The problem is I don’t know enough about other people’s jobs to answer that. But I think most people have parts of their job that a machine can’t really do. You would think grant writing would be one of the first jobs to go, and it might take a hit, but so far it seems to be helping me more than hurting my job prospects. I can’t remember if it was this thread or another one where we talked about the importance of legacy information. AI can write a grant narrative pretty well, and it can organize an application or related process, but it doesn’t understand the context of our specific community and it’s hard for me to imagine an AI that would understand it. In order to understand it you have to go out and talk to people. You’d have to understand the rhythm of the agency and our culture. You’d have to meet with program staff, and attend a shit ton of community and funder events. You’d have to be human.
I suspect a lot of jobs are like that. But I only have my own to go on. I can see the highest risks being the tech sector and customer service jobs. Which does suck because there are a lot of people in customer service. I do think knowledge workers are relatively safe.
Now that doesn’t mean CEOs aren’t going to be suckered into firing a bunch of people and then finding out the hard way AI can’t do their jobs. I expect some of that kind of turmoil.
The lack of a sustainable business strategy is one reason I’m not particularly worried about being replaced with AI. I expect this is going to end with a lot of people losing money, which will suck perhaps just as badly.
Yeah, killing that pipeline is one of the seemingly more short-sighted parts of all this. And yet…
…this is true too. Generations of programmers have worked with higher and higher levels of abstraction. I never had to learn assembly, or even basic algorithms + data structures, or even C++ or C# or Java (which are considered low-level today!).
By the time I really got established in my career, the dominant market force (Next.js) was already an abstracted framework built on another abstracted framework (React) built on a third-party extension (TypeScript) of an interpreted language (Javascript) that gets just-in-time compiled by a web browser script engine to modify HTML, all running on top of several more OS frameworks… all to render some stupid marketing picture and paragraphs of nonsense 0.13 seconds faster. And yet I was paid more than the electrical engineers writing low-level life-or-death firmware controlling dangerous battery electronics…
A lot of the tech world is just plain wasteful, and more fashion-driven than scientific. The buzzwords du jour always pay better based on regular ol’ supply and demand, not so much their underlying merit.
I guess AI just extends that. I can’t speak for all programmers, but to me, the work of a “prompt jockey” isn’t fundamentally different from what I’ve always done in my career: use words to tell a computer what to do.
It’s just that the most popular programming language has transitioned from Javascript to English.
Yeah, exactly. The classically-trained programmers still have a (significant) leg-up over the layperson, for now, because we better understand the AI methodologies and failure modes and can correct for them. But that gap is narrowing quickly as the models and harnesses get better, and will probably be completely gone in a few more years, if even that.
In a nice world: AI expands productivity for everyone, like the computer did, and there ends up being a net increase in needed labor and the commensurate pay.
In our world? Probably not so much. Nobody is investing in AI because of its benefits to the average office worker drone or mid-career programmer. Its value is in being able to replace expensive human labor with cheap, good enough machine labor.
So, yeah, my boss, or more likely his boss’s boss’s banker’s boss. In a nicer world, we’d see millions of small businesses pop up. In our world, more likely we’d end up with even bigger multinationals or multi-planetary businesses with a few uber-capitalists and AIs owning most of the production and wealth.
I don’t think we even really have the societal structure or words to describe this sort of disparity anymore. “Trillionaire” doesn’t really capture it quite completely, when a single person can be wealthier and more powerful than most of the world’s countries combined, and that’s with AI in its infancy.
It’s funny, when I was a kid, Syndicate Wars seemed so, well, fictional. Now it reads more like an instruction manual for the future… sigh.
I think that’s one of the more interesting side effects of AI. It devalues not only labor in the general sense, but also the value of specialization that our particular knowledge economy so depends on.
More than an accounting of mere hours saved, its ability to be a master (or at least journeyman+) in most trades and scientific disciplines all at the same time puts it in a league far beyond any single individual human or company. Yes, at any given task, there is today — and perhaps always will be — an individual with more expertise. But no human individual can draw from the same breadth AND depth of an AI… there’s just not enough time in a human lifespan, even devoted to 24/7 study.
Specialization used to be scarce and valuable because it took real effort and time. We’ve removed time from that equation and shifted the bottleneck from human expertise to silicon manufacturing capability. That transcends borders and cultures and populations and risks diverting the global economy even further towards the hands of a very, very few people who can produce “good enough” AI suitable at most tasks, even if it’s not the leading expert at any of them.
Huh. Despite my overall worry about how AI is going to affect us in the longer term, in the short term… I have to admit I really enjoy using it.
I never really liked programming languages. They were just so limiting and stunted compared to natural languages or illustrations, lacking the expressiveness of even a 5-year-old or a housecat.
If it were up to me, really, Frontpage or Dreamweaver would still be the way everyone made websites We’re not very far from an AI version of that becoming reality — there are many in-the-works prototypes of that exact thing.
I would love a future where I could just doodle on a napkin, chat about what it all means, and have it become reality in a few hours. Everyone gets their own personal Holodeck, except it’s all real too.
Not just web dev, either. Claude’s been the best teacher I’ve ever had, and helped me explore topics and fields I could never afford to learn about before. If only it could also act in the real world, outside the knowledge economy.
What about when it does expand into the physical world with AI-powered robotics (or human slaves, sigh)? “Claude, will you build a log cabin with me here?” would be great, but what would the costs of that be?
Like the Internet, AI has so much potential. Like the Internet, I’m sure that potential would be squandered and co-opted by the worst among us. It’s just how our species has always worked. I hope I’m wrong, of course, and that some breakthrough open, democratic AI ends up fomenting a proper proletariat revolution and leads mankind and machine into some new golden age… but yeah, right. More likely some other crazy Kennedy or Muskspawn ends up Galactic Emperor instead.
But, oh well, guess we better enjoy that tiny bubble of happiness before the enshittification really sets in.
My husband is a back end web developer. And his job, also, looks like it could be easily replaced by AI. But he says, when people ask him about that, his answer is:
If our customers knew exactly what they wanted, they could code it themselves with AI. So we aren’t worried about our job security.
I think that’s true in a lot of fields. You need a certain level of expertise to know what to ask for. And also, what he says isn’t even true. He tells me every week about some boneheaded mistake someone is making that’s way too technical for their customers to understand.
Written shortly before the internet / www became generally available, it’s a gallows humorous look at how a massively connected human and machine world could develop with all of human’s worst instincts turned up to 11.
The author did not correctly predict the way the www would develop; he figured something we’d call total immersion VR would be the standard way people used his internet-equivalent infrastructure.
But still and all, it’s eerily plausible in concept if not in detail.
It is a great novel, but NB it is not really about A.I. in any way. (Even though the protagonist does incidentally use an AI, as people do today, but which would have seemed futuristic when the book came out. Also, interestingly for a sci-fi novel, it is set in a specific year (IIRC 2002 or 2003) rather than generically in the future.)
There are a surprisingly large number of wealthy people in the United States.
But also look at the progression from assembler to C to client server and web development to no-code platforms. In created more abstracted, less technical interfaces, these developments have made analytical tools accessible to a greater number of people who can become productive without requiring masters degrees in computer science.
But more to your point about aviation, I see a similar trend in corporate management and leadership where people are questioning the value of middle management and administrative roles like project managers. “Oh that’s just all administrative bureaucracy.” Well guess what? 90% of a large business IS administration and bureaucracy. And I’ve come to the conclusion that the reason a large corporation can function is that it maintains a large enough middle management headcount such that within all that bullshit and paperwork is enough institutional knowledge and actually run the company. Which is why I always find it surprising for a job like mine where I might spend months and months learning the minutiae about a corporation’s unique collection of processes so they can actually build something, that knowledge becomes so disposable.
My point was, there are a large number of wealthy people because the middle class is employed with income to spend, which flows to the investor class. Once AI has hollowed out the middle class, that cash flow stops. How long do the wealthy stay that way, without a prosperous middle class?
Good point. If the US middle class becomes more like the e.g. Mexican middle class, what will happen is a large contraction in the number of wealthy Americans. but not as much a collapse of the wealth of the uppermost. Mexico has homegrown USD billionaires too. Just not as many per capita as the USA has. Same for folks with net worths of e.g. USD 100M and USD 10M. Per capita USA has more such people than Mexico, but there are still a goodly number in Mexico.
As long as each of those now-wealthy Americans can convince themselves they’ll be one of the survivors, the game can continue as-is.
I’ll also point out that there’s a big difference between the “Investor class” who own shares, and the “Tycoon or CEO class” who actually have their hands on the controls.
I’m investor class. As are most retirees who aren’t living solely on SS & a fixed pension. Unlike most, I’m fairly far up the food chain percentile-wise. Which is nice enough, but which still means I’m practically broke compared to the Tycoon / CEO class. I have exactly zero influence on what they do or don’t do. And for damned sure they don’t care about me and folks like me.
I don’t consider that everyone who owns stock is a member of the investor class. A person with a moderate IRA or 401k is not in the investor class. The 10% of investors that own 87% of the equity and bond markets are the investor class. I don’t see how they prosper when everyone else has to liquidate their 401k to survive because they have no income. That will be the beginning of the end as stock values crash. Even the ultra-wealthy will be decimated, and the surviving oligarchs will be dragged into the streets by starving mobs.
Hasn’t most of our history been similarly bimodal, with an elite class and everyone else, with a tiny bit of trickle down in between?
It’s just that recently, the wealthy became much more so. Maybe there was a brief century or so where we thought a sustained middle class democracy might be possible, but that was probably the exception rather than the rule? AI just brings us back to the “natural” order of our species, maybe, where most of us are fodder for the few kings.
I mean, most of us just want to eat, breed, play for a few hours, and then sleep. It takes an outlier to have both the ambition and talent for world conquest. Capital and now AI have always amplified their power, and the rest of us simply don’t have the will or means to do anything about it, so we beg and talk about things like “ethics”, but those are altogether irrelevant to the actual power dynamics. .
Revolutions will come, but they will similarly end. I would say that billionaires and slaves are the steady state for our species, not equality.
At a company I know of, they laid off half their technical writers.
They didn’t replace 100% of the the technical writers, just half. What they did was the AI writes the backbone of the document, then the human edits and updates it. So they only needed half the labor to do that.
Thats likely how AI will affect the economy in the medium term. It won’t eliminate 100% of jobs, but in certain industries they will find that they can eliminate 30-80% of the workforce and the combo of human workers and AI can have the same outputs. Thats what happened in the industrial revolution. For things like shoemaking, weaving, etc you didn’t get rid of all the workers, but you got rid of maybe 2/3 of them.
However in the long run, decades from now, there will be virtually nothing a human can do better than AI and robots. So human minds and human muscle will become like horses, totally replaced in the economy by non-biological machines due to inefficiency. The only jobs left for humans will be jobs where regulation prevents automation, or jobs that require meaningful social connection (babysitter, caregiver, etc)
A big one is pathologists and radiologists. AI is getting better and is already much faster at looking at scans and slides and coming up with a diagnosis.
We’re not very far from an AI version of that becoming reality — there are many in-the-works prototypes of that exact thing.
.