🦾How You Can Leverage:

We built the exact same puzzle that Apple claims reasoning AI models can't solve.

In 30 seconds.

And the AI model solved it.

Without flaws.

 Apple's "research" paper went SO viral it showed up in a business newsletter that usually ignores AI completely.

That's when we knew something was seriously wrong.

So we did what any investigative reporter would do. Read the study twice manually, fed it to three separate AI models, then built the exact same Tower of Hanoi puzzle Apple claims is impossible for AI to solve.

(You can try it here.)

Plot twist?

The AI model solved it PERFECTLY. In optimal moves. Using the same AI model Apple tested and said that it didn’t work.

What we uncovered next will blow your mind.

Apple didn't just fail at AI research. They appear to have weaponized questionable science to cover the biggest business fumble in corporate history.

We're talking a potential $2 TRILLION that Apple left on the table because they couldn’t crack AI.

So instead, the tech juggernaut has resorted to throwing shade at AI’s development via “research papers.”

Those quotes? Yeah…. jump into today’s episode.

Let’s goooooooooo.

In Apple’s “research”, they banned AI models from using code.

Huhhhh?

You're testing reasoning capabilities but removing the most effective reasoning tool available. It's like testing race car performance but yanking out the engine first and saying NOW GO FAST CAR GOOOOOO!

Then calling every car ever made slow.

But here's what really caught our attention. Apple capped the AI models they tested (Claude and DeepSeek) at 64,000 tokens output when a 128,000 token version was available the exact same day they ran their tests via Claude. We checked the internet archives to verify this.

Why’s that matter?

One of the big metrics Apple relied on was logic-based puzzles like Tower of Hanoi.

A 13-disc Tower of Hanoi puzzle, though, requires approximately 65,000+ tokens to solve properly when you output every move as Apple required.

In other words, Apple knowingly tested models on questions that were actually/physically/legit impossible, then acted surprised when it faile?

Fishy, right?

It was through “failures” like these that lead Apple to label THE ENTIRE CLASS OF REASONING MODELS as "AI failures" when models couldn't break the laws of physics.

That's not reasoning collapse.

That's hitting an artificial ceiling that Apple intentionally built.

Try This

Next time you see research claiming AI "fails" at something, check if they artificially limited the AI's capabilities first.

Look for banned tools, capped resources, or weird constraints that don't match real-world usage. If you can't replicate their setup without those limitations, the failure might be by design, not by AI inability.

2021: Apple dominated the world at $2.1 trillion market cap. Microsoft trailed at $1.6 trillion.

Half a trillion dollar lead.

Today: Microsoft sits at $3.5 trillion market cap. Apple remains stuck at $3 trillion.

If Apple had figured out AI and kept pace with Microsoft’s modest growth over the last four years, they'd be worth $5 trillion right now.

Instead, they've left approximately $2 TRILLION in market value on the table by completely botching artificial intelligence.

(The ARROGANCE of trying to rebrand AI as “Apple Intelligence” last year. lolz)

The numbers tell the whole story.

Reports from 2023 show Apple spent millions of dollars DAILY training their internal AI model codenamed Ajax. They claimed internally it was more powerful than ChatGPT.

When they finally released it?

So embarrassing they didn't mention it by name during their main keynote presentation.

A small language model that lives on device. That's it.

After reportedly spent billions on AI development, Apple only walked away with some lawsuits for failed promises on AI delivery.

Apple remains the only major tech company that failed to produce competitive AI offerings. They're now facing multiple class action lawsuits for promising Apple Intelligence features they haven't delivered.

How’s this relate to the The Illusion of Thinking paper that Apple dropped?

This research paper dropped DAYS before their Worldwide Developer Conference where they essentially announced an AI gap year.

Coincidence?

We think not.

Knowing their stock would take a nosedive for failing to release anything noteworthy on the AI end, they instead weaponized a piece of corporate marketing masquerading as a research paper to say that AI kinda stinks.

Try This

When evaluating any company's strategic moves, calculate their opportunity cost using competitors as benchmarks. T

ake their current market cap, find the growth rate of their closest competitor over the same period, then apply that rate to see what they could have achieved.

The gap reveals exactly how much their strategic decisions cost them in real dollars.

This isn't Apple's first time pulling this corporate maneuver, because they've released multiple papers downplaying AI capabilities while frantically scrambling to catch up behind the scenes.

Real breakthrough AI research typically involves multiple institutions working together like Stanford, MIT, OpenAI, Google, and Microsoft collaborating to provide cross-institutional validation like this one.

Apple's paper? ALL Apple researchers. Just them.

(Our hottest take: we don’t think any distinguished AI researcher outside of Cupertino woulda ever signed off on this flimsy thing.)

That's what marketing disguised as science looks like.

The research community isn't buying it either, with AI researchers openly criticizing the flawed methodology on social platforms. When you essentially tell every other researcher that their work is invalid due to "contaminated data," then use puzzles that are plastered all over the internet anyway, you're basically dismissing the entire scientific community.

Apple's desperate because everyone else is winning the AI race. Google's making breakthrough discoveries, OpenAI's models are revolutionizing industries, and Microsoft's integrating AI into everything.

Apple? Still struggling with basic voice recognition after spending billions.

Try this

Next time you see explosive research claims about AI limitations from a single company, dig into three specific areas.

• Be wary if a study limits model capabilities without a clear reason—real research tests models at full strength.

• Check if papers are released right before a company's product launches; that timing often isn't accidental.

• If all authors are from one company and making big claims, it's likely marketing, not objective science.

Wanna dive even deeper?

Check this great Twitter thread from Andreas Kirsch, a research engineer who's worked at Google, DeepMind, Midjourney and others.