Here’s How AI Will Come for Your Job
theatlantic.com
theatlantic.com
I think it's a red herring to focus on the "quantity" of work. If I can push a button and a thousand things are produced, I am not inconvenienced as a worker to the point that I should feel slighted by my employer for expecting more than if that button only produced one thing at a time.
The output quantity possible per unit of time isn't the thing that stops employers overworking or abusing their employees, that's what the law is supposed to be for.
AI adds very little to the conversation of increased productivity not increasing leisure time of employees. No amount of increased productivity through technological gains is going to suddenly make this happen. We need to decide as a society that the physical and mental health of all citizens is more important than maximum economic output and enact laws that enforce these limits.
People say it’s “the man” that encourages this out of their own avarice and greed, but my observation (partially by having been the man for some time now) is “the man” is a product of the system of beliefs, and is more subsumed by the delusion than most people. They certainly do encourage the belief in labor for labors sake - but more so because they do whole heartily believe in it themselves to the core of their being that they can’t see a functional society without everyone emulating the life they’ve led and that has rewarded them so heartily.
There was a time up until about the 70’s that universal basic income and a dream of a post labor oriented society was vibrant and fairly mainstream, with even presidents advocating for it. It got lost, particularly (IMO) in the conservative backlash against social programs that equated “welfare,” racism, and a bunch of other touch points, to drive out the notion we can function without an overwhelming need to toil. We firmly tied health care and housing and any form of security to a life of unending toil. Retirement has been eaten away to ensure people toil long after they would have. As productivity improved, we demanded more toil. What our feckless leaders diverted away can be regained, and I hope that AI ushers in an era of human toil that’s so utterly pointless we finally can see the fnords en masse.
It's important to guide smart, intelligent people towards meaningful work for the betterment of society. There isn't much separating a clever scientist from a clever criminal, beyond ethics and discipline. To produce for the betterment of society is ultimately a moral goal. The notion that work is its own reward is not poisonous to society in and of itself, rather it's beneficial. When taken to excess, of course it is harmful as are most things taken to excess. But the complete absence thereof would be equally harmful to society.
It is important for whom to guide these people? What are their motivations?
I think the paternalistic view that people need jobs pushing a pointless button to keep them from degenerating into Lex Luthor assumes evil is the natural state of man, and we need a constructed hamster wheel of society to keep the bad guys at bay. It hasn’t kept the bad guys at bay so far, and I believe the bad guys are very visible but a stark minority. We all have to spin wheels just in case there’s a bad guy that is stuck in a wheel somewhere too?
As we move to an era where, potentially, production improves without human input, maybe the betterment of society can be achieved in other ways than toiling in pointless corporate salt mines. Maybe artists can have health insurance? Maybe people can pursue labors of love for their lifetime, not laboring a lifetime in jokes they have enough money, health, lifespan, and soul left to do what they love briefly before they die.
Some people already do what they love. Great! Do it. But look around you more carefully and look at the culture of derision about modern work life. Almost no one loves modern work life across all callings. It burns us out. And when its done it casts us into a public retirement home or worse and hopes we die faster for actuarial efficiency. If we can’t fit into the corporate mold, it relegates us to poverty, disenfranchisement, food and shelter insecurity, and lack of basic medical care.
So, as a post scarcity society is possible, will we construct fake work to keep people chasing health care and other basics, or will we move beyond that? Will we open the door to full autonomy? Or will we do a PPP/WPA style guided employment by the state model? Or will we just turn a blind eye and let massive segments of society rot unaided until it falls apart? I suspect in 20 years we will know the answer.
It depends what happens when you push the button. Is all that happens really that you boss makes a profit, what is your company's business? Does it not have customers? Does it not do anything of benefit to anyone other than your boss? How could that even work, where does the revenue come for your salary?
It's called productivity because it's about producing. Somebody somewhere is paying for that. There's no mystery here, we know how this works. If productivity goes up, the product gets cheaper and demand goes up. The customer benefits, the company benefits and your society is now better off. That means in aggregate it can afford more stuff, including services, which generates jobs. That's how progress works. It's why we don't have thousands of unemployed horse drawn buggy drivers on every street corner because people can drive their own cars.
But if you can dramatically increase productivity per worker and for a given capital investment, your costs stay the same but your production is now higher and you can cut prices. That stimulates demand which mops up the available supply. Now very broadly several things can happen, it's more than the below but this isn't an essay.
One effect is the price of production drops, you raise margins a bit and cut prices which increases demand. The company makes more money and society gets more stuff at a cheaper price. If your product is itself productive, like it's a tool, then your customers can produce more with it more cheaply. Your company is more profitable, your society gets even more stuff. Your customers are better off, so they can hire more people, buy more products, pay people more. It's great all round.
Another possibility is demand goes up so much that in addition to the above your company can expand, increasing production further with more scaling effects, hiring more people, maybe also developing new products because it can invest more.
So productivity increases can wash through the economy raising all boats. That's partly what GDP growth is, and it's one reason why for example the lowest 10% by income in the UK now earns twice as much in purchasing power parity than they did in the early 80s. That sounds great, and it is, but middle incomes have stagnated the most and inequality has risen. Fairness matters, but raising all boats is worth doing. Let's not let the real social issues we do have blind us to economic reality.
Certainly once AI takes the task of improving AI, things go off the rails. But there’s a lot of steps along the way and the materiality of human labor I think is an early thing to go.
Only if I have money to buy it. That's why monetary velocity is a very important economic measure. If I have a 1 trillion dollar economy, and 999 billion is in the hands of one person, and the last billion is spread across 1000 million people, then the average monetary velocity is apt to be both low and only exchange between a few actors. It will be a very unhealthy economy.
Bill Gates and Warren Buffet don’t have huge money piles in warehouses. Their wealth is active in the economy.
Wealth is actually economic control. The people who have it run large chunks of the economy. Historically there have been two answers to the question of who do we want running most of the productive forces in the economy and directing investment in it, private citizens or politicians and government bureaucrats? The former is capitalism and the latter is socialism. Even the Chinese Communist Party eventually went with option 1. Who should run Microsoft or Ford, investors and professional managers, or the federal government?
Actually there is another option, Russian style oligarchy. That’s what real corruption looks like. Every system has some corruption and waste, the question is how we manage and limit that.
Think about shoes. Tens of thousands of years ago, it was just some animal skin wrapped around your foot. Maybe a wooden sole if you're advanced enough. Now we have incredibly comfortable shoes that you can use to run a marathon. I know this is a simple analogy but I'm hopeful that more advanced tools allow us to have more advanced products and services, which benefits everyone.
And how does AI fit into that equation? Will we regulate the use of all kinds of technology that doesn't clearly improve "all citizens' physical and mental health"? Surely the expiration date for that initiative passed eons ago...
I can't think of a single example where the adoption of AI does not involve diminishing a cost, like a avoiding a threat to human health or life. What's the role of AI if not to increase reward to the owner of the 'bot?
> People speak derisively about middle managers because their outputs are hard to define and monitor—they are viewed, sometimes unfairly, as a mere link in the chain
I speak derisively about middle managers for being short-sighted, petty, political, outright crooked and lacking any interest in the success of the business. In the software development world they're at best useless and often so full of spite and contempt for employees as to be net negative. I've had a few first-level managers that I really appreciated - genuinely good people - but the middle is just about all garbage. I think a lot gets blamed on executives when the middle managers are where the actual toxicity is coming from. However little or much your employees might care about your business, your middle mgmt almost always cares less.
Executives are hiring these middle managers and they don't do anything about the dysfunction although they have the power to do something. "The fish rots from the head".
Middle management is hired by and incentivized by the structures the top management installs (from HR to pay to morality & culture).
If your middle management sucks, it's because your top management are setting priorities that drive that outcome.
Top level management is often driven by short term incentives, they want dramatic big bang 'transformative change', while middle management is largely in it for their career. So they're pressurised for short term goals, but their actual career stability and advancement depends on middle to long term success.
What's really important is making sure the signal from the ground level gets to middle management. A lot of reporting process and bullshit is because middle management want to understand what the heck is going on and that's the only reliable way they have to get information. So politely and as constructively as possible go to them and tell them what you think they need to know. I'm not guaranteeing they will listen to you, but if you don't try then all they will have is reporting spreadsheets.
Yes -- it's uncomfortable. But it's also necessary.
Learn the tools well enough to be comfortable with them.
And get ready for more change. ChatGPT is already iterating: https://youtu.be/C_78DM8fG6E and it's not the only General AI tool widely available.
The google that added AI and turned their search results to an unusable adspam?
The google that released Bard which can't even reproduce a fact that can be simply looked up by their AI adspam search ?
The google that had a motto of "Do no evil" and later they decided to just sweep it under the rug ?
Yeah no I dont trust that google, thanks.
AI is a tool (at least for now). And like many tools, when used proficiently, it can boost your productivity and therefore make you better, not redundant at your job.
> is it right 100% of the time, no.
No human teacher is going to be right 100% of the time either. I suppose ChatGPT is a little more overconfident / doesn’t have a sense of tone or anything to assess confidence; even still, though, given how overconfidence is incentivized in the workplace, it’s not that different in this regard from an average coworker.
100% agree. Its like having a coworker who you can constantly pester with technical questions. I really love how you can ask ChatGPT follow ups to earlier questions and it understand the context.
At some point, I expect companies to put their internal Wikis/documentation into LLMs so you can learn and onboard staff faster. You'd probably need a large amount of data to learn on, but I guess some of the large companies have that.
Basically, anyone who thinks that having a tutor-like experience with a robot is worth trading away correctness and consistency, is revealing that they care more about getting an answer in a form they can easily ingest than actually having any depth of understanding of the topic at hand.
And before someone makes the silly argument that "well your human tutor could be wrong too!", all of the training data that trained chatGPT was written by humans, and so the potential human error is still the same. ChatGPT is adding a new source of error that is unpredictable and based on dozens of anecdotes I have seen here and on Twitter, none of the enthusiasts are double checking it's output for bullshit
I don't have a human tutor available and to get one would represent a significant cost for me.
Slightly wrong and available is worth far more than always right and unavailable, especially at the early starting point where you're unsure of how much you want to invest into an endeavor.
You can still do this today without a financial cost, because the choice was never an either/or between paying for a human tutor and learning. And just because there are now nearly free and shitty tutors available, it doesn't mean learning should be done via these means / is effectively done via these means.
A recent example of this was some crap CSM my wife was working on that had piss poor documentation from the vendor. As in the vendor documentation was just flat wrong and would never work the way they wanted it implemented. We dumped a code snippet off in GPT4 and it came back with "This is wrong, do this instead" which worked.
The key point here is it solved a problem in a system that is rarely worked on. Going outside the vendor documentation to learn the python/js interaction the vendor obviously did not understand would have represented a huge investment over the payout of correcting the output of a single page.
Try asking ChatGPT about more complex stuff. Design patterns, computer theory. I think you'll be impressed at the results.
This is the strength of ChatGPT, for some (and possibly many) topics, it does a really good job. If it didn't, it would not be getting all the attention it currently commands.
That's a leap. Chances are you'll still be seen as a producer and not a manager, so your job won't look like a middle manager of humans. In this case, your output is defined, and quality would be correlated with your prompting skills and any edits you make. You'll likely be in less meetings than middle managers, but maybe we'll have AI secretaries (avatars?) that go to those meetings for us. And yes, the pay will be way less than middle managers.
All that rebuttal to illustrate that job enhancement is a far more likely outcome for most of us, with only a small amount of lower skilled jobs at risk for complete obsolescence. You won't be joining the management caste unless you intentionally go for it. Plus you can still be a developer and glue things to AI systems. That's certainly not going away either.
If one side of that is instead AI, you won't have that issue.
No, it's more perversely ironic than that.
It's that the stuff we find "easy" and "boring" (like driving, walking around, picking up a cup and setting it back down) is just stuff eons of evolution has hardcoded into most of the neural network that compromises our brain.
The "hard" stuff we pride ourselves on is just the stuff we've begun to be able to do thanks to the smallest and newest addition to our brain, the neocortex. But the reason we think it's hard is because we suck at it, which is because it's an evolutionary novelty.
Our computers' evolution is inverted. The "easy", "boring" stuff we hoped robots would do for us, like collecting garbage, is the last thing they'll be able to do and perhaps the only thing left for us to do. :D
I think this possibility accounts for a good deal of the overblown hysteria about AI's potential impacts in the corporate media world. I wonder if any of these outfits (The Economist, the Atlantic, the Washington Post, the New York Times) would ever explicitly discuss this, e.g.
"Here's How AI Will Come for Your Executive Bonus and Dividends"
they're the ones ultimately deciding how AI gets used in a company.
As per The Dictators Handbook, I doubt a CEO with an inner circle replaced by AI could effectively direct a company.
The question is, we know AI is going to do take over doing work, how will that change the system when like 80% of people have no means of earning their income. How will those people abusing those machines earn theirs and who will generate input for those machines. Will those machines be able to solve genuinely new problems? Their knowledge comes from whqt people created. When people are no longer creating because they can't afford to live, who will feed the machines? And who will those few abusing those machines sell their machine generated products to?
It will lead to a revolution and it will be bloody. Already the 99% are discontent and it's brewing. With the artificial price increases for the cost of living more will fall into social security. AI replacing workers means the bloat in the IT sector will be reduced and a lot of unemployed people on social security will be left. With steadily less money being available for the 99% they will either turn on themselves or, more likely cut the head off the 1% boiling their water steadily but surely. There will always be people trying to defend their salve masters, if we learned anything from WW2 and even the invasions on Iraq and Afghanistan, it's people suck and want someone they can feel superior to.
Add to that the climate disaster...
Does anyone think we couldn't right now, with our current tech, even without AI, provide food, shelter, and healthcare to everybody with everyone only working four or even three day weeks?
But the worse thing is that it's never rational for economic actors (e.g., employers) to pay out a "social dividend" by cutting everyone's hours but paying them the same.
30 workers got replaced by 1 worker at a computer? Cool, but are you gonna pay that 1 worker 30 times the wages of a laid off worker? No; otherwise, why lay the 30 off to begin with? Conversely, are you gonna NOT lay off the 30, pay them the same, but have them work 1/30th of their regular 8 hours because their base productivity just got achievable in 1/30th of the time by computerization? Economically, that's absurd.
The only thing that can force those economic absurdities is the government. (Not having slavery is also economically absurd. Voluntarily refuse to have literal, self-replicating, AGI automata working for you? That's crazy! But it's enforced by the government and international norms, so the absurdity is now normal.) The scary thing about that is how hard it would mess with the economy, having the government tell you how many people you can fire and how much to pay them based on some measure of productivity or just business profits.
In a future of "AI, math PhDs, and serfs", though, we really need to figure it out. If we're headed toward that, it's going to be insanely asymmetric cyber-feudalism where 9.9/10 people are GrubHub peasants OR The Culture's post-scarcity luxury space communism. The latter seems way better. On the plus side, I think we'd only be GrubHub peasants for like 15 years, max, before that kind of AI ignites the Singularity, at which point who knows what happens? GrubHub would probably be gone though.
Only if you're very greedy. The truth is just the opposite, it's not absurd at all to pay well.
Lets imagine a hypothetical economy where there is only one company. "Pool party supplies". That is they sell things like beach balls and pool floats. So in your economic system it would make sense for the company to get rid of many employees as possible, and have the ones remain work 16 hours a day...
...and then they go out of business???...
People that are working 16 hours a day don't go to pool parties. The small economic upper class doesn't buy enough pool party supplies to justify the companies existence. The unemployed don't buy party supplies. Economic profit maximization lead to less profit in this case!
In this derived scenario having more employees that earn well and have plenty of time off leads to more pool party supply sales.
In contrast, in the early 19th century, when the western system was disrupted by labor unrest, it was 60+ hr work weeks were common and 40 hour weeks with weekends off was unthinkable. There were literal bloody fights to shift the system - and I hope we have progressed enough to not require that this time around, but it seems within the realm of possibilities in our modern day.
It will need to grow hands first. We’ve been “replacing” people with robots for a few decades now yet manual labour is still a thing. We have software that can literarily do all accounting tasks for us yet accountants are fine. I think will be an aid but not a replacement.
Most companies don't get rid of employees until economic pressure is put upon them. So companies will increase capability for some time into the future. Then we typically run into a trigger of some kind. For example a housing crash that causes a recession. Then mass unemployment occurs and we'll consider this 'normal'. Then we'll start to have an economic rebound, but white collar jobs just won't come back in mass (unlike eventually after 2008 which was very slow).
Then people will push hard into manual/blue collar labor, but the new influx of labor will drop pay significantly especially where low skill jobs are under pressure.
The economy can generate an unlimited number of jobs... The real question is can it generate 'paying' jobs that allow a person to survive the ever increasing cost of assets.
But I am wondering. Will we always want what we have today? Did we simply exhaust all need for travel once we developed cars? Or did we remove the need for food once we switched from hunting to farming?
My point being that new tech has always led to demand for new services and things.
Once everyone has access to ai and ai does everything for us would we not want new stuff?
And if as per your comment we have a couple of years until a possible replacement may have taken place wouldnt we have developed a demand for new services and jobs by then?
We had no clue we needed a tiktok app before we made the iphone and we had no clue we needed digital artists before we invented the computer.
For example the development of vehicles directly contributed to the scale of WWI. This then directly fed into the rapid expansion of flight and the development of city destroying nuclear weapons, which again lead to rocketry and civilization ending scale weaponry.
In that same time frame we went from an anti-fragile society to a very, very fragile one. In 1900 if the town next to you burnt down, at the end of the day it didn't affect you "that much". Today if the town that produces fertilizer for half the nation burns down, we'll there's a good chance you'll eventually have a bad day. You used to grow your food in your back yard. Now that food may be grown on the other side of the planet.
This is why can't just say "Well this worked in the past". When the underlying assumptions change you can have rapid destabilization. For example, world wide war because AI figures out how to disable nukes at launch time reliably undoing the last 70 years of MAD.
Indeed new tech led to new weapons and potentially more deadly conflicts.
But conflicts comparable to or worse than ww1 already occurred well before mechanised weaponry: in china, the mongol invasions, the colonisation of the americas, and so on.
MAD was mostly posturing, with few close encounters. In effect there was more anxiety than death.
Also i agree that there are far more single points of failure. Your example with fertilisers sounds plausible and logical. Both due to logistics and also due to losing farming skills. Yet somehow we suffer less from famine. Now theres more anxiety over it happening rather than it actually happening.
The pattern i see is that yes the risks are real and plausible but the odds of them occurring are slim.
And almost always (almost because i dont know of a specific counter example) technology only posed a risk due to politics rather than the tech itself.
There is no single example i am aware of technology posing an existential threat on its own - but at the same time we never created autonomous machines and software before at this level.
No matter what i do i keep getting back to the idea that the only danger is us. Our emotions, politics and instincts.
I feel conflicted.
We also need to, in effect, generate more tax revenue per productive working age person to support the aging population. AI helps that (from a government standpoint) by essentially "1:Many" multiplying the work 1 person can do, and be paid, and taxed.
Link to UN graphic on business insider: https://www.businessinsider.com/6-charts-explain-worlds-demo...
Then you have the middle ground of expensive enough to replace with a robot and AI, overlapping a skill range that an AI and robot can replace.
This middle ground is what we commonly call middle class and further gutting it presents many risks to an economy.
In regards to software engineering, automation and abstraction in our industry (compilers, high level languages etc.) has always increased the demand for software and people working in the industry. This is very unlikely to be any different unless we'll come up with ultimate general purpose AI that can solve any problem humanity can ever face.
Except that one of the major problems is that the people who built their lives and careers around those jobs are historically hung out to dry.
It is looking like software and most white collar work will be replaced before the trades. How many 45-yr senior software workers will be happy to give it all up to become a plumber? Don't forget they will be competing with all of the other displaced workers which will drive down wages. Will they be happy begging for minimum wage jobs unclogging sewers?
More points:
Reskilling will become increasingly impossible as it becomes faster and faster to train an AI model than a human on new types of work.
Net job creation is far from guaranteed. Most law work is competed by paralegals and other staff, but replacing them doesn't mean there will be a natural growth in legal cases. A lawyer might be able to process cases 10x as fast, but that doesn't mean there 10x as many lawsuits and criminal cases will magically appear - at least I certainly hope not.
> The only problem? Middle management is stressful, grinding, usually thankless work. People speak derisively about middle managers because their outputs are hard to define and monitor—they are viewed, sometimes unfairly, as a mere link in the chain.
Middle management is absolutely screwed though. Good ones will go back to some IC/lead work and bad ones will probably be squeezed out/kept around for political yes-manning.
- Beavis
If something can be replaced by a better option, it will be. Earlier king/people used to send messenger/pigeon to deliver messages. Later came letters(postal services), then phone, then emails, then SMS, then current gen of IM.
Any one doing/done some proper research ?
Americans are not putting bidet toilet seats into their homes despite how cheap and easy it is.
Just because something is better doesn't mean we will start using it.
Social proof related behavior are not changed easily, but they eventually. Despite availability of email, people were still writing letters. In longer time frame, people do adapt.
Impact seems unimaginable. May be we need an AI to AI all imaginable/unimaginable scenarios.
Then you're completely missing the point on road building... This job is already super highly automated. 2000 years ago it took a Roman division to build a mile of road. Now it's just a few people and a lot of equipment that is getting more and more automated every year.
Welcome to the 21st century, software developers.
Take Shopify which makes it so you don't have to hire devs for your ecommerce store. A lot of custom work for ecommerce stores dried up, but then developers moved on to things that could not be automated. Let's say AI gets so good it can handle ALL types of web development tasks. The tools for AI would have to be extremely advanced, so much so that a normal web developer now has super-powers.
You might not think a web developer has the ability to automate a construction workers job... but in a future where AI is so good that web development is obsolete? Robotics frameworks would be as easy to use as React by then.
Of course there are many edge cases in automating the real world, and that's why you'll need hoards of developers to iterate, debug and customize robots coded using AI. The job won't involve typing code in an IDE anymore, so maybe you can question whether it is actual "development." But the only jobs left in the future will technical people who understand how software and AI works.