The conventional wisdom is that AI will redefine entry-level jobs, improving employee productivity while freeing them to train for more senior roles.

No, it will destroy those job opportunities, and use human workers to help it do it.

A better trainee.

Entry-level jobs have been all but a right of passage for generations of employees. Roles that require little experience and rely mostly on rote actions are perfect for people who’ve only learned about the world in school, or simply have never worked before.

However exotic or complicated the business model, there are always tasks that can be distilled into a series of “if this happens, do that” commands that newbie workers can follow and, therefrom, launch their careers.

AI are perfect for those roles right out of the, er, box.

It’s already a reality, especially in businesses that have to screen arrays of data and look for patterns and standouts (which means most businesses). People walking in and out of commercial buildings. Applications for jobs. Registrations for discounts or other perks. Vehicles in parking lots. Some fact or insight hiding in a stack of documents.

These are jobs, broadly speaking, that could be given to people with little experience. Make the kid do it was often the sigh of relief for more seasoned employees when the newbie worker showed up and could take over some onerously repetitive or uninspiring responsibility.

No more. Make the AI do it.

Same goes for tasks generating the words used by marketing, PR, HR, and most other corporate departments for promotional and reporting material.  Businesses have a voracious and insatiable hunger for this stuff, and AIs (especially generative AI like ChatGPT) are really good at coming up with it.

Now think about non-corporate entry-level jobs like manning retail counters, serving fast food, even babysitting. Such gigs have traditionally trained people to understand requirements for taking responsibility and on-time performance, among other core skills, if not acquire activities related to specific careers. AI are assuming these roles, too, though the transition has been limited by their general lack of dexterous robot bodies.

But AIs are already monitoring grocery store shelves, vacuuming carpets and moving lawns, and making pizza crusts. Attach more advanced actuators to the sensors they already possess today and they’ll have the movement capabilities of any human newbie worker.

Employees are training their replacements from Day One.

There’s an ugly not-so-secret truth to working with AI: When you learn from it, it learns from you. In fact, AIs only forward, or share information with people that was at some earlier point created by people. It’s an amazing engine for gathering and repurposing.

So, every time entry-level workers learn while using AI, the AIs learn, too, and then they share it with all the other AIs to which they’re connected.

The same goes for any other activity in which we engage through a digital interface, whether at work or play. Clicks and posts on websites. Using credit cards. All that data gets gathered and repurposed by AI do understand those activities better, faster, and more reliably than we do.

Add all of the surveillance of our movement in the physical world (where we go, how we drive, when we do things), and AI are going to get better at automating evermore complex tasks into series of “if this, do that” commands.

When it comes to entry-level work, AI doesn’t help train people. People are there to train AI.

Competition that never rests or ends.

OK, so imagine that you’re an employer and you have a choice: Either hire some kid who is unskilled, perhaps unmotivated, probably in need of constant reminders that the work has purpose, and likely expects to be included in senior-level decision-making, or buy an AI that is the exact opposite on all of the above.

Or just contemplate how long that entry-level employee can hold onto the job attached to a co-worker that on top of everything else doesn’t need healthcare or even proper lighting or HVAC.

Human beings won’t have a chance competing for those jobs, let alone keeping them.

But never fear! Another argument for the benefits of AI is that it’ll free them to engage in more advanced, thoughtful work activities. This assumes that they’ll be able to get the jobs in the first place, but even if they do there’ll likely be AIs installed at all of those new “upskilled” positions waiting to be trained.

You can see the pattern: People get jobs, train AIs to take over, and then do their best to prepare for new jobs that AI will assume once people have trained them. Don’t forget that networked AIs share what they learn with one another with a relentless efficiency that no individual initiative or collaborative study-group could ever touch.

No matter how smart or inspired people might be to improve themselves, they’ll lose to AI.

Closing the door to employment, not opening it.

What am I missing?

Oh yeah, the argument that AIs taking over existing jobs will create entirely new occupations for people that we can’t yet imagine. The closest thing we get to an example is a reference to jobs servicing AI in some way, along with vague references to the new employment opportunities that arose when machines replaced or augmented work done by people during the Industrial Revolution.

No aspect of this argument holds any water.

Any newly-created job would likely be at risk of AI or some form of automation, by definition, so people would always be striving and losing to AIs competing with them. As for comparisons to the Industrial Revolution, looms didn’t get smarter over time, and the new jobs creating by displacing people with machines didn’t emerge until a generation or two after the initial job loss and didn’t replace its economic or social impacts.

Belief in the authoritative and restorative powers of technology is a fantasy, and using it as an explanation for what is happening with AI at work, including its impacts on entry-level opportunities, isn’t remotely legitimate.

It’s a lie.

[This essay originally appeared at Spiritual Telegraph]