Ep. 43: AI Isn’t Coming for Your Job
In this episode I make the case that AI isn’t actually coming for most people’s jobs, backing it up with real examples like Klarna’s AI customer service reversal and Amazon’s quietly-human Just Walk Out technology. The central argument is that when human interaction is baked into the product itself, that part of the job simply can’t be outsourced to AI. My hope is that this episode serves as a springboard for a bigger discussion about destroying community in the name of efficiency, and the myth of rugged individualism.
Dystopian Dramatizations
Y’all, I spend a lot of time on social media. Like, a lot. Admittedly, too much time, but I’m not changing. Either way, basically every single piece of content made these days is optimized for clicks, which means it’s dramatic, it’s polarizing, it’s got no nuance, and the goal is to make you either angry or scared…because that keeps you on the platform.
So, yes, a large part of the reason I wanted to do this episode is because I’m honestly just sick of the fearmongering and dramatizations and dystopian predictions.
CEOs Are Doing the Math Wrong
Since pretty much minute number one, the headlines surrounding AI have focused on AI taking people’s jobs. Don’t get me wrong, I think so many white collar jobs out there are absolute bullshit and I’d love to see them go away and have all that money put towards things that are actually going to help society, but AI ain’t taking these jobs. I honestly don’t think it’s going to end up taking AND keeping many people’s jobs.
Yes, people are currently being laid off because these dumb CEOs believe they can decrease expenses (aka cut salaries) and increase profits and shareholder value by firing humans and deploying AI. But any of you reading this who have actually used AI for any task with any level of complexity know how the story is going to end.
No guessing required, we’re already seeing companies backtrack on this and rehire people!
Klarna laid off 1,800 people when they replaced their customer service staff with AI chatbots, only to have the CEO admit that it led to reduced service quality and customer satisfaction. They have since rehired human customer service staff.
In the IT job market, one report showed that of the companies surveyed (those that enacted layoffs secondary to AI optimization), 50% were expected to rehire for the same roles and profiles by 2027.
The Task Is Not the Entire Job
Customer service is a role that many companies chose to outsource to AI very early on, and I intentionally referenced Klarna’s catastrophe because it so clearly demonstrates the bigger point I’m trying to make: AI isn’t coming for your job, because most jobs that these whack ass companies are looking to outsource to AI aren’t merely tasks.
In many cases where a human is involved, the task is not the entire job.
Using customer service as the example: the task of identifying the issue and remedying it is only part of the job. Another part of the job is the human component: expressing remorse, demonstrating understanding, apologizing, sympathizing, doing something about the caller’s emotions, taking a personal interest in the situation and going the extra mile to solve a problem that is out of the ordinary, etc., etc., etc. Whether you agree with it or not, these things are part of the job (and are grounds for paying these people more), and they simply can’t be outsourced to AI.
So of course Klarna saw reduced service quality and decreased customer satisfaction when they had the robots take over: Only part of the job was being performed.
What Is the Actual Product? What Is the Actual Job?
When folks say that “AI can do X,” they’re talking about a task. We must remember that a job is not just a single task. And if it is, well then real talk, yes, there is a good chance it can get outsourced to AI.
So the real question here is: What is the actual product and what is the actual job?
For a lot of work, especially what I’m guessing that you, dear reader, are personally familiar with (especially if you’re in the service industry), the product itself is deeply intertwined with human interaction. Think: therapy, coaching, personal training, teaching, barbers, restaurants.
With these examples, we see that the human isn’t just the delivery mechanism, the human interaction is absolutely part of what the customer is paying for.
That Word: Desired
AI cannot replace desired human presence.
Worth noting, that word: DESIRED.
When we want a human and perceive them to be an integral part of the experience (aka, an integral part of the product or service), we won’t accept AI. When we don’t desire a human to be part of the experience, because we don’t consider them to be an integral part of creating said experience, robots can replace them.
A perfect example: an ATM. For simple transactions like just depositing or withdrawing money, we don’t need a human or a pleasant greeting. They are nice to have, perhaps, but not need to have.
A more polarizing example, but one that we’re currently witnessing: Waymo. Folks are willing to pay more for driverless experiences, because in many cases they actually do not even want a human to be part of the experience; a common sentiment expressed by female riders and other demographics that are often the victims of violent crimes committed by men.
And then there’s self-checkout at the grocery store. A bit of a tangent perhaps, but I do think that self-checkouts give us insight into what will likely happen in the future: a hybridization of many things, where customers and consumers are given choices. I think this will even be a thing with domains like music. People will always want to go hear live music performed by humans, but I think they’ll be far more open to streaming AI-generated music. Honestly, we’re already seeing it.
Something Worth Sitting With
At what point are we just contributing to self-isolation and the destruction of community?
One of the reasons I started Prompting Curiosity was to have a space to articulate these things and start conversations, because there is so much more to AI than just talking about the technology itself.
As we think about the tasks that make up a job, and the parts that are uniquely human, there are so many things to be considered when deciding whether or not a human is an integral part of the experience. I do think that much of the time we inherently know, and we can feel it. But, it’s also worth noting that what we’re willing to accept (just, in general) and what we enjoy can definitely be shaped by what we’re continuously exposed to.
It Always Comes Down to Money
Basically the only variable these companies are looking at as it relates to what is and isn’t uniquely human, is money. Nothing else.
Ironically enough, it will likely be money that ultimately drives companies to rehire humans, because unfortunately it will always be cheaper to simply underpay a human to perform the task. Global capitalism both requires and creates exploited cheap labor. The cost of AI is rapidly rising as VC funding is rapidly depleting, and companies are starting to see the real numbers as AI companies are being forced to do real math. All of this has led to new questions circulating as to whether or not it is in fact cheaper to use AI.
Honestly, none of this is new. Amazon’s Just Walk Out technology at its Fresh grocery stores was touted as being powered by AI, but it actually relied heavily on manual monitoring by around 1,000 workers in India. In 2022, 700 out of every 1,000 transactions required manual review by that team. Talk about some Wizard of Oz shit. Amazon eventually scrapped the whole thing.
Yes, the Amazon example demonstrates an actual technological inability to perform the task, which I do think is the case with some of the things that companies are currently looking to outsource to AI. However, I do foresee this gap closing over time.
The bigger point I wanted to make, and the gap I don’t foresee ever closing, is that for many jobs, human interaction is actually integral because it’s baked into the product. As such, AI isn’t coming for that job anytime soon.
Where AI Actually Fits
Like I’ve said eleventy billion times before, where AI truly shines and fits in best are the administrative, repetitive, behind-the-scenes tasks that eat time. The goal, IMO, shouldn’t be to outsource these tasks to AI so that humans have more time to do more work, but rather, outsourcing to AI so that humans can spend more time being human. Let’s all go touch some grass.
So, if your job has parts that are deeply human, and if human interaction is part of the actual product, AI isn’t coming for your job.
Nuance here: AI isn’t coming for your job at scale.
There will always be individuals who don’t value, or don’t want, human interaction as part of the experience. In those cases, yes, AI could be tagged in. But just remember: we’re a social species by nature. One-off preferences are not at all representative of the whole, and rugged individualism is a cultural myth.
Humans need and want other humans.
How I Used AI This Week
Each episode I share a quick example of how I used AI that week.
This week I used the DocHub dupe PDF filler that I built out with Claude Code a few weeks ago. I’m referencing this “repeat” because something I want you all to be aware of when it comes to vibe coding is that maintenance is a very real thing.
I love vibe coding my own personal web apps and I really do think it is the ultimate use case for vibe coding. But the reason why vibe coding won’t kill SaaS (SaaS = software as a service; think ConvertKit, HubSpot, Notion) is because just like with a house, or a car, or anything you can think of that has been built, over time things can break. My DocHub dupe worked just fine, but I have had to use Claude Code to troubleshoot some issues with my podcast automation in the past, when software updates caused a part of the flow to break.
In my opinion, this is by no means a reason to not vibe code your own solutions, but it’s definitely something to be aware of.
Speaking of vibe coding, I put out feelers on my other newsletter, and I will be hosting that small group intro to vibe coding training. I gotta figure out logistics (aka the date), but it’s definitely happening (ideally by the end of this month), so keep an eye out for registration and definitely DM me if you’re interested!
Da Wrap-up
I’m not denying that plenty of jobs are currently on the chopping block courtesy of “AI optimization”, but I truly believe that for service providers, AI isn’t coming for your job because most of these jobs aren’t just tasks, and human interaction is often baked into the product itself.
Companies are already learning this the hard way (hi, Klarna), and the math on AI costs is starting to catch up with the hype. Where AI actually fits is in the repetitive, behind-the-scenes stuff, with the ultimate goal of freeing up time for humans to be more…human.
As always, endlessly grateful for you and your curiosity.
Catch you next Thursday.
Maestro out.