192: Mixed Bag
Zuck's threat, WPP's response, why LLMs hallucinate, complexity bias and more
I took an unintended hiatus from writing. But all is now well.
Many thanks to all of you who joined this newsletter recently. I endeavor to publish weekly; but the past two proved insurmountable.
Today’s post is a diverse collection of reactions in a loose central theme. We’ve long surmised creativity has been at risk because of AI, and now some threats are becoming clearer. But there’s still numerous indications the AI hype isn’t all that.
Our human capacity to risk continues to serve us well.
💥 “You don’t need any creative.”
It’s probably wise to take Mark Zuckerberg seriously when he says he’s going to dismantle your industry. If I was running a creative agency, I’d ask my leadership team to read Ben Thompson’s interview with Zuck; and ideate. Because our clients will. Zuck says the easiest task his team can focus on now is, “Use AI to make it so that the ads business goes a lot better.” To be clear, they’ve already built the measurement and targeting pieces of the puzzle. To wit, “we are just better at finding the people who are going to resonate with your product than you are.” Now he aims to absorb and automate ideas, concepts, writing, design, and art direction. The oft-cited, money quote:
“We’re going to get to a point where you’re a business, you come to us, you tell us what your objective is, you connect to your bank account, you don’t need any creative, you don’t need any targeting demographic, you don’t need any measurement, except to be able to read the results that we spit out. I think that’s going to be huge, I think it is a redefinition of the category of advertising.”
The good news out of all of this is we’ll see—we’ll see if and how algorithmic creativity functions, or not. And this will open a gap for those marketers who see value in strategy, in novel human ideation, in risk. (See my final note at the end of today’s post.) More on this from The Verge, Campaign, Digiday.
🧑🏼🤝🧑🏼 We’re not an agency, we’re a concierge
So if you’re an ad holding co like WPP, you hear Zuckerberg, you examine the AI landscape and conclude: Our value to brands is being a concierge. Sure, ideas matter, but what matters more is demystifying, simplifying and connecting the diversity of AI and related media options (e.g. TikTok’s Symphony) into a single point-of-contact. Because brands can’t rely on Meta alone. Hence the cloning of WPP’s Chief Creative Rob Reilly (Instagram, TikTok) amidst Cannes to sell that story. More via Adweek.
🗺️ Have hope! LLMs don’t (yet) have world models
If you’re looking for something to slow the seemingly unstoppable rush of AI, consider the concept of world models. As defined by Gary Marcus, professor emeritus of psychology and neural science at New York University, a world model is, “a computational framework that a system (a machine, or a person or other animal) uses to track what is happening in the world.” Each of us uses world models to track where and when we are, enabling persistent context.
But here’s the thing: LLMs lack this central, critical capability, writes Marcus.
“The fact that they can get by as far as they can without explicit world models is actually astonishing. But much of what ails them comes from that design choice.
§
LLM are giant, opaque black boxes with no explicit models of at all. Part of what it means to say that an LLM is a black box is to say that you can’t point to a model of a particular set of facts etc inside.
§
The lack of stable, robust world models matters enormously not just for business, and legal writing, agents, math, everyday understanding, video comprehension and video generation (to name a few) but, also, crucially for AI safety.”
In other words, the hallucinations will continue. As will the opportunity for humans to leverage what makes our approach and output meaningful.
🦾 In the meantime, don’t compete with AI’s peaks
I agree with
Jones in this short piece on the collapsing knowledge economy that AI is not (yet) all-powerful and is, in fact, “not super great at” many tasks and operations—which is another way of saying don’t compete with what AI is good at. Instead, focus on where and how your artistry, your brilliance, stands beyond AI’s current grasp.
⚖️ And maybe don’t fall prey to complexity bias
presents another worthy perspective in the continuum of Zuckerberg, WPP, Marcus and Jones—suggesting, “As humans we often prefer complicated solutions over easy ones (complexity bias) and we can miss what’s right in front of us. In working with and applying AI I think we’re at risk of falling foul of complexity bias in some key areas, from prompting to products to AI systems.
§
Spending a little time being clear about what it is we’re actually trying to do rather than going straight to ChatGPT always pays dividends.”
Amen.
💰 Language matters, doesn’t it?
A recent Gallup-Walton Family Foundation survey of AI use by K12 teachers, found “teachers who use AI tools at least weekly save an average of 5.9 hours per week,” and labelled this “a dividend.” Kind of an odd frame, since dividends flow to capital owners and pedagogical efficiency doesn’t deliver a cash payout to shareholders. So I agree with the University of Mississippi’s
who summarizes:“Calling time savings an ‘AI dividend’ reveals everything wrong with how we're thinking about this technology. Dividends go to shareholders, not workers. Real gains for teachers would look like smaller class sizes, better support, or actual schedule flexibility—not just more efficient ways to complete the same overwhelming tasks that form unrelenting workloads.”
📺 Once more into the breach, dear UGC aficionados
YouTube is giving Bazaarvoice, Billo, Social Native, et al a run by announcing Open Call during Cannes 2025. The tool, “allows brands to place broad requests for sponsored content to any creators in the YouTube Partner Program,” notes Adweek. Seems inevitable; but success will depend on brand leaders engaging.
🏆 DM9/Cannes cheating
I thought Ashley Rutstein offered a cogent opinion piece on agency DM9’s faux entries debacle. More on the three 😳 withdrawn campaigns via The Drum. Yes, AI made cheating easier. But then, cheating was always possible without AI—as was and is deciding to do the right thing. (Here’s Ashley on my podcast earlier this year.)
📝🚨 BRXND AI Conference - September 18 in NYC
I just bought tickets to attend Noah Brier’s BRXND AI conference on September 18 in NYC. If you’re looking for an intense, memorable experience, this might be it. Here’s my podcast conversation with Noah.
⛔ WNW will officially close on June 30
13 years later, the infamous freelance creative portal, Working Not Working is closing following its acquisition by Fiverr.
🦉 “The thing you’re missing from your music [or creativity] is risk.”
And let’s conclude with some wise advice (via TikTok) from the Grammy-award winning audio engineer Greazy Wil.
“The thing you’re missing from your music is risk.
It’s plain and simple.
That’s it.
That’s what you’re missing.
You’re not taking enough risks.
It’s not a microphone.
It’s not a compressor.
It’s not a plugin.
You aren’t taking risks.
You know how I know that?
Because the piano in John Lennon’s Imagine is the worst sounding fucking piano l’ve ever heard in my entire life.
It sounds like shit.
It’s like mono as fuck.
And it’s just all spongy and soft.
And it’s got the slap delay from his voice on it.
But when I listen to that song, fucking heaven, dude.
That is the best sounding piano l’ve ever heard in my entire life.
So, record that shitty vocal.
Record that shitty guitar.
Record that shitty drum set.
Put it up in the front of the mix too.
Make it just the most obvious thing in the song.
That’s where true artistic creation shines and people really acknowledge and love your music.”
In this age where “mediocrity is now free” thanks to AI, Greazy makes a salient point—history making, brand building, culture changing comes not from technique or safety. It comes when you and I take the risk to put our ideas into the world as they are.