BY JOE PROCOPIO, FOUNDER
TEACHINGSTARTUP.COM@JPROCO
"Your job isn't going to be replaced by A.I. Your job is going to be replaced by people who know how to use A.I."
Man, that sounds clever. It's also wrong.
Well, I mean, it's not entirely wrong. It's just not right.
I read that exact statement not long ago in one of those "A.I. is coming for your job!!!" articles I was trying to debunk (and I feel like I did). I didn't bother to keep the article or even go back and look it up again for this article because I've since seen the same quote in slightly different forms attributed to a number of different people -- all of whom I believe are very smart and I have no beef with.
The problem, as it always is with A.I., is context.
I've been playing around with generative A.I. for over a decade, including co-inventing the first publicly available platform that turned data into words back in 2010. We sold that company, Automated Insights, to a private equity firm in 2015.
I'm still messing around with all forms of automation and A.I., including generative, and I'm actually really impressed with the progress since the long-ago days when we started cranking out fantasy football recaps for Yahoo or quarterly earnings reports articles for the Associated Press.
But then I see dire warnings like "You must learn to use A.I." and I'm like: Wait, are you telling people that they need to get really good at writing prompts? Because that sounds like a road to nowhere. And also a really boring career. Oh, and also a lot of wasted time developing the exact kind of skill that leads to getting replaced by a machine pretty quickly.
Here's what I mean.
The Obviously ChatGPT-Generated Sales Email
Yesterday, I got a spam email in my inbox at Spiffy, the VC-backed mobile vehicle maintenance startup where I'm chief product officer. Now, I get probably 100 spam emails a day to this inbox, because a lot of people want to sell us a lot of things we don't need -- and that might be the most pathetic humblebrag in humblebrag history.
Anyway, this one spam email stood out because, well... honestly, because I was looking for it, the same way I was looking for all the A.I.-job-eating-fear-mongering articles a couple of weeks ago,
The sender of this particular spam email gushed with praise for all our company's recent accomplishments, which they then listed in the easily identifiable bullet-list-in-paragraph form that ChatGPT uses. I say "easily identifiable" because that's how we started building the early versions of our own natural language generation software, but to be honest, if you see it once, you'll see it every time after that.
Back to the spam email. With each of Spiffy's recent accomplishments -- a promotion, a fundraise, and so on -- the sender included the reason why they felt these accomplishments were impressive, reasons which just happened to be the same as those we ourselves touted in our own press releases.
Ergo, the email was obviously ChatGPT fed with LinkedIn data. I know this because of a little LinkedIn glitch we've been using for years to figure out if the person emailing us has actually taken the time to figure out who we are and what we do.
See, that last bit is what a good salesperson is supposed to do. They're supposed to take the time to learn about our company in order to be able to detail the fit between what we need and what they're selling.
They're not supposed to blow smoke up our inboxes by having A.I. regurgitate all the accomplishments we already listed somewhere else. That just shows that they actually haven't done any of that research, that they have no idea what we do, and that they have even less of an idea whether or not we need what they're selling.
Here's where I get harsh. When we talk about A.I. replacing jobs, the person responsible for that email needs to be fired on the spot because A.I. has not only already replaced them, but has called them out for being completely inept and lazy at doing their job -- a phenomenon that technology, especially A.I., is quick to do and merciless when it does.
And I'm not talking about the sender of the email -- the name listed in the email signature. For all I know, that person may not even exist. I'm talking about the person who "learned to use A.I." (my quotes) to send a ChatGPT sales icebreaker email to probably hundreds or thousands of companies.
Yes, that person who took the time to "know how to use A.I." (my quotes again -- snark level at 100) to create something just as useless as the hundreds of spam emails I was getting before them from humans who just manually cut-and-pasted the same stuff into the same spam email over and over again.
See, the key to not being replaced by A.I. isn't to get good at what A.I. does. Far from it. The key is to get great at the things A.I. sucks at.
A Contrarian Bet Against A.I.: Part 2
Back in May, I wrote an article about making a contrarian bet against the A.I. money train by exposing the scarcity, and thus the increased value, of uniquely human knowledge, experience, and intuition.
Again, I'm not a Luddite. I'm a big fan of A.I. I just want it done right. So allow me to make the contrarian argument for the uniquely human side of business.
When Automated Insights started producing public company quarterly earnings reports recaps for the Associated Press, boosting their output from 300 stories per quarter to 4,400, the managing editor at the time, Lou Ferrara, made public statements about the fact that not only would no AP journalist be made redundant, they were now able to do what they did best, and what was ultimately better for the business. He said:
If anything, we are doubling down on the journalism we will do around earnings reports and business coverage.
We are going to use our brains and time in more enterprising ways during earnings season. Rather than spending a great deal of time focusing on the release of earnings and hammering out a quick story recapping each one, we are going to automate that process for all U.S. companies in the 4,400.
Instead, our journalists will focus on reporting and writing stories about what the numbers mean and what gets said in earnings calls on the day of the release, identifying trends and finding exclusive stories we can publish at the time of the earnings reports.
Learning to use A.I., in this case, and in most cases, is a matter of pushing a button and getting information. It's the kind of thing we do every day with our phones. But the magic isn't knowing whether or not it might rain, it's the skills to plan your day around it.
In the case of generative A.I., getting the information is just creating a prompt that's "close enough" to get started. There is nothing wrong with that. It's what you do with that information -- that's where you compete.
Just stuffing emails full of ChatGPT content is redundant work that goes all the way back to the envelope-stuffing jobs of the 1970s. The content is almost immaterial, especially without the human connection -- the kind of human experience used by those AP journalists who were now freed up to unleash their uniquely human knowledge and intuition.
So yeah, if you're in danger of losing your job to A.I., you were probably already in danger of losing your job to someone or something else. Don't lean into that. Instead, double down on those skills that A.I. can't pull off, yet or ever. They're uniquely human, and they're the same skills that have been used to tech-proof jobs for decades.