If you’re going to try Semrush, get 2 week instead of 1 free here

👋 What’s growing: AI’s appetite for the planet.

ChatGPT sucks up around 10-15x more electricity than a normal Google search, according to Goldman Sachs research. And as more people turn to AI for everything from quick look-ups to hook-up advice, things aren’t getting any better. As it is, Google searches alone cause about 1.7% of global emissions each year. Multiply that by 10-15 and that number gets worrying.

Treat AI like turning off the lights when you’re not in a room.

AI CONTENT

You can rank with AI-generated content. But should you?

That’s a firm maybe.

Ask a computer to write you an article. It takes seconds, and it’s sometimes better than what you’ve been getting from pricey freelancers. 

It sounds like BS, but the internet is full of people who swear they’re ranking with articles spit out by ChatGPT.

Yes, you can rank with AI-generated content.

But should you?

It’s cheap, it’s easy, and it’s popular. But it’s not risk-free.

How We Got Here

Those of us who measure our SEO experience in decades remember the rise of content spinners, the spiritual predecessor of generative AI. Cheap content agencies were built around these. You’d pay a few bucks for a post, and they spun up 300ish words that almost made sense together. 

The internet was flooded. Some of those $5 blog posts even ranked.

And then Google updated their algorithms. Bye-bye, spun content, and good riddance. Enter the reign of well-written content. Until…

Now, almost anyone can spend a couple of minutes fiddling with an AI and come up with an article that sounds better than the average person can write. 

But whether it actually does it’s job, i.e. pleasing both Google and users, depends largely on you.

E-E-A-T is what matters

It all comes back to the SEO standby E-E-A-T: experience, expertise, authority, and trustworthiness, i.e.

• Insights driven from experience

• Expertise in your topics of choice

• Domain and page authority

• Accurate content

👉Learn how to shine with E-E-A-T here.

The most important thing your content needs is this, though — a reason to exist beyond driving traffic.

If you know why you’re creating a piece of content, who it’s for, and what they want out of it, it doesn’t really matter if it’s written by a computer or a person.

💡 Of course, smart keyword selection and SEO-optimized content are big factors here too.

There are risks involved with AI usage

There’s a reason that most AIs have a warning about accuracy on the front page. Think of LLMs as overeager people-pleasers. They’ll try really, really hard to give you what you asked for, even if it means inventing some facts. 

But there are other downsides that almost nobody talks about. 

Estimates from a popularly quoted 2024 Amazon study say that 57% of the whole internet is AI-generated. Some experts think that figure is higher. (Ever heard the dead internet theory?). Here’s the problem. As AI content takes over, AI increasingly trains itself on AI-generated content, which, according to a study in Nature, makes it really dumb.

Somebody has to put more human-created stuff in there, or the machine brain basically eats itself like an ouroboros.

Plus, as explained in the intro, data centers are particularly energy-intensive.

Unless we put some work in of our own and by limit it’s use where possible, AI is not going to work out for any of us.

How to generate AI content that ranks

The people who win in the age of AI will be the ones who figure out how to use artificial brains to maximize their human advantages.

Top content teams use AI, it’s a given. But here’s how they intervene to make it work.

Unless you get involved, your content won’t quite get “there”.

💡 Want an AI bot that gets SEO too? Try ContentShakeAI and make content that ranks 12x faster.

CULTURE

Give an AI Legs, Sure. But a Face?

In November last year, Kim K unveiled her “new friend” on X, but he was a far cry from the people you usually see on the red carpet. This one was a Tesla robot named Optimus. 

Kim called him “cute” as she showed off how he can blow kisses from his feature-less face and play Rock, Paper, Scissors. He can do far more though, like unpack your groceries and assemble your IKEA furniture, among other things.

As long as you have $20,000.

Shaped by the Uncanny Valley theory

Standing at 5’8” and weighing 125 pounds, Optimus is more or less built like a human — but he doesn’t look like one.

Tesla designed the bot to look like, well, a bot. That’s because of a theory coined by the Japanese roboticist Masahiro Mori in the 70s that says we feel disdain for non-humans that look kind of human. It’s called “Uncanny Valley”.

For a visual understanding, think of the characters on The Polar Express.

They rode the line between cartoon and human too closely, and it was a little disturbing. Early iterations of Shrek’s Princess Fiona caused feelings of anxiety in kids because she was too life-like.

Why though?

“Why were we equipped with this eerie sensation? Is it essential for human beings?” Mori mused in an essay, over 50 years ago. “I have not yet considered these questions deeply, but I have no doubt it is an integral part of our instinct for self-preservation.”

DATA POINT

Gen AI is driving traffic — but not conversions (yet)

AI-assisted shoppers are flooding in. But they’re not pulling the trigger.

According to Adobe’s 2025 survey, 55% of consumers now use generative AI to research products, and 43% use it to hunt deals. AI-driven traffic is up 1,300% year-over-year between Nov–Dec. Engagement metrics are strong: more pages viewed, lower bounce rates. But conversion rates lag — 9% lower than non-AI traffic.

Why? A few working theories

• AI-assisted shopping is early-stage
  These visitors are likely still in the top of funnel (ToFU) phase. Gen AI makes it easy to explore, but most people still revert to familiar habits when it’s time to purchase.

• Decision fatigue
  AI shoppers browse ~12% more pages per session. Could indicate overchoice — a UX issue, not an intent issue.

• Trust still matters
  AI can suggest. But it can’t guarantee. When money’s involved, trust cues — reviews, guarantees, social proof — carry more weight than machine summaries.

But should you be optimizing for it? 🤔

First consideration: We’re looking at a 1300% increase in traffic for a 9% decrease in conversion over the past year. Seems worth it. Second, conversions from AI traffic have already increased since mid-2024. 

If you want to stay ahead of the curve, here are a few tips on how to optimize AI traffic conversion. 👇 

1. Track pages that are getting AI traffic

Go to your analytics tool (like Google Analytics) and look out for these in your referral reports:

Next, create a segment or alert to track visits from these domains.

💡Alternatively, you can use Otterly.ai to track all AI mentions of your brand along with the URLs it links to. 

2. Optimize for AI traffic

You’ve identified the pages that attract AI traffic. Now:

• Implement top-of-funnel strategies. Offer assets for further convincing, e.g. competitor comparisons. 

• Instill trust. Make sure there is social proof of some kind, be it customer reviews or TrustPilot scores. 

• Address decision fatigue. Make top options prominent, or use pop-up quizzes like: “Not sure what you need? Answer 2 questions →” to guide them to the right place.

• Retarget. You should already be doing this, but any page attracting top funnel traffic should have lead magnets to capture email addresses for retargeting. 

💡You can leverage the Semrush Topic Research tool to brainstorm ideas for your lead magnet.

Bottom line

Gen AI is shaping how people find you. It’s not yet how they choose you. Ultimately, you want to give the ‘just looking’ crowd a reason to choose you. Even if AI traffic doesn’t convert on the spot, you want to be the business they return to when they’re ready to buy.

This week’s issue was written by Amy Hawthorne and edited by Catherine Solbrig.