For decades, AI researchers fought over one question: what does intelligence look like?

One camp said it looked like expertise. Deep knowledge, rigid rules, logical reasoning. The other camp said it looked like… checks notes … uh, well, they said it looks like a baby. Pattern recognition, broad inputs, learning from scratch.

Interestingly, the baby camp seems to have won. Modern AI, the stuff rewriting industries right now, is built on their approach.

I was listening to a podcast on this (The Last Invention), and it made me rush to my keyboard to write this newsletter. The takeaway for founders is terrific.

When you hit a wall, the instinct is to fix it with expertise: take a course, hire a consultant, read the book everyone recommends. But that keeps you stuck preparing to fix the problem, rather than actually doing it.

It’s funny, though. When you’re an entrepreneur, the smartest move is often to approach problems as if you know nothing at all.

Staying behind on tech trends can be a career killer.

But let’s face it, no one has hours to spare every week trying to stay updated.

That’s why over 100,000 engineers at companies like Google, Meta, and Apple read The Code twice a week.

Here’s why it works:

The idea of AI, as we currently understand it, originated with Alan Turing amid the backdrop of WW2. Originated, then went dormant for over a decade. 

That is, until a group of researchers gathered at Dartmouth College in 1956 to “solve intelligence.” Not study it. Solve it.

They believed they could build a machine capable of truly thinking. And so, they coined the term “artificial intelligence.” 

They also kicked off a war that would last decades.

On one side: the symbolists. Marvin Minsky, Allen Newell, Herbert Simon.

They believed intelligence was expertise made explicit. If you could encode enough rules, enough logic, enough domain knowledge into a system, it would “think.” The goal was to build machines that reasoned like experts; like people who were masters of their category. Chess grandmasters became their north star. 

On the other side: the connectionists. Frank Rosenblatt, and later Geoffrey Hinton (whose name you may recognize, as he’s often referred to as the “Godfather of AI”).

They believed intelligence came from pattern recognition. No rules. No explicit knowledge. Just layers upon layers of simple connections learning from raw input, the way a baby learns to recognize faces before it knows the word “face.” Our brain’s neural networks became their north star.

As a summary:

The symbolists dominated for thirty years.

They had the results. And in turn, the funding and prestige. Today, we’d consider their expertise-focused systems boring, but their work powers everything from medical diagnosis to financial trading. 

Meanwhile, the connectionists were sidelined. They spent the 1980s in relative obscurity, becoming the laughing stocks of the AI world. Despite this, they remained convinced the field had taken a wrong turn.

Then things started to break down.

The symbolists’ expert systems didn’t scale nicely. Every new domain required painstaking knowledge engineering.

At the same time, connectionists’ neural networks got bigger, faster, and able to pull from greater amounts of data. By the 2010s, deep learning was beating expert systems at almost everything: image recognition, language translation, game-playing.

As far as the “who’s right” debate goes, the AI “babies” won. They just… needed some time to grow up. 🤷🏻‍♂️

I’m not just here to recount the drama of AI development (even though it would make a great TV show). Here's where it matters for you.

When founders get stuck, they usually diagnose it as a knowledge gap. “I don't understand Facebook ads well enough.” “I need to get a lot better at sales.” “I should take a course on operations.”

Sometimes that's true. Most of the time, though, it’s procrastination masquerading as productivity. What great psychotherapist Alfred Adler would refer to as “hesitating.” Doing things that feel productive, in order to postpone what you actually need to do.

But there's a second trap. One that hits founders who have experience.

Cognitive scientists call it the Einstellung effect. In a 2008 study, researchers tracked the eye movements of chess players solving puzzles. The experts kept looking at familiar patterns, even when they believed they were searching for new solutions. 

They just couldn’t help it. Their knowledge was blinding them.

The symbolists fell into this trap at a civilizational scale. They had such firm ideas about how intelligence should work that they couldn't see how it actually worked.

The connectionists had a different advantage. That is, they were willing to look stupid. (just a note here: they didn’t actually look stupid. But that IS what our minds are afraid of)

They were willing to let a system learn from scratch with no guarantees, legible reasoning, or expert validation along the way. They stuck with it until someone proved them wrong… But no one did. Their willingness to appear stupid let them win.

For founders, the lesson cuts both ways.

Either way, the prescription is the same: think like a baby.

_{*Article is paywalled, which, yeah, is kind of annoying. But I believe good research and journalism is worth paying for.}

The history is interesting, but you didn’t subscribe to this newsletter for an advantage at trivia night. Chances are, you're still reading because you're stuck on something.

Maybe it's a problem you've been circling for weeks. Maybe it's a part of your business you've convinced yourself you need to “learn more about” before you can act.

Here's a three-question filter to cut through the overwhelm and start thinking like a baby.

Strip away your assumptions about what should be happening. Forget best practices. Forget what the course taught you. Forget what worked last time.

Describe the situation as if you're seeing it for the first time. A baby doesn't know that “conversion rates should be 2-3%.” A baby just sees: people visit, people leave.

Write it down in plain language. No jargon. Just what you observe.

Expertise makes us sophisticated. Sophistication makes us dismiss simple solutions as beneath us.

The connectionist approach looked stupid to the symbolists. No elegant logic or explicable reasoning. Just throwing data at a system and hoping patterns emerged. And yet, they emerged.

What solution have you been dismissing because it seems too obvious? Too basic? Too unsophisticated for someone at your level?

That's probably worth trying first.

Founders stop asking naive questions once they feel like they should know the answers. This is how blind spots calcify.

Hinton kept asking “what if the brain doesn't work like a logic machine?” for decades. Even when it was an embarrassing question in his field, he asked it anyway.

What's the question you've been avoiding because it might make you look ignorant? Ask it. Send that email. Post in that community. Call that mentor.

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide