

SEO is dead. Long live SEO.

For 27 years Google has conditioned all of us in SEO to focus on the head. Create content such as “What is SEO”, rank in the top 3, and watch the money pour in.

Unless you’ve been living under a rock, you know that that hasn’t been the case for a while now. Search Engine Land has the #2 organic position for “SEO”. But that position doesn’t mean what it meant 10 years ago.

For years, we’ve looked at these SERP changes as nothing but a cash grab from Google. The reality? It might be Google trying to survive.

Google sees the writing on the wall. For searches with informational search intent, people just want the answer. If they don’t find it on Google, they’ll go to ChatGPT or another AI chatbot. 

Does that mean SEO is dead? No, as long as humans are looking for things, search engine optimization will exist. But the concept of a “search engine” is changing right in front of us.

If your whole business model is built on being the #1 Google result for a generic question that AI can now answer in two seconds, you’re on borrowed time. Those “searches” will be happening somewhere else.

In the past, SEO was about getting people to find your content.

Going forward, SEO will be about getting people to find your brand, making sure it’s in the conversation when your topic comes up.

The days of ranking generic content for a head term, sitting in the top spot, and driving huge traffic are almost over. When users ask AI about your topic, it will give a solid answer without necessarily citing you (and even if it does cite you, most people won’t click).

“Ranking” in Google is about positions, clicks, and impressions. “Ranking” in AI means something altogether different: Will AI recommend your brand over your competitors when it’s having a conversation with the user?

The opportunity now is to make sure your brand is top-of-mind for AI, so it mentions you naturally in the flow of conversation or in response to a follow-up question.

How to thrive in this new world

Ironically, there has never been more opportunity than today to be a search expert. 

Think about it. When someone has a “conversation” with AI and asks it something that’s beyond its training data, how does it learn more?

It searches.

And not just Google or Bing. AI will query every searchable place it can find, including Amazon, YouTube, Reddit, and your own site’s search engine.

It’s not looking for broad, generic info. It’s looking for detail and nuance, the kind of specific, in-the-weeds knowledge most SEOs have been ignoring for the last 27 years.

That’s where the opportunity is.

Search isn’t dead. It’s about to explode.

Here’s what you need to do to get ready:

1. Be an indispensable brand

Nobody searches “books” to find Amazon. Nobody searches “hamburgers” to find McDonald’s. Nobody searches “video game console” to find Nintendo.

These brands have executed so well that people skip the search step entirely. And their brands have reached the point where if Google or AI will look foolish for NOT recommending them when the subject comes up.

A lot of companies hiring SEO talent don’t have SEO problems, they have brand problems. They’re frustrating customers, alienating prospects, and burning employees. No amount of “SEO” will ever fix that, and their problems will only be compounded in the coming age if they don’t fix them. 

2. Stop ignoring the niches

Google has trained us for the last 27 years to chase the big head terms. If SEMrush shows a keyword with under “10” searches a month, we dismiss it as worthless.

In the AI era, those extreme long-tail queries are where the real opportunity is. In fact, AI will likely be the one conducting those searches on the user’s behalf.

The first step is to figure out where those underserved niches are, whether that’s by geography, by audience segment, or by a specific use case.

And no, the answer is not to hire an army of inexperienced freelancers, and it’s definitely not to have AI churn out watered-down content.

3. Publish deep knowledge and original insights

Go way beyond “FAQs”. Build rich knowledge bases and web content that take the expertise locked inside your team’s heads and internal documents and makes it public for AI, Google, and most importantly your customers to find.

Make sure it’s crawlable. Go deep. Provide insights and helpful information, especially around your brand and your unique value props, including things that have never been told before. The stuff you think is “too detailed” or “too obvious” is often exactly what people need and will be asking their AIs about.

Yes, you’re letting AI “learn” from your content. But you’re also training it to trust you. If AI sees your site as a reliable source of useful, relevant, and solid information, it’s going to remember you when someone asks for recommendations.

4. Share real stories, not soulless content

The “uncanny valley” used to refer to AI-generated work that just seemed “off”. These days, of course, generative AI can produce images and videos that look completely real.

But there’s a new “uncanny valley” taking its place, especially as generative AI starts to get used universally. You may have experienced it yourself. The words or images may be perfect, but there’s still something hollow and soulless about them.

Tell stories on your site. I’m not talking about fake-sounding “testimonials” or overly polished case studies. I’m talking authentic conversations with real people about your products and services. How they used your product. What went right (and wrong). What they were able to accomplish. What they learned. These are all details that humans will crave after AI gives them generic answers.

5. Build communities

Google and OpenAI are paying millions to access Reddit’s data for a reason: it’s still one of the last places on the Web where authentic conversations are happening.

A lot of those conversations aren’t exactly high quality. There’s an opportunity for brands to bring back their own moderated spaces where people can have real, troll-free discussions.

Well-moderated forums continue to be by far the most effective method of generating relevant, helpful, timely, and detailed content at scale. It always has been, but brands couldn’t justify the ROI. Since I wrote this article last year Reddit’s market cap has gone up $25 billion. It might be time to re-do the math on that one.

And don’t forget the “community” of reviewers, whether on your site or third-party sites like Trustpilot, Amazon, Yelp, or Glassdoor. Those are going to become even more valuable as AI leans on them for signals. That doesn’t mean faking reviews, it means doing #1 above.

6. Stop acting like a robot

For years, people in SEO (and the people hiring us) have thought of our job as “understand the tech.” That used to be true. Not anymore.

In the AI era, the technology matters less. Understanding humanity matters more.

Stop thinking “How do I get a million visitors from one keyword?” and start thinking “How do I be the brand that can answer a million different questions in ways that actually help people?”

7. Understand your users and how they’ll use AI

Here’s a good exercise. Just like I did with my diet plan, go to your favorite chatbot and try to accomplish a goal, the way your customers would.

Don’t just type a keyword into the box like you would on Google. Have a conversation. Try to get something done.

You’ll find some things AI gets uncannily right, some it gets embarrassingly wrong, and some it can’t do at all. Those last two are your opportunities.

Conclusion

Hall of Fame baseball player “Wee Willie” Keeler famously said “Keep your eye on the ball and hit ‘em where they ain’t”. 

I think that’s fitting advice for the coming AI age. 

The ball is and always has been your customer.

AI can do a lot of things well. Scarily well. If it ultimately helps your customers, embrace it. And figure out how to get your brand top-of-mind in conversations.

But there are some things it can’t do yet. There are some things it’ll never be able to do. Figure out what those are, and do them, starting with this list.