Howdy, marketer!

If you’ve opened LinkedIn, Twitter, or your inbox at any point over the last two years, you’ve seen the shift. 

We live in an era where anyone can generate a full campaign brief, ten variations of ad copy, three landing page concepts, and a set of hero images before finishing their morning coffee. 

The barrier to producing marketing material has effectively dropped to zero.

But as the volume of marketing output goes to infinity, the average quality and distinctiveness of that marketing is collapsing into a giant sea of mid. 

Everything sounds slightly polished, perfectly structured, vaguely energetic, and completely forgettable.

When any brand can press a button and output a clean, logical campaign, logic stops being a competitive advantage. When production is solved, creativity ceases to be about manufacturing assets. 

It becomes about taste, angle, positioning, and having the guts to say something that an algorithm trained on average historic data would never suggest.

Large language models operate on probability. 

They look at billions of parameters across the internet and predict the most likely next word. By definition, AI gives you the statistical mean of existing human thought.

In marketing, the statistical mean is the absolute worst place to operate. 

Growth never comes from doing what everyone else is doing slightly faster; it comes from finding an asymmetrical angle that hits a raw nerve with a very specific audience.

Strategists like Alex Murrell (in “The Age of Average”) and commentators on AI slop argue that over-reliance on algorithmic and generative tools pushes brands toward safe, homogenized messaging that blends into competitor ads.

When your ad looks like your competitor's ad, your customer acquisition cost (CAC) goes up. 

Your click-through rates plummet because users have developed an unconscious filter for "AI-flavored" polish. 

They scroll right past it. 

The true cost of relying on AI for your creative direction is campaign fatigue and decaying brand equity. 

Real creativity today means actively resisting the statistical mean. 

It means looking at what the model suggests, acknowledging that every competitor in your niche is getting that exact same suggestion, and intentionally turning 180 degrees in the opposite direction.

For a decade, marketing careers were built on execution speed. 

The team that could churn out twenty landing pages, run fifty multivariate ad tests, and publish five blog posts a week was usually the team that won on distribution. 

AI has completely flipped that dynamic. Execution is now commoditized.

Because execution is free, the value has shifted almost entirely to curation, positioning, and taste. 

Your role as a marketer is no longer to sit in front of a blank document trying to write a copy from scratch. Your role is to act as an executive director who knows what great looks like, knows what feels fake, and understands the underlying psychological triggers of your customer.

Growth operators like Casey Winters argue that AI can accelerate testing velocity, but it can’t surface non‑obvious frictions on its own - because models lack the contextual understanding of what buyers fear, how they talk, and which words feel “too corporate” in their world.

Creativity in an AI-dominated landscape requires you to be an aggressive editor. You use the tools to generate raw material, options, and structured frameworks, but you apply an uncompromising human filter over the output. 

You strip away the performative adjectives, inject real customer quotes, adjust the cadence so it sounds like an actual human talking, and anchor the campaign in a truth that only comes from qualitative research.

How do you find creative angles that an AI model will never output? You go to places where data hasn't been scraped, aggregated, and sanitized yet.

If you input "Write a high-converting Facebook ad script for a B2B project management tool," you will get an ad script that sounds like every B2B SaaS ad written since 2018. 

But if you take a raw transcript from a customer success call where a buyer says, "Honestly, I just wanted a tool that didn't make me feel like I needed a PhD every time I assigned a task," now you have something real.

Creator and former a16z podcaster Steph Smith argues that real leverage online comes from surfacing under‑utilised, non‑obvious internet signals - what she calls “clues” people leave about what they actually want - and using those to validate and differentiate ideas before everyone else catches on. 

The best creative growth teams are spending hours in customer interviews, reading Reddit threads, listening to sales call recordings on Gong, and digging through negative reviews of their competitors. 

They gather raw, unfiltered human emotion - the frustration, the specific vocabulary, the weird workarounds people use - and use that as the foundation for their campaigns. 

AI can help you format and scale those insights across multiple channels, but it can’t go out into the world and extract those truths for you.

Look at how classic campaigns were built. 

They weren't built by pleasing everyone; they were built by taking a distinct, opinionated stance that actively repelled the wrong audience to deeply attract the right one. 

Generative AI tools are inherently programmed to be helpful, polite, and balanced. They struggle with edge, irony, counter-intuitive arguments, and polarizing hooks because those elements sit outside the safest statistical probabilities.

For Seth Godin, playing safe is very risky and he warns that products built for everybody end up being for nobody. His solution isn’t to remove all friction, but to choose the right kind: the hassles that create traction, trust, and tension - and cut the ones that just get in the way. 

What does intentional friction look like in practice?

When you introduce a strong opinion into your marketing, you give your audience a reason to care. AI can help you scale the distribution of that opinion, but the bravery to take a stance in the first place has to come from human strategic leadership.

If you want your growth marketing to actually work today, you cannot treat AI as a strategy engine. 

You have to treat it as a hyper-efficient production line, while doubling down on the one thing algorithms cannot replicate: human friction, lived empathy, and radical strategic differentiation.

✌️,

Tom from Marketer Gems