Hey, it's Will and I feel like I’m getting stupider…

The convenience of letting AI do the heavy mental lifting is addictive and the speed becomes expected in my day to day workflow, but I feel like I’m constantly trying to fight the cognitive atrophy that sets in.

Whenever I try to write something from scratch; my best man toast, a friend’s birthday card, anything, it feels harder than it used to.

I talk to founders every week who feel the same cognitive shift. They used to think through problems clearly. Now they reach for ChatGPT (or Claude or whatever their model of the moment is) before they finish reading the question.

MIT's Media Lab just put numbers to this feeling. Four months of regular ChatGPT use reduced brain connectivity by 47%. 83% of users couldn't recall sentences they'd written minutes earlier. Mental effort dropped by 33%.

In this week's newsletter, I'm breaking down what AI is actually doing to your brain, and more importantly, how to use AI intentionally as a force multiplier rather than a cognitive crutch.

Welcome to Catalyst—your bi-weekly insights on emerging fintech and Web3 trends, a behind-the-scenes look at some of the top players in the space, and actionable strategies you can implement today. No fluff. No basic takes. Just clear insights on what's actually happening in fintech. 💻

Forwarded this email? Get Catalyst in your inbox each week.

Will’s Picks

A few things I’ve loved reading this week—

• MIT Media Lab: Your Brain on ChatGPT (Full Study)

• TIME: ChatGPT's Impact on Our Brains

• Nextgov: AI and Cognitive Decline

• Content Marketing Institute 2025 Research

The Study That Changed How I Think About AI

In June 2025, MIT Media Lab published research that should make every knowledge worker stop and reconsider how they're using AI tools.

They recruited 54 participants and divided them into three groups:

Group 1: Used ChatGPT to write essays
Group 2: Used Google Search for research
Group 3: Wrote essays with no tools (brain-only)

They tracked each group over four months using EEG headsets to monitor brain activity across 32 regions.

The Brain Connectivity Collapse

The brain-only group showed the highest neural connectivity, especially in alpha, theta, and delta bands (the regions associated with creativity, memory formation, and deep semantic processing).

The Google Search group showed moderate connectivity. Their brains were active and engaged.

The ChatGPT group showed the weakest connectivity, and it got worse over time.

By session 4, participants who switched from ChatGPT to brain-only writing showed reduced alpha and beta connectivity. Their brains had literally adapted to outsourcing cognitive work.

The researchers measured this as "cognitive debt;" the accumulated cost of replacing human thinking with AI assistance.

The Memory Failure

83% of ChatGPT users couldn't recall a single sentence they had written just minutes earlier.

They "wrote" essays. Submitted them. Claimed ownership. But when asked to quote their own work moments later, they couldn't do it.

The brain-only group? They remembered their essays clearly. They could quote passages. They felt ownership over their work.

The difference is encoding. When you think through something, your brain encodes it into memory. When ChatGPT thinks through something and you copy-paste the result, there's no encoding. No learning. No retention.

The Cognitive Engagement Drop

Mental effort, which is what leads to learning, decreased by 33% in the ChatGPT group.

The EEG data showed low executive control and attentional engagement. Their brains literally weren't working as hard.

Two English teachers assessed the essays. They described the ChatGPT-generated work as largely "soulless."

All the essays from the ChatGPT group were extremely similar. Same expressions. Same ideas. Same structure.

Meanwhile, brain connectivity dropped from 79 neural links to 42 in regular ChatGPT users. A 47% decrease.

The wildest part is that even after participants stopped using ChatGPT, the effects persisted. Their brains stayed sluggish. The cognitive debt didn't disappear when they put the tool away.

What This Means for B2B Founders

I use AI every single day.

For research. Outlining. Editing. Generating variations of concepts I've already developed, but I never use it for my actual thinking.

In an increasingly automated world, that is your one true competitive advantage.

In B2B, buyers trust thought leadership more than marketing materials. The 2024 Edelman-LinkedIn study found that 73% of decision-makers use an organization's thought leadership to assess capabilities.

If your thought leadership is AI-generated, you're outsourcing your competitive advantage and your buyers can tell.

The MIT study showed that they developed an "AI filter" that gets suspicious when writing is too clean or pristine. They can sense when there's no human thinking behind the words, and more importantly: you're making yourself dumber.

Every time you let AI do your thinking, you're weakening the neural pathways that make you valuable. You're trading short-term speed for long-term cognitive capacity.

That's a shit trade.

The AI-Assisted Thinking Framework

After experimenting with this for the past year and analyzing the MIT research, I've developed a framework that preserves cognitive function while leveraging AI's benefits. At least one that works for me.

Step 1: Do Your Own Thinking First

Before you touch AI, spend time with the problem.

Write your initial thoughts. Outline your argument. Identify your key insights. This is where the hard cognitive work happens.

This is where your brain builds those neural connections. Where learning occurs. Where your unique perspective emerges.

For content, I spend 20-30 minutes brainstorming before I ever open ChatGPT:

→ What's the core insight?
→ What's my perspective on this?
→ What examples support this?
→ What would make this valuable to my audience?

That upfront thinking is what makes the content mine and prevents the "soulless" quality the MIT study identified.

Step 2: Use AI as a Sparring Partner

Once you have your thinking, bring in AI to challenge it.

Prompt: "Here's my argument: [your thinking]. What are the holes in this logic? What counterarguments should I address? What evidence would strengthen this?"

This uses AI to make your thinking sharper without replacing your thinking.

You're still doing the cognitive work. You're just pressure-testing it against a different perspective.

Step 3: Use AI for Execution, Not Strategy

AI is brilliant at taking your strategic thinking and executing the tactical work.

Once I know what I want to say and how I want to say it, I'll use AI for:

→ Writing multiple variations of hooks to test
→ Optimizing for specific platforms (LinkedIn vs newsletter)
→ Editing for clarity and concision
→ Formatting and structure improvements

The core thinking, strategic insight, and unique perspective stay mine. AI just helps me execute on it faster.

Step 4: Use AI for Research, Not Synthesis

AI is incredible for gathering information quickly.

"Find me five examples of companies using X strategy."
"What are the current benchmarks for Y metric?"
"Summarize the key arguments in this paper."

I'll use ChatGPT to gather raw material. Then I close it and do my own analysis.

Step 5: Create Forcing Functions

The MIT study showed that the more people used ChatGPT, the lazier they got. By the third essay, many participants just gave ChatGPT the prompt and copied the output.

You need systems that force cognitive engagement.

My forcing functions:

→ The blank document rule: I write the first 200 words of anything without AI. Forces me to think through the core idea.

→ The explanation test: If I can't explain an AI-generated concept in my own words, I don't use it. This ensures I understand what I'm publishing.

→ The 48-hour rule: I let AI-assisted work sit for 48 hours, then re-read it. If it doesn't sound like me, I rewrite it.

→ The source citation requirement: Every piece of research AI gives me, I verify with original sources. Prevents copy-paste thinking.

These rules slow me down. That's the point. Speed without thinking is just cognitive decline.

The Cognitive Trade-Off in Content Marketing

Let's make this specific to content marketing, since that's what most of you are dealing with.

Content marketing is where the AI cognitive trade-off shows up most clearly.

AI can help you publish 10x more content. But if that content has no original thinking, you're just adding noise to an already noisy market.

The Content Marketing Institute's 2025 research found that only 22% of B2B marketers rate their content marketing as "very successful." Meanwhile, 91% are using content marketing.

That gap exists because most content is generic. It sounds like everything else. There's no unique POV. No original insight. No distinctive voice.

The AI Content Spectrum

I think about AI-assisted content on a spectrum:

Level 1: AI Generation (Worst)
"Write a LinkedIn post about B2B content strategy."
Result: Generic, soulless, indistinguishable from competitors

Level 2: AI Enhancement (Better)
Human writes core content, AI optimizes for clarity/format
Result: Maintains human thinking, improves execution

Level 3: AI Research (Better Still)
Human does thinking, AI gathers supporting data/examples
Result: Human insight backed by comprehensive research

Level 4: AI Sparring (Best)
Human develops thesis, AI challenges it, human refines
Result: Sharper thinking through adversarial testing

Level 5: Human-First, AI-Last (Optimal)
Human does all strategic thinking, AI handles tactical execution
Result: Preserves cognitive function while gaining efficiency

Companies are hiring content teams and telling them to "use AI to scale production," so they publish 5x more content, but it's all the same.

Meanwhile, the companies that are winning are using AI differently:

→ They're doing deep original research (human thinking)
→ They're developing unique frameworks (human thinking)
→ They're sharing contrarian perspectives (human thinking)
→ They're using AI to distribute and optimize (AI execution)

;

Here's my current framework for using AI intentionally:

For Content Creation:

1. Spend 30 minutes thinking before AI (builds cognitive engagement)

2. Write first draft completely alone (forces encoding)

3. Use AI for structure/clarity optimization (preserves core thinking)

4. Always re-read as if I wrote it (ensures ownership)

For Research:

1. Start with my own hypothesis (forces original thinking)

2. Use AI to gather supporting/conflicting evidence (speeds research)

3. Do my own synthesis without AI (preserves cognitive work)

4. Verify all AI-provided sources (prevents false confidence)

For Strategy:

1. Develop strategy in analog (whiteboard, notebook)

2. Use AI to pressure-test assumptions (adversarial thinking)

3. Make decisions without AI input (preserves agency)

4. Use AI for execution planning (leverages efficiency)

For Learning:

1. Read/study without AI summaries (forces comprehension)

2. Write my own notes and synthesis (builds retention)

3. Use AI to find related concepts (expands knowledge)

4. Teach concepts without referencing AI (tests understanding)

The pattern is consistent: Human thinking first. AI assistance second.

What This Means for Your Team

If you manage a content team, this research should change your AI policies.

Most companies are pushing teams to use AI to "do more with less." That might work short-term. Long-term, you're degrading your team's cognitive capacity.

Better approach:

1. Make AI optional, not mandatory
Let team members choose when/how to use AI. The ones who preserve their cognitive function will produce better work long-term.

2. Measure thinking quality, not just output volume
5 pieces of deeply original content beat 50 pieces of AI-generated content every time.

3. Create forcing functions for original thinking
Require teams to submit thinking documents before using AI. Forces cognitive engagement upfront.

4. Train on AI-resistant skills
Pattern recognition, synthesis, unique perspective development; these are the skills that remain valuable when AI commoditizes everything else.

5. Audit for cognitive debt
Regularly check: Is your team getting better at thinking, or just better at prompting? The former builds competitive advantage. The latter builds dependency.