The Real Uncanny Valley

Why AI Messages Feel More Human Than Human-Sent Scripts

Are the "robotic" messages in your inbox actually written by humans?

I got three LinkedIn messages this week that all started with "Hey John, I noticed you're doing interesting things with AI." Two of them used the same Dan Martell template I saw on YouTube a few weeks ago. The third misspelled my company name.

It's very possible that all three were typed by human fingers. But even if they were literally typed by a person, none of them involved a single thought about me.

Here's what I'm working through: We've been taught to worry about AI pretending to be human. But the real uncanny valley might be the opposite.

The Original Uncanny Valley

The classic uncanny valley comes from robotics. A cartoon character doesn't creep you out because your brain knows it's not real. A human doesn't creep you out because your brain knows it is real. But something that's almost human, not quite right, falls into the valley. It's the slightly-off face, the choppy movement. Your brain can't categorize it, and that dissonance registers as wrongness.

We've applied this idea to AI. Chatbots that try to seem human feel weird. Deepfakes feel unsettling. The worry is that AI will get good enough to fool us, and that deception will be the problem.

But I think we've got the direction backwards.

It's Not AI vs Human

The messages that feel most hollow in my inbox aren't the automated ones to reset my password. My bank's AI doesn't pretend it's been thinking about me all week when I want to check on a transaction. The expectations are clear. It's a system. It acts like a system. Fine.

The hollow messages are the ones that signal "I thought about you specifically" when the opposite is true.

And here's what I'm realizing: that hollowness has nothing to do with whether AI was involved.

Both AI and human interactions can be any of these three things:

Thoughtful. A human who spent an hour researching you and crafted a message about something specific to your situation. Or an AI that was trained on someone's genuine perspective, researched your work, and applied that thinking to write something that actually addresses your situation. Both feel real. Both are valuable.

Robotic. A chatbot that says "How can I help you today?" Or a human reading from a call center script word for word. Both are clearly automated. Neither pretends to be personal. Fine.

The Weird Middle. A human who copied a guru's template and sent it to 500 people with just the first name swapped. Or AI programmed to blast generic "I noticed you're doing interesting things" messages at scale. Both signal personal attention that isn't there. Both feel hollow.

The Problem Isn't AI

The problem is the weird middle. And humans live there just as much as AI does.

When someone copies a "script that's filling calendars right now" and sends it to everyone in their network, they're not being more authentic than AI. They're being less authentic than a well-programmed chatbot. At least the chatbot isn't lying about whether it thought about you.

A thoughtful conversation with AI beats a mindless conversation with a human. Every time.

If you're sending outreach right now, be honest about what you're doing:

Option 1: Go high volume, high quality with AI. Train it on your actual voice, your actual offer, your actual perspective on who you help. Let it research each prospect and apply your thinking specifically. This works because the thinking happened, even if the typing didn't.

Option 2: Go low volume, high touch manually. Pick 5 people. Research them yourself. Write something you could only write for them. This works because each message is genuinely personal. People will notice and your reply rate can skyrocket once you figure out who exactly you should be connecting with.

Option 3: Stop. Pick option 1 or 2. If you're copying templates and swapping in first names, you're in the uncanny valley. You're not doing sales. You're doing spam with extra steps. The recipients can feel it.

There are still tons of courses teaching this method and marketing companies selling it. Why? Because it worked in 2010. Really well.

But here's what changed: In 2010, sending automated outreach at scale was hard. You had to actually think about your offer. You had to build a real prospect list. You probably paid serious money for tools that could message at volume. The barrier to entry filtered out people who weren't committed. If someone put in all that work, there was a decent chance they'd deliver quality on the back end too.

Now those tools cost $50/month and take 20 minutes to set up. The barrier is gone, so everyone's doing it. And most of them aren't going to provide a quality service. We've learned something from all this noise: the effort in the outreach signals the effort in the delivery.

The fix isn't "use AI" or "don't use AI." The fix is: was there thought about the recipient? That's the only question that matters.

Pick a Path

You have two choices now:

Build systems with AI. If you want scale, this is the path. But it requires real investment: training AI on your actual thinking, building research workflows, iterating until the output genuinely reflects you. Your competitors are doing this. All of them.

Go manual and high-touch. If you want to stay hands-on, commit to it. Five prospects. Deep research. Messages that could only exist for that person. This works, but it doesn't scale.

What doesn't work is the middle. Templates with first names swapped. Scripts copied from YouTube. The uncanny valley is getting lonelier every day, and soon you'll be standing there alone wondering why nobody responds.

Before you hit send on your next message, ask yourself: "What's in this that I could only have written for this specific person?"

If the answer is nothing but their name, pick a path.

(If you're a corporate professional figuring out how to do sales without becoming the thing you hate, I wrote about that in Your Corporate Genius Doesn't Know How to Sell. The paradigm shift there connects directly to this.)

FAQ

Isn't all sales outreach kind of hollow by definition?

No. The best sales conversations I've had started with someone genuinely understanding my situation. When that's real, you can feel it. The problem isn't outreach. The problem is outreach that pretends to be specific when it isn't.

Why should I care whether it's a well-reasoned AI message or just sophisticated spam?

Because a thoughtful AI message might actually lead to real value for you. Just like the rest of us, you probably need better tools, ideas, or solutions. Especially considering the pace of improvement we're seeing. If someone invested the effort to design an AI that researches your situation and explains specifically why their service might help, they're actually trying to solve your problem. That level of reasoning takes genuine work to build. And there's a good chance that if you need what they offer, it'll be delivered with that same effort and quality. That matters more than who typed the message.

What makes outreach feel authentic?

One thing: Does the message contain information that required actually paying attention to the recipient? Something specific enough that the sender had to read, think, and form an opinion. The mechanism matters less than whether that thinking happened.

---

I'm figuring this out in public. If you're thinking about authenticity, AI, and what it actually means to connect with people, follow along.