Without AI… it could be worse.
If AI wasn’t “taking jobs”, these other factors might really hurt.
What if it could be worse without AI?
First, a disclaimer. The fear over dystopian outcomes and the need for safeguards is real and warranted. Some smart regulation, yes please!
But what about the other waves happening around the world of work right now? From productivity flatlining to generational shifts, a jobless AI run world isn’t the only possibility for what’s next.
There’s real potential that this all converges at (roughly) the right time. Here’s why.
Productivity. Are we getting enough output now?
What is “productivity”?
In general terms, productivity is the amount of output produced per hour worked.
For example: a single hour worked in 1920 would produce less than a single hour worked in 2024 due to technology, process, and other improvements.
What’s going on with it?
Over the last completed business cycle (from 2007–2019), productivity growth averaged roughly 1.5% and has continued to slow. The prior cycle was over 2% with some time in the 1990s reaching 2.7%.
Consider cases like in the UK. From the 1970s to today, workers are 2x productive. You can see the slow down dramatically when you consider that in 2024 workers are only 1% more productive than in 2007. That’s a long time to only scoot up 1%.
It’s not just employers squeezing workers for more output.
When productivity stops or declines, the expansion of the economic “pie” also stops.
That can ultimately lead to falls in standards of living for everyone. We all know what it feels like in a “boom” vs. a “bust”. Does the word “stagnation” imply anything good to you? Yeah, because it’s bad! Slow or flat growth doesn’t mean that things stop being expensive. In many cases, it means that things get or stay more expensive. For example, without productivity gains a TV wouldn’t be cheap. A TV today is 99% less expensive than in 1950 in equivalent dollars. But thanks to productivity (through technology) gains, we get significantly better tv’s for way less.
Productivity is not the only measure of well being. But we should notice when it starts stalling.
AI exposed industries have thus far seen 4.8x higher growth in labor productivity.
All leaders care about is “efficiency”, right?
OMG if productivity is falling employers must be freaking out because all employers care about is efficiency right? They must be totally ready to wipe out teams in favor of AI.
Efficiency is having a moment with layoffs in the news. Expensive capital (see: high rates) and many orgs have an over-hiring hangover from 2021/2022’s hiring bonanza. It’s also a very tempting time to trim down a team when “everyone else” is doing it.
But this sentiment makes it seem like efficiency is all people at work think about.
And yet, there are thousands of hours in every organization that could have been automated already without AI. With just regular code. Or plain ole process improvements. And they’re not. Not because it wouldn’t be more efficient. It would be!
Humans at work have other real incentives:
Patterns and habits. “That’s the way we’ve always done it” is a trope because repetition is safe and easy. Our brains are set up to repeat the same patterns and conserve energy.
Personnel real estate. Who gets credit in a corporate environment for crushing it with 5 people? Hardly anyone. Instead, more employees means more important. A VP managing 1000 people is highly incentivized to keep a big team.
Productivity vs. Productivity Theater. How much time is really going to output vs. summarizing, planning for and reporting on “output”?
Consensus > Momentum. Deciding means responsibility. And responsibility means risk. So it’s often easier and safer to make sure that a group of people makes the (usually worse) decision. See: Innovator’s Dilemma.
Stories like this coffee conundrum happen every day:
My company provides coffee machines on every floor but charges 20 cents per cup (except for “meeting coffee,” which is free). There are lists. People on every floor whose responsibility it is to refill coffee, sugar, and milk. Deputy people for this job….
At some time someone made an official “proposal for improvement” to eliminate the charge for coffee, the lists, the cash boxes, and the whole system: Have a single person whose job it is to refill the coffee machines daily and be done with it. There was a short calculation of how much time and effort could be saved. (A lot.) That proposal has gone through the improvements committee (yes, that’s a thing), the sales people, the union, the CEO, and back to the improvements committee. It is still under consideration after 18 months.
Is that to say no one will implement AI because they all want to keep big teams? Of course not.
There will absolutely be adoptions and smaller teams doing the work of previously bigger ones. And smaller companies will gain in competitive potential. But we can’t ignore that there are absolutely more to a decision at work than “Is it more efficient?”.
Consider our other systemic friend, bureaucracy.
In 1980, a report came out with alarming news - the US would soon have a surplus of doctors. The government reacted by limiting financial support for residency programs and thus began our current system in which there are more med school graduates than residency spots. In theory, the idea was to keep medical costs down by not flooding the system with too many doctors. In reality, we’ve created our own shortage resulting in massive salaries. Those already inside of this system are incentivized to keep it.
I once worked at a company where we had a very kind and smart employee whose entire role was to keep us OFCCP compliant. This was a position just to meet the compliance requirements to be a government contractor. Well intended for DEI outcomes. And bonus, a job was created.
“Where once universities, corporations, movie studios, and the like had been governed by a combination of relatively simple chains of command and informal patronage networks, we now have a world of funding proposals, strategic vision documents, and development team pitches—allowing for the endless elaborations of new and ever more pointless levels of managerial hierarchy”
― David Graeber, Bullshit Jobs: A Theory
I haven’t seen many signs of bureaucracy slowing down. Have you?
Frustrating? Yes. True? Yes.
85M reasons the labor market needs AI.
In our lifetime, we’re unlikely to unravel our societal build around the idea of a “job”. Politicians think about jobs. Jobs are how we survive. Jobs are way too often our identity. So, a healthy system has just about enough jobs for enough people with only a few percentage points of difference.
Did you know our current AI-free trajectory is actually kinda scary? WEF and other studies estimate that by 2030 on our current trajectory, global labor shortages will reach 85M people.
How so?
In the United States, the ratio of prime working-age people (25-64) to retirement-age people (65+) is currently around 3 working-age person to every 1 retiree. By 2030, that will be 2.4 working age people to 1 retired one. And by the end of the century, 1.6 workers for every one retired person.
Due to economic challenges and/or longevity gains more workers may continue working beyond 64 but the core issue remains that more workers are leaving the workforce than joining.
At the same time, labor force participation rates are falling. This includes growing challenges for caretakers who can’t find affordable child care or elder care. By 2030, the labor force participation rate is expected to drop 3% to 60.4%. That’s a change of 1.8 million workers over less than a decade. US workers also are working fewer hours than before, reducing the pool of labor even further.
It’s easy to see that this won’t be fixed near-term. Birth rates are falling globally with the US seeing a 23% drop between 2007 and 2022. So, there’s no next big wave of population growth here.
Immigration has been recently buoying the US population with even recently accounting for 65% of the population increase from 2021 to 2022 after a pandemic lull.
But this system is far from perfect. For example, consider the OPT Visa. A student comes here for an advanced degree then can work and gain more skills and experience within the US for 1 year (with some STEM exceptions to a few years). Then they’ll need to find someone to sponsor an H1B (giving them 7 more years at considerable cost to an employer) or leave… taking those skills with them.
Without major structural changes (quite a tall order in a time of increasing bipartisan divide), immigration would not be enough to solve this problem.
At the same time, the jobs themselves are changing.
Ok but what about people who are really struggling now to find work? There’s a silent struggle happening beyond “low unemployment numbers” especially for white collar workers.
The “half-life” of a skill is the amount of time that it takes for a particular skill to lose half of its value. We used to see a half-life for skills of 10+ years. That’s now down to 5 years and declining.
It’s important that we differentiate between skills and talent. When we discuss skills in this context, we mean the tactical knowledge used to perform a job at hand.
The tricky part here is that increasingly employers tend to want to hire people who have been there/done that in a professional environment. But training programs today focus on practicing the skills in a training environment.
When budgets are tight and resources are limited, the fear of hiring the wrong person begins to outweigh the pain of the open role. And so teams begin to invent more and more hoops for candidates including wanting more proof of experience rather than more proof of skill acquisition.
This creates a market tension in which ML engineers have their pick of jobs but other roles might be struggling to find steady employment. But just saying you know machine learning (which you really might!) isn’t always enough for an employer to make the hire. The second job is always 100x easier than the first in a career transition.
These supply and demand differences also create wage compression in which a candidates sees “market compensation” as their last salary and an employer thinks they have equivalent options at lower salaries. Their views of the value being created by a role shift and it takes time to find the new equilibrium.
But what if AI could hold a solution to that tension? What if someone could learn in a way that was built for them and more easily replicate the real world experience that employers are focused on? What if we could more readily truly understand someone’s capabilities beyond a conversation packed with human bias and room for error?
But people can’t possibly handle change can they?
Are you less busy now than you were 10 years ago?
But don’t you use more technology?
Technology changes, advancements and incorporation into our work are not new.
Changes to skills, duties and the kinds of jobs we have are not new. By 1980, the majority of jobs were ones that didn't exist in 1940.
Fear isn’t new either.
Consider that theater owners were once quite upset about bicycles:
The word "Luddite" comes from British textile workers in the 1800s who worried new machines would take their jobs. And they did. But it also created new jobs. And the new jobs were year round instead of seasonal. Those new job workers banded together and early versions of unions came together ultimately ushering in (hard fought) better working conditions and improved economic productivity.
We’re already making adjustments within the system. Enrollment in schools for the skilled trades are at their highest peak since 2018. Why? Demand and job prospects.
75% of knowledge workers are already using AI at work.
We’re acting as though this is a far off dystopia when the reality is that it’s already in motion and we’re already making adjustments.
Yeah, but change hurts.
Am I saying an efficient AI enabled/run startup couldn’t or won’t displace a “safe” Fortune 100?
No. That could certainly happen. There is a destructive quality to any evolution. Jobs will be lost. And job loss sucks. Job has health implications. Job loss has mental health implications.
But we also can’t avoid it by pretending this isn’t happening. We have agency to take action now. We have time to rethink reskilling now. We can face our own demons around change starting now.
Because the choice was never between the good ole days of back-then and the horrors of what’s next.
Thanks to slowing population growth and productivity, without AI we could have easily found ourselves with an overburdened working population, an underserved elderly population and a real tumble in overall economic growth. All converging at the same time. That doesn’t sound idyllic to me.
All this to say, what if it could have been worse without AI?
