I can train AI to sell, but can I teach it to smile?

Mark Zuckerberg, proudly wearing his Meta Ray-Bans, announced on a panel, that the landscape of advertising with the addition of AI, could create a ecosystem that disrupts the $1.5T industry through one ‘end goal:’ automation. After watching the video, I read a comment that said:

“To sell a product is not the same as selling a brand. Performance sells products. Connection builds brands. That’s something AI alone can’t replicate.”

As a marketer and avid user of ChatGPT, that made me think a little more about the future of media buying and where manual work and automation may butt heads. It goes without saying that AI excels at execution. One day, it may even rival the world’s best creative directors in speed and output. But can it create in the way the human mind has for thousands of years?

From Darwin to Zuckerberg

• 1000 BC – Grunt and Point
  The advertiser draws a picture of a goat on a rock and points at it while yelling. This is called “branding.”
• 300 BC – Scribe-for-Hire
  Advertisers write ads on papyrus scrolls. Their biggest challenge: papyrus doesn’t support tracking pixels.
• 50 AD – Street Yellers
  Advertisers pay town criers in bread and wine to yell deals in the market. No targeting, just volume.
• 1300s – Sign Masters
  Advertisers craft wooden signs with hand-painted icons. Market testing = “does it get stolen?”
• 1600s – The Pamphlet Hustle
  Advertisers learn to place ads in printed newsletters. A/B testing involves walking to different towns and seeing what works.
• 1920s – The Radio Revolution
  Advertisers start writing dramatic radio scripts. They now must convince people using only sound and absurd claims like “A smile so clean, even your mother-in-law will approve.”
• 1950s – Madison Avenue Era
  Advertisers wear suits, smoke indoors, and invent jingles. They pitch ideas with storyboards, martinis, and slogans like “Because you’re worth it.”
• 1994 – Banner Ads & Desktop Clickbait
  Advertisers learn HTML (sort of) and make flashing banners. Click-through rate: 44%. ROI? No idea.
• 1998 – Spam Lords
  Advertisers realize email is cheap. They blast everyone, everywhere. First CRM is just an Excel file named “emails_final_final2.xls.”
• 2003 – Pay-Per-Click Awakening
  Advertisers discover Google Ads. They become obsessed with keywords and stop caring about actual people. Conversion = king.
• 2010 – The Facebook Pixel Era
  Advertisers now stalk users across the internet. They obsess over targeting, build audiences, and measure click-through rates like day traders watching tickers.
• 2016 – Influencer Wrangling & Multi-Channel Wrangling
  Advertisers start diversifying across Meta, Google and soon to be TikTok, however incrementality is still cloudy. Advertisers slide into DMs of attractive strangers and negotiate product placements with emojis. “Exposure” becomes a currency.
• 2023 – AI Becomes Their Coworker
  Advertisers start using AI for copy, images, and even media buying. Their job now involves editing robots instead of designing ads. Some people continue editing ads because they realise many robots still suck.
• 2025 – Funnel Architect Vibes
  Advertisers don’t “make ads,” they build entire user journeys. Prospect sees an ad, reads a blog, clicks an email, visits a quiz, joins a retargeting sequence, and eventually buys socks.
• 2030 – The Full-Stack Advertiser (Zuckerberg’s plan)
  Advertisers log in to an all-in-one dashboard. They enter company details, connect Stripe, sync their website, upload product info, and hit "Launch Campaign." AI builds landing pages, writes 50 headlines, auto-tests 10 channels, manages budget, and tells them what’s working. They sit back, sip coffee, and wonder if the AI needs a raise.

How can AI combat branding?

When it comes to creative development, AI is still relatively new. Its ability to design and execute high-quality creative with minimal prompts will no doubt improve over time. But that evolution doesn't account for the psychological tendencies that precedes the birth of a brand or the human impulse to make something meaningful from nothing.

I recently looked back at some of the biggest DTC brands in the world (or at least partially DTC) and asked myself: where did these giants really get it right? No matter the angle I took, the answer kept circling back to one thing, ideation. Great brands tend to win by delivering value in one of two ways: Example A: They compete on price and convenience, like Walmart, Amazon or eBay. Alternatively, Example B: They generate intangible value through brand and experience, like Apple, Coca-Cola, Red Bull etc.

AI will likely accelerate the growth of brands in Example A, where winning is more about logistics, financial optimization, and mechanical problem-solving. But Example B is where things are challenging without human intervention. Creating emotional resonance, cultural relevance, and aesthetic nuance isn't just a task. It's a deeply human process rooted in psychology and perception.

Imagine it’s 1976. Jobs and Wozniak are sitting in a garage, sketching out the future of Apple. Now picture someone handing Steve Jobs a tool that could generate endless ideas, automate execution, and respond with incredible intellectual depth. Would that tool help him create Apple Computers? I’m not so sure. It would help him create an extremely advanced and seamless product, but how would it understand or think to ideate the rebellious nature and persona that came with the brand?

Until AI can sit across from someone, sense their energy, and translate the invisible mechanics of emotion into a brand, it’s still just a machine executing instructions, not a mind prioritizing how it can build a generational cult through firsthand, live interaction.

Brand Integrity vs Efficiency

There’s a tension between creative integrity and performance efficiency. AI leans hard toward efficiency, but brands that resonate deeply often take creative risks that algorithms wouldn’t recommend. This is where Zuckerberg’s plan of pure automation runs into a pretty challenging task for founders looking to build brands with a heavy reliance on consumer experience.

Meta’s targeting, as I wrote in a previous article, is unsuccessful roughly 99.875% of the time, at least within its own attribution window. But what about consumers who convert outside that window? How does Meta measure the long-tail impact of a great piece of creative - the kind that builds brand equity over weeks or months, not just seconds? The truth is, it can’t, for now. So how can an algorithm be expected to create for long-term emotional resonance when its core function is to optimize for short-term conversions? As a result, most mid and lower-funnel creative is engineered to push a sale, not spark a feeling and as long as the platform’s financial interests are tied to immediacy, that bias won’t change.

This is why, in the ideation phase of branding, likewise in media buying, human intervention remains essential, especially when developing creative for prospecting audiences. As I’ve written before, the future has shifted toward media mix modeling (MMM), where brands rely more on holistic metrics like MER (marketing efficiency ratio) rather than obsessing over in-platform data from Meta. So while the world may be convinced we’re heading toward a Starship-level reckoning of AI-led automation in marketing, and we’ll all be out of a job, the reality is more nuanced. The future will still demand emotional intelligence in media buying, so long as Meta remains blind to what happens beyond its own attribution window.