top of page
Search

Apple Just Gave AI a Reality Check, and It’s Exactly What I’ve Been Saying

By Michael Cunniff, Founder & Creative Director, TheSuperAgency.com


Ai & Automation Isn’t New — The Hype Is


I’ve been building and developing for a long time. Automation has been critical to my work for over 20 years, long before “automate” and “AI” became trendy buzzwords.


Back then, we called it process engineering, scripting, or system integration. The work hasn’t changed nearly as much as the marketing language around it.


Now, every week, AI dominates the tech press. Another “groundbreaking” model launches, promising to “reason,” “think,” and “plan” like a human. It’s sold as the inevitable march toward Artificial General Intelligence.


But I’ve been around enough hype cycles to know when we’re selling vision more than reality.



Monster Mind Marketing, located on Galt Ocean Mile, the original Super Agency Office. Remember when it was "Graphics, Printing, Websites" now if I mention the word "website" someone will rapidly say, "Ai can build your website" to which I respond, "Can it though?".


My Perspective


I’ve run an award-winning marketing agency for two decades. I’m all-in on Apple — if they make it, I probably own it. I’ve developed platforms from scratch, built systems for industry leaders, and worked with titans of business.


So when I see many of today’s “AI models” looking suspiciously like rebranded open-source programs that have been floating around for years, I notice. The tech might be more refined, but the underlying concepts aren’t new.


Ella Cafe ordering system and Ai integrations for Uber Eats and Ordering Apps
Ella Cafe ordering system and Ai integrations for Uber Eats and Ordering Apps

I use AI daily — in workflows, client management, and creative processes. It’s a valuable tool. But I’ve never believed the narrative that these models are truly thinking. Strip away the hype, and you’ll find vibe coding is just platform development. Search engines are being introduced as the backbone of what the data and language model AI leans on... and plenty of AI workflow tools are just pigs in lipstick, dressed up to look like something more revolutionary than they are.


Apple Put AI to the Test


This isn’t just my opinion. Apple recently published a study that systematically tested these “reasoning” models — the ones from OpenAI, Anthropic, Google, and others — to see if they could actually think.


Instead of cherry-picked demos, Apple used controlled, classic logic puzzles: Tower of Hanoi, Blocks World, River Crossing, Checker Jumping. These tasks are designed to scale in complexity, making it easy to see exactly where a system’s reasoning ability peaks and where it fails.


The results were telling:


  • Low complexity: Basic models sometimes beat the “reasoning” models.

  • Medium complexity: Reasoning models pulled ahead.

  • High complexity: Both collapsed to zero accuracy.


And here’s the kicker, even when the models were given the exact algorithm to solve the puzzle, they still failed at harder levels. In some cases, they actually reduced their reasoning steps midway through, even when they had tokens left to keep going.


It’s like watching someone give up on a puzzle halfway through, despite having the instructions and the time to finish.


The Reality: AI Doesn’t Think


Apple’s research confirms what many in the industry already suspect: these models don’t think in any human sense. They predict. They’re extremely advanced pattern-completion systems.

That’s not a bad thing. Prediction at this level can be incredibly useful. But let’s not confuse utility with cognition.


The Horse and Buggy vs. the Steam Engine


This is where my analogy comes in. Right now, AI is the horse and buggy at its peak. We’ve optimized it — better wheels, smoother suspension, stronger materials. It’s faster and more comfortable than ever.


But it’s still a horse and buggy.


Quantum computing is the steam engine. When it arrives at scale, it won’t just make the buggy faster — it will make it obsolete.


The steam engine changed the world. It altered migration patterns, created new industries, and destroyed others.


Michelin stars, the restaurant awards, started as a guide to get people to drive more, which sold more tires. That’s the kind of domino effect quantum will bring.


Using an Apple X Phone on a 2010 Lamborghini, both aren't even close to the modern day cell phone iPhone 16 or iPhone 17 or the $500,000 technological wonder that a modern Lamborghini or Ferrari is. In my example; That Red Gallardo is Ai and a Rosso Revuelto is Quantuum.
Using an Apple X Phone on a 2010 Lamborghini, both aren't even close to the modern day cell phone iPhone 16 or iPhone 17 or the $500,000 technological wonder that a modern Lamborghini or Ferrari is. In my example; That Red Gallardo is Ai and a Rosso Revuelto is Quantuum.

Where This Leaves Us

I believe AI will one day evolve alongside quantum computing into something that does resemble true reasoning. But today’s models are the final, refined iteration of pre-quantum tech.

Apple’s study is the reality check the industry needed.


Even the most expensive, hyped-up reasoning models hit a complexity wall and collapse.


So yes, I’ll keep using AI for what it is — an incredible tool. But I’m saving my real excitement for the day the steam engine finally arrives.


And when it does, everything we’re hyping now will feel like the last chapter of an old book.

 
 
 

Comments


bottom of page