Trust AI Like You’d Trust a 22-Year-Old Banker with a PowerPoint Addiction

Good morning you animals —

While I Was Writing Today’s Signal // Noise:

Who knew the Germans could produce such funk?! (no shade Germans, much love 🙌)

Now that I’ve probably indirectly insulted someone, let’s move from sound to signal—and get this baby rolling with what’s on my mind…

// The Signal

AI’s Not Your Co-Founder, Silly. It’s Your Goldman Intern.

If you're already running a business, a lot of AI headlines may feel a bit… off.

On the one end, it’s completely panicked paranoia: AI is coming for your job.”

On the other end, it’s all hype: “AI is your next co-founder.”

But here’s the frame that I prefer:

AI is just like your coke-railing, deck-building, investment banking intern.

It’s smart. It’s fast. It’s willing to work 24/7.

Also: it’s great with spreadsheets, easily confused, & highly over-confident.

Joking aside, this is actually backed by data (the smart & fast part, not the coke part, just to be clear).

➡️ Case in point: MIT and BCG ran a large-scale study with over 750 consultants, assigning them real business problems to tackle. Some consultants got access to GPT-4, and some didn’t.

The AI-assisted consultants:

• Completed tasks 12% faster

• Saw a 40% boost in output quality (based on third-party reviewers)

• Made fewer errors, especially on writing-heavy and open-ended tasks

• 🔴 But on tasks outside the AI’s capability frontier, those same people were 19 percentage points less likely to get the right answer compared to the control group.

Why?

Because people trusted it too much.

Instead of gut-checking outputs and knowing when to call bullshit, they leaned back and let the crazy ass intern take the wheel. The result is polished, but it’s still bullshit.

This is what I’m seeing more and more of lately - a nuance that can easily be missed.

Used well, AI absolutely compounds productivity.

Used poorly, it creates a false sense of speed and certainty—and can quietly jack up decision-making.

So what do you do with this?

• You don’t just let everyone “play around” with AI and call that supportive.

  You need to teach your team how to use it—when to trust it, when to pause, and how to audit what it spits out.

• You build workflows that assume some level of oversight.

  AI should generate a draft, not a decision. Review and ownership should still live with your A-players.

• You screen for interaction style.

  Some use AI as a sidekick. Others go back and forth with it the whole way.

  Both work— as long as you're not just pasting its output and calling it done.

We’re still early. The tools are improving fast (consider how much has changed since that study was done in late 2023 when GPT-4 just came out).

But here’s what’s already true:

AI helps people move faster — but that only matters if they’re headed in the right direction in the first place.

🔖 The Blog Round

If you’re a CEO or founder (or senior exec) and your team constantly comes to you with problems—but not solutions—this is a framework you can use to straighten it out fast.

I’ve used it to uplevel how my team communicates, how they take ownership, and it’s given my great managers something to use for their direct reports, too.

The best part, it prevents me from being a decision bottleneck.

Simple, clear, and it actually works.

Check out the post below.

👀 A few Jawns to Check Out

Was this helpful? Relevant? Totally off?

Tell me — I read every reply, promise.

Or, pass it to someone who might dig it.

Have a good weekend, you animals.

Love yous.

Jordan