The job market is in flux. On the one hand, employers are expecting new hires to bring an ever expanding AI skillset to the table, on the other, jobs for junior- and in some cases mid-level positions seem to be getting scarcer.
It is no secret that AI automation will require a significant shift in how the workforce is utilized. The writing is on the wall: junior jobs are disappearing, and junior employees are expected to fill mid-level specialist jobs with the aid of AI. Entry-level workers are expected to use AI to jumpstart their careers and skip a crucial level of early skill acquisition. What does it mean for the generation entering the job market?
Who fears the AI ax the most?
Employees already feel AI breathing down their necks. According to a D2L study in the US, 43% of full and part-time workers are apprehensive about being replaced by employees with stronger AI skills within the next year. The youngest generation in the workforce, Gen Zs are the most worried: more than half of young employees (52%) fear for their jobs because of the AI revolution. Millennials follow them closely with 45% expressing similar concerns. The older generation, Gen X (aged 44 and above), shows far less trepidation about the future of their employment, with only a third of them expressing fear over being replaced.
It may seem puzzling that the youngest employees, often referred to as “digital natives” are the most worried about what the AI revolution has in store for them. Perhaps they understand the technology well enough to see its true potential, beyond the current scope of image generators and Large Language Models.
They may also realize what their Achilles' heel is: they have not yet had the time to acquire real professional experience and the skills that come with it. And what’s worse – no one is willing to teach them.
How do we climb the ladder if the first few rungs have been removed?
Skilled jobs without skill
Skills are built over months and even years. Granted, some specific tech know-how can be taught over a single 2-hour webinar, but broad-scope skills are accrued over 10,000 hours of diligent work, as described by Malcolm Gladwell in his bestseller book “Outliers.”
Can we really devote 10,000 hours to honing a single skill these days? The first problem is that with the fast pace of change in technology, few have the luxury to bet on a single skill to propel their careers. We have had to learn to be adaptable above all else, being ready to switch from one technology or platform to the next and the next after that.
However, frequent switching is not conducive to in-depth understanding.
From MS Word to gen-AI: an editor’s tale
Here’s an example: Years ago, back when I worked in a physical newsroom, I honed my proofreading skills to a point where I became a "superuser" of MS Word. I knew all the shortcuts and advanced functions. I could edit and proofread text faster than anyone. As befits a superuser, I did all the work with the keyboard, never even touching the mouse.
Then my newsroom switched to a different text processor, an online one. All my key combinations became invalid in an instant. I had to relearn basic key combinations, like checking for word count. Granted, I gained other useful features with the new software but for a while, I became much slower at editing and proofreading.
Then came more advanced editors, like Grammarly and other AI-powered tools, which made my job notably easier. Once I mastered them, I became even more efficient at editing. When I transitioned to marketing, I started using AI SEO-optimization tools, at which point my writing skills became a secondary consideration. I learned new skills: how to include a set of keywords the right way, how to follow a set structure and how to deliver concise, informative and persuasive copy. All of these skills came on top of the writing skill, which still serves as a foundation for my profession.
When ChatGPT came online, I too was fearful. I wondered: Is there still demand for all of my hard-earned experience and skills? If a Large Language Model can produce marketing copy in an instant, translate between any pairs of languages, and edit for grammar, style and punctuation, many of my skills that took years to master have clearly become outdated.
Was it worth spending 10,000 hours to learn them?
LLM quality assurance - AI skillset in use
I think many of us had that moment when it dawned on us that the skills we'd poured countless hours into mastering were no longer marketable, because a gen-AI model could perform many of our tasks better. Or at least faster.
Then we all realized that gen-AI is not infallible and it makes mistakes, not to mention sometimes pulls data and sources out of thin air. Not for nothing did Dictionary.com name “hallucinate” the word of the year 2023.
Many professionals heaved a breath of relief when they realized there is still room for them in the newly minted “AI quality assurance” jobs. And this is where we are now. We use AI on a daily basis, we keep finding new applications for the technologies available, we learn new ones, switch back and forth until we find the best fit for our task. We are uptraining existing models and testing new ones. We are getting comfortable with gen-AI.
There is one caveat. The “we” in that example refers to employees with significant experience to lean back on.
How good at spotting errors and correcting LLMs are people who have never had the chance to learn writing, editing, translating or coding the hard way, making errors and correcting them?
Once you’ve spent years poring over your own and other people’s work you instinctively know where errors may occur. Switching to curating and editing AI-sourced content is not that far removed. You quickly learn where AI tools trip up and put those bits under a microscope.
As someone who works with LLMs every day, I know where an AI translator can fail and which language structures can lead to misunderstandings or hallucinations. I have enough experience correcting mistakes that recalibrating my toolbox to search for AI screwups was relatively easy.
The key word is experience.
What if your resume lacks the years of editing, coding or doing research the old-fashioned way?
You get your first job, which comes with a suite of the latest gen-AI models and your boss says “go.” What then?
There's no need to panic. Studies show that your productivity will jump very quickly. According to Business Insider, gen-AI tools are the great equalizer in the workplace. People who were underperforming in their jobs suddenly caught up to their more productive colleagues when given gen-AI tools.
Gen-AI is great at writing copy and code, translating, summarizing, sifting through data sources etc. Occasionally it will fail, but on average it speeds up simple, repetitive work tenfold or even more. Junior jobs are where gen-AI shines the brightest. But that’s just it.
Getting a leg up
AI can give a junior employee a leg up and make them more efficient very quickly. But it won’t teach them the basics, it will do the work for them. It’s like always copying your friend’s homework and getting straight “As.” But then you want to move on and go to college. That’s when you realize learning advanced skills without mastering the basics is oftentimes impossible.
Can a junior employee skip the initial part of the learning curve altogether, gliding over the bumps in the road on feather-and-wax AI wings? Or will they forever remain just that: juniors with really cool tech at their disposal?
Common sense benchmarking
It takes skill to realize when an AI is underperforming. It takes experience to know where to look and how to assess the work product AI tools deliver. When we see an artificially created copy or code, we contrast it with a human-made benchmark we have in our minds to see if it passes off as “acceptable.” But we need to acquire the benchmark first.
If all you ever learn from is AI output, how are you supposed to learn what is good code or good copy? How would you know if a novel you’re reading is good if you’ve only read plot synopses?
Apprentice wanted
Knowledge that comes from experience allows us to look at the world with a critical eye, discerning what is valuable from what is a clear waste of time.
We learn a lot by doing, by trial and error, by failing and trying again. We go through a series of jobs to acquire skills that education alone could never provide. And this process starts with entry-level positions.
As the use of AI for entry level jobs expands, we may have to devise an alternative training scheme for the new generations entering the job market. Perhaps something more structured than “on-the-job training” that used to work in the past.
Maybe we will see the return of apprenticeships: seeking out promising individuals and expending time and energy to teach them all the intricacies of a profession based on the promise that they will become proficient. Alas, this also implies that the less fortunate ones will be given AI crutches and will remain at their entry to mid-level positions without much hope for advancement.
Commenti