When Alexandra Samuel started her career in the 1980’s, she followed what was at the time one of the only paths for women seeking white-collar work: secretarial jobs. In high school, her mother had bribed her to learn typing, and Samuel spent her high school and college years churning out correspondence, reports, and memos. “As I kind of worked my way through these jobs, you know, little bits of things that weren’t secretarial would get thrown into the mix so that by the time I finished college, I actually had a resume that showed I was capable of having a junior professional job,” she said.
But in today's digitized offices, those sorts of clerical roles have all but vanished. "There was so much paper in the world that I grew up in, and paper just isn’t very important anymore," Samuel reflected. "That erased a whole category of entry-level work."
Today, with the emergence of AI, it’s not just one category of entry-level work that is being disrupted—it’s hundreds. Writing unit tests and boilerplate code, sending out cold emails, and drafting up legal memos currently form the bulk of early-career training for junior software engineers, sales reps, and lawyers. But they’re also the very tasks likely to be automated first.
At first glance, the implications may seem foreboding: new technology could make it much harder for recent graduates to get jobs, not to mention the risk of credential inflation to compensate. On the other hand, companies will now be more incentivized than ever to upskill their junior workers—fast.
“We’re definitely going to have people who are able to accelerate the learning curve [and] go into what we now think of as a mid-level job, much more rapidly and earlier in their careers,” said Samuel, who is now a journalist and speaker who focuses on remote work topics. In other words, the way we learn on the job could change drastically, sure, but the disruption could introduce innovations in the way we think about training, learning, and development.
“We’re going to have people who are able to accelerate the learning curve... much more rapidly and earlier in their careers.”—Alexandra Samuel
As the modern culture of knowledge work has rapidly evolved, the way we learn and develop on the job has struggled to keep pace. With work environments shifting alongside the onslaught of new technologies like AI, the methods in which we take on career training are ripe for a collective audit. For some, this starts with asking the right questions. As Shriram Krishnamurthi, a professor in the computer science department at Brown University, puts it: “What are the cognitive factors [of training] that have worked until now? How can we get the painful parts to go away? But then also, how do we recognize that the painful parts might actually be necessary for future success?”
Access to mentorship is more important than ever
Human learning, as we currently understand it, is predicated on pattern recognition and repetition. Every time we repeat a task, the neural pathways associated with that task are strengthened. Historically, the early years of a knowledge worker’s training are often filled with repetitive, relatively routine tasks for this reason—whether it’s junior marketers building slide decks or entry-level administrative assistants jotting down meeting notes and organizing calendar schedules.
Yet even the simplest tasks can benefit from another person guiding you through the process. Imagine mailing out a letter: you can throw someone in front of a pile of letters and have them figure it out on their own or you can actually provide instructions. The latter is obviously going to be faster than the former. This is why the number of repetitions it takes to learn something is not only heavily dependent on whether the learner is doing deliberate exercises and is properly motivated, but also has access to mentorship and coaching.
Samuel is a strong advocate for organizations shifting to a more deliberate mentorship model, with actual training and coaching. It makes sense that as more working environments are hybrid or remote, the way we address training should evolve and cater to the landscape. “The reason we’re so obsessed with this idea of on-the-job learning is because we’re very bad at providing true learning opportunities, where you actually have to take 20 minutes out of your day to teach people something,” she says. That could happen just as easily in remote settings versus in-person.
“Thanks to AI, every person in the world is now able to have a teacher in many different fields working with them one-on-one on whatever they want.”
So, what do these “true learning opportunities” look like in practice? According to Samuel, this could mean one-on-one technical mentorship programs where junior employees are paired with senior mentor programs that answer their questions for an hour each week. It could also look like organizational overview sessions where new hires are explicitly instructed on the structure of the company and what each department is responsible for. There could also be performance incentives for senior employees who spend time explicitly teaching.
The key is being deliberate and proactive about upskilling employees through structured learning and mentorship, which can help develop talent more quickly than ad hoc, on-the-job learning alone—especially in remote work environments. This is also an area where the advent of AI will help. “We know that individual instruction is more effective than classroom instruction, and thanks to AI, every single person in the world is now able to have a teacher in many different fields working with them one-on-one on whatever they want,” Samuel says.
The impact of AI and early career training
The evolving culture of modern work is being shaped by tools like AI—and vice versa—as new technologies and more flexible working environments develop in tandem. Shadowing a colleague throughout the day or getting pulled into a meeting because you happen to pass by is more challenging when everyone works different hours and is distributed around the globe.
With fewer jobs even requiring ever stepping into a physical office, early-career knowledge workers are missing these seemingly insignificant but crucial touch points and they often enter the workforce with little context or tangible examples of on-the-job training. “Orientation” is usually just a litany of software demos and compliance videos. But these kinds of introductory programs can be refreshed to meet the challenges of remote work with tools like AI.
Suppose you are a data engineer working at a software company. Right now, the only way to really know how to react if there’s a data outage is to actually experience one. Hopefully, the company you’re working at only has a few data outages a quarter, so your pace of learning is slow. But what if part of your training is “mock outages”, where the corrupted data points are generated by AI? The concept can be extended to other fields: junior therapists can conduct practice sessions with voice AIs, radiologists can look at AI-generated scans of rare diseases, and sales reps can be trained by virtual customers with specific niches and concerns. Such responsive simulations provide contextual learning through practice.
“Even as new technologies come into the fold, they're not really putting people out of business as much as they are giving us more skills.”—David Malan
There’s an argument to be made that refreshing job training with AI is just the tip of the iceberg when it comes to the potential impact of this technology to early careers. AI could be used to power new programs that reinvent how a junior employee experiences a job orientation. But when used most effectively, these tools can potentially help people skip levels altogether by taking on more of the repetitive, routine tasks that often characterize entry-level positions.
This means that, rather than spending all their time programming, junior engineers could spend more of their time solving more complex problems. Sales reps could spend time analyzing sales data and identifying new opportunities rather than cold-calling prospects from a list. Lawyers could focus on drafting substantive legal briefs and conducting research to build cases rather than just formatting documents and compiling references.
AI isn't the first technology to make an impact on entry-level jobs. In the 1970’s and 80’s, the advent of ATMs automated most routine tasks of bank tellers—until then, a stalwart entry-level gig for working in finance. But the new technology, while greatly reducing the overall need for tellers, didn't make the role totally obsolete. Tellers continued to handle more complex customer needs that required human interaction. More importantly, the automation of basic transactions like deposits and withdrawals enabled banks to shift their entry-level hiring to roles that better suited the new landscape, like customer service reps and risk analysts.
The same has been true over the last ten years, in the world of IT operations. Entry-level server maintenance and internet connectivity duties have merged into "DevOps" roles, where developers can now provision cloud servers themselves. This has meant the death of the “database administrator” role, but it has also made individual engineers more productive and multifaceted.
“I think that we already have ample evidence that even as new technologies come into the fold, we're sort of assimilating them and they're not really putting people out of business as much as they are just giving us more skills ourselves,” said David Malan, a professor of computer science at Harvard who teaches the university’s introductory computer science class.
The job market is flexible, and what employers expect from entry-level employees is already shifting; job applicants report tech companies asking fewer algorithmic coding questions, which AI can often easily solve. “It's a lot more system design now, I think, where you have a conversation with the other person,” said Vicki Xu, a college senior studying computer science.
As for those lost secretarial jobs that Samuel cut her teeth on—they, too, have had a second, and even third life. The secretary role of the 1960s and 70s, which involved mostly typing, dictation, and administrative support, has evolved into more strategic positions like executive assistant and chief of staff. With basic administrative tasks requiring less dedicated time thanks to technology, assistants can take on responsibilities like event planning, client relations, research projects, employee surveys, and acting as an executive’s right-hand advisor.
Those jobs may look different than they did in the past, but the human elements remain the same.