AI Replacing Salespeople Perfectly Defines the AI Redundancy Problem

AI for sales outreach sounds great in theory. I know. I thought the same thing 15 years ago. For 10 minutes. Until I really thought about it. Then I laughed and went on to the next thing. 

Then last month, I wrote a controversial column called “We All Know AI Can’t Code, Right?” in which I argued that while generative AI is good at creating compilable syntax, the fact that it can’t create anything that it hasn’t already seen means potential disaster when trying to deliver real code to real customers. 

A reader commented, “It’s not just code. Our CEO recently looked at signing up for a £1,500/month AI sales outreach tool for generating leads. It sounds great in theory, and the price tag is almost certainly calculated to make it seem like a premium product, but surely it’s obvious that the actual outcome is likely to be a low quality, low conversion, brute force campaign that alienates as many (or more) customers as it generates? Or is that just me?”

Sigh. No, it’s not just you. 

Here’s why a lot of the hype around generative AI is built on duplicating use cases that are already broken.

800 Websites for Pennies on the Dollar!

Let me quickly explain AI redundancy. I won’t get technical or wordy.

As I said, I was swimming with these AI sharks as far back as 2010, when Automated Insights changed our model from next-gen sports stats visuals to automating sports content based on those stats. It was the precursor for generative AI, but we didn’t know that. It wasn’t even called natural language generation until a couple of years later.

Anyway, to prove how cool our tech was, we stood up over 800 websites, one for each pro and college football, basketball, and baseball team in the U.S. And we populated each of these websites up to five times a day with automated sports stories — game recaps, previews, players of the week, and so on. 

This was super cool and we got a ton of attention. A ton. Remember, this was something no one had ever seen before. 

What we didn’t get was money. Because there was no viable revenue stream. Because all these teams had at least one human writer covering them for at least one news outlet. So we were cool as hell, but we were redundant.

Then something serendipitous happened. A number of the smallest colleges from the smallest athletic conferences started sending out press releases whenever our writer bots selected one of their team’s players as the player of the week for that conference. Because those small schools had no other coverage besides us.

That’s when it hit us. There was the use case. Write stories that humans can’t. Redundancy removed. Revenue stream achieved.

Now, humans can and do indeed write sales outreach cold emails. They’re just awful at it. So if AI does the same thing, it’ll be just as awful, but it will cost much less. 

This is the promise of a lot of generative AI use cases right now: doing things terribly that humans do terribly today, and making it cheaper to do those terrible things terribly.

Hi, My Name Is Already Deleted

I’m not telling any secrets when I admit that when any one of the hundreds of sales outreach cold emails I get makes it through my filters, I immediately delete it without reading more than four words. 

Oh, and here’s a pro tip. If someone “came across” you or your company, they came across you in the list they bought that was probably illegally compiled and sold.

But here’s the thing.

“Hi Joe, I assure you I’m not a bot.”

I assure you I don’t care.

In over 20 years, I have never once bought anything, seen any webinar, downloaded any demo, or even responded “not interested” back to a cold email, human-written or otherwise. And I am far from alone. Yes. I know it works, or Nigerian princes wouldn’t still be doing it, but this is not something that needs to be artificially intellgensed. 

I know. I’m preaching to the choir. So let me extend the metaphor.

Scrolling Until the Content Fulfills the Promise of the Title

You hate this too, right?

One of the main reasons Automated Insights was able to succeed in sports, finance, marketing, real estate, and so on was not because it could replace real writers in those subjects, but because those areas of interest were already plagued by low-quality SEO-driven content farms.

Now, our “cool tech” resulted in our doing a lot of interviews. And every time a journalist asked us how soon we were going to put them out of work, our confident response was, “Never. Not you. We might even make your job easier.”

I’ve written about this several times. Over a year ago I was talking about how generative AI was coming for the content farms. 

Nine months ago, when Sports Illustrated got caught using AI content and avatars, I noted that its AI was replacing mostly paid promotional product reviews. 

And then a few months ago, I talked about the difference between generative AI and “real” AI and how the line between the two is being blurred to imply the possibility of richer, more useful use cases that aren’t quite viable with generative AI yet, and may never be.

Pennies on the Dollar Instead of Dimes

Add all that up, and it’s a river of redundancy being produced by generative AI. I was spot on about how this was going to play out. genAI replaced the awful SEO content farms first, then the spaghetti coders, and now the worst of the sales outreachers.

But the only thing gen AI is actually doing, in a lot of cases, is offering a 1-cent-on-the-dollar solution for tasks and content you were paying only 10 cents on the dollar for prior to gen AI.

You’re still getting only 10 cents on the dollar worth of quality. 

And the bigger problem is it’s being sold as a whole dollar of quality.

But …

Any writer or content creator worth their audience will tell you that the bots can’t do what they can do. 

Any software developer with real-world experience sees AI as a tool, not a replacement.

And any sales professional selling a product with any value knows cold outreach emails aren’t worth the effort, and would rather spend their time actually selling the benefits of a product they love as much as they want you to.

Yes, AI will keep getting better, and truthfully, I can’t wait. Because I know from 15 years of experience that the better it gets, the more these redundant use cases will be called out for what they really are, and AI will eventually be universally understood correctly as an assistant to talent, not a replacement for talent.