Knowledge workers use it primarily to "augment" their work, answer questions with AI, get it to create documents, etc. But it's usually 1 individual to 1-2 conversation threads.
Meanwhile in software development were running fleets of agents. At work we have it setup where firing off one agent request will eventually lead to like 10-20 AI calls at different parts of the process.
I make this distinctions because it changes how each group thinks about AI. For software devs it just feels like "moving up the stack". Yeah you're not writing code anymore but there is still plenty of work to do and problems to solve within your "software factory". Contrast that with knowledge workers that feel like the AI is making all of their decisions for them.
I almost think of this as a factory worker versus an artisan. The software devs are "factory workers" where we used to be on the floor but now we're managing the machines that took over the work. There are still tons of interesting problems at that stage and you still feel like you have worth. Contrast that with a knowledge worker that's more of an artisan in this analogy. For them they aren't running a factory but instead watching a robot do the work they used to do while they sit idle and watch it.
I think your general thesis about how AI is used of agents vs. the general chat interface is true, but not the rest.
Your knowledge worker example is where people fail when using AI, and it is usually very obvious.
Knowledge workers who know what they're doing recognized very quickly that AI often goes down very bad rabbit holes, often cites incorrect sources, makes up sources, or will use a random Facebook source or Reddit post as an authority on some topic. In legal circles you have to fight it to understand what venue(s) apply and to get it to stick to the right ones, in medical research it will sometimes pull in crackpot papers from left field and intermix them with well defined research.
People in the knowledge realm caught on very quickly that you have to use AI iteratively and constantly back check what you get and correct it. Let it run on its own and it will give you a legal theory based on some weird case in Kentucky that has no bearing to your issue with a court in Vermont.
Using AI is like using adaptive cruise control: you're still in charge of where the vehicle will ultimately end up and if you decide to go take a nap you will wake up in the hospital or end up in the morgue.
It's true though that the 'time between interventions' is lengthening steadily.
The primary difference is that it's much easier for the AI to figure out itself if it's wrong on the programming side. Knowledge workers don't have automated testing for example.
The end result is the knowledge worker needs to be more heavily involved in the process and be more mindful of mistakes which makes it a worse experience for the user and generally less productive than it is for programmers.
For most professionals it’s just safer to avoid having output in your workflow. Keep using AI, but stop copying and pasting AI output.
In practice, lawyers generally don't ever disbarred except if they lie to clients or steal from them.
All of this will sort itself out and eventually bio, legal, software will all just work.
Don't hold onto any single assumption about the world. Things are going to be weird.
We aren’t talking about the new version of SV spiritualism here, we’re talking about systems people actually care about.
If lawyers and article writers are in the knowledge realm. Then your thesis has plenty of counter-examples of people who just use AI to create stuff and don't correct it.
And I'm probably on the more amateurish end of AI development. As I learned these techniques from YouTube.
When you hear about the work Anthropic researchers are doing, they seem to be using hundreds or thousands of agents to solve problems.
There's a reason they're Anthropic researchers. The value proposition tanks when you actually have to foot the bill for the tokens you use.
Subagent flows can improve overall performance, but it generally also comes with disproportionately high token use. That gets worse with parallel subagents that are often doing the same things at the same time and cannot share a cache.
So the median developer is probably using 5-20 agents at a time for a task. I'm not sure if that crosses the threshold of being a "fleet of agents".
I think that sums up the "fleets of agents" approach fairly well.
I don't actually use AI much at all simply because in my line of work the knowledge is industry specific and niche - an AI can't tell you how to fix a proprietary machine. I have used it a few times but to do dumb things like write some Excel script or some other piece of software I have no interest or need for learning.
It’s a phase.
Fortunately I'm in a shop that isn't being judged by feature output quantity, but is absolutely judged on whether or not our internal or external customers go out of service. We don't have pressure to pump output.
In the sense that "down the stack" is well-captured by existing language, AI is absolutely moving programmers up the stack. I can describe my goals using high-level abstractions, and the low-level implementation is somewhat constrained by that.
How well constrained it is depends on how well I described the problem and the quality of the model - much like how good human-written code is depends on how clearly the authors were thinking when they designed the structure.
I know knowledge workers who throw what you give them into the slop machine and return the slop, but I don't need them, I can feed the slop machine myself. I still need those who have experience and taste that the machine hasn't got yet.
I’m just old enough to remember some older adults in my life who refused to use email. They would only talk to coworkers in person or on the phone.
It did not work out very well, unless they could afford to retire at that point. When your career evolves, the only way to keep participating is to evolve with it.
I like to call it “keyboard cooties.”
I can see why someone who never used email until their fifties would never get fully comfortable with it.
(actually, i forgot i have 2 or 3 others, I check them even less, perhaps a few times a year, really only when i need to reset a password associated with them)
It's a fucking log dump for machines.
You see that 90% of the daily volume is sludge and think this is a healthy communication network for humans?
I can't afford a pay cut, and AI isn't making my job worth more.
Why at lower wages? Most automation has led to people transitioning to higher wages. Utility works get paid 10x more than lamplighters. Why would that change here? Your job will have a bigger impact, affect more people, and you'll be doing more with less.
I am not saying all employers will immediately start paying people more, but economically it doesn't make sense that pay will go down.
The economy now is very different, no gold or silver standard, less isolation and more globalised product chains and processes, and 7-times as many people.
There is a good chance that a large enough number of people will suffer from this transformation, and they will have to fight with whatever they have, because if they don't, they will risk being left on the roadside and not getting any raise.
People having some direction over how jobs are automated should keep that in mind, otherwise it will be too late to convince the masses of people fearing automation that it is their fault not seeing the positive angles of automation.
But, the story doesn't end there does it? Most loom workers left their jobs and became factory workers (part of the same technology revolution) and critically: made MORE money. They were wealthier. And the loom machine operators left had a better job. There are very good sources for this, not just vibes [1].
I don't deny that the transition process is traumatic, and there is much we should be doing this time around to assuage the process. But all the evidence points to a general rise in pay, material conditions and standard of living.
[1]: https://yalebooks.yale.edu/book/9780300205251/libertys-dawn/
Individuals and entire communities can live out the rest of their lives destitute and hopeless in the wake of changes that leave them behind. Things might be brighter for those who come after, but let's not pretend we're not doing real violence to those left behind.
There's a lot of questions right now.
1. Will there be a net-new creation of jobs? Is positive-sum knowledge work a real thing, or have we run out of new jobs to create?
2. How quickly will workers transition to an AI version of their jobs?
3. Will AI even succeed in a broad way across knowledge work or will the frontier of AI remain jagged and only impact some jobs (coding) while leaving others alone?
4. For workers left behind in this transition, what will their lives be like? We know from looking at previous technology transitions (industrial revolution, factory automation, computer revolution, now maybe AI revolution) that many workers really do end up in bad places. It's perfectly reasonable to demand that in today's world in our advanced societies that even if those workers are unable to adapt that they are able to lead lives with dignity.
There's a lot of interesting conversations and investigations to be had but my biggest fear is just that nobody is really willing to listen to the other side with empathy. This entire conversation thread is just people snipping at each other, hitting the vote buttons based on where they want to steer their favorite opinion. If you want to do social and economic science in this space but people are so polarized, the work will be thwarted. In general offline spaces are less polarized (not to be mistaken for pro or anti AI disruption) than online spaces, but my fear is that we continue to polarize, making understanding the fabric of our social change challenging and letting down everyone in the process.
Try to come up with a job that both doesn't produce anything physical, and can't be done by AI.
This is vague marketing speak and simply not the reality on the ground. As someone implementing all new workflows, most of the time I’m create new product classes wholesale. sometimes I see people upskilling tasks so the same human can do more.
I can’t think of a single knowledge job where they are removing humans from the loop completely.
Random news headlines of struggling companies doing “AI layoffs” don’t count. No amount of AI can save bad leadership.
people who spout the trite “everything will be better at some nondescript future point in time” are essentially screaming “the ends justify the means and screw everyone else, i’m going to get my bag / be ok”.
No, what you mean to say is that automation has eventually (decades later) led to higher purchasing power in the system as a whole. The outlook for people that were directly automated away was pretty bleak, actually. The hundreds of thousands of lamplighters did not largely become the thousands of electricians that were needed in this new industry.
This tracks for all industries. Automation literally means far fewer people will be employed to do the same job (otherwise what's the point of it?). That dumps these workers out at the low end of the general job market, where they have to compete with an increasing number of people. Because of the desperation to get any job, the wages fall. It's only decades later that societies settled into a new place and created enough brand new jobs to make up the shortfall, or applied regulation to reign in the crazy power imbalance. That's the stuff you miss out on if you only look at statistics that span centuries and treat the population as one homogeneous mass. The people who are directly targeted by automation are very likely to have their lives messed up for a very long time.
No technology involved? Other than I guess tangentially, containerized shipping. I don't see how that analogy applies here at all, every AI workflow I've seen has been either:
a) Totally new thing that was never done before that is possible now and is therefore creating a new class of work
b) Worst case scenario: a whole spate of jobs being consolidated under one person, but its still a human in the end.
So everything I'm seeing is pointing to this being a lamplighter/utility worker situation.
Gonna need a citation for this.
A grub hub delivery man is wealthier today than a farmer 100 years ago and that's purely because of two reasons: Technology and Unions :)
This ball is going through even your ever-moving goal posts. My statement never claimed people's careers would stay the same and they'd just make more money.
There are some great reads on this topic[1].
[1]: https://yalebooks.yale.edu/book/9780300205251/libertys-dawn/
Currently, I don't see any reason at all to suspect that AI will make my life better or even just not make it worse.
AI boosting seems to be very close to pulling shit out of ass.
It was never about software. That was just the means.
Even in the software era, fads developed and tech stacks changed frequently. Now it's just moving a hundred times faster.
You need smart systems thinkers to do this. That's the role. Be smart in managing processes and complexity.
I'm not going to give coding interviews anymore, but I am going to give architecture, systems design, and SRE interviews. People still need to know how not to fuck up a database and how to design scalable and resilient systems.
The number of people who never seem to have understood what software and IT is in most companies is eye opening to me. Your job was never really to "write code" - you were always automating things.
That the field got somewhat stuck for a couple decades seems to have made many people forget this.
The vast majority of programmers who started in the 80's and 90's are basically directly responsible for massive waves of job losses in certain categories. Lots of cases of entire departments of secretarial level paper-pushing work being replaced with a small team of programmers. Same will happen with AI tooling. The difference is programmers are no longer the "winning" party here.
Computers and software in the end are the most configurable and complex automation system society has ever created. They are simply a continuation of industrial automation, and AI is a continuation upon that.
So ... perhaps more secure than email, but not necessarily as much as you'd think. Though maybe it isn't as common now to have phone lines with fax receiving included. I worked for a CLEC at the time and got my phone & internet through them too, so maybe I was just getting the special deluxe plan. Had my very own public class C, too, but that's a different problem.
Lots of people, myself included, understand llms and can and do use them at work because we have to, but we might choose to not use them outside of work because they rot our brains, are unethical, funnel even more money into the hands of a few billionaires (selling a service trained on stolen property) and haven't actually been proven to make the world better. Unlike email!
What the fuck have LLMs done yet? Nothing! We would be no worse off without them so far and no one has anything to show to prove otherwise, except some vibe coded websites that all look the same (edgy dark theme and hacker vibes)
Traditional professions share essential characteristics. They provide a vital service to society, requiring expertise and skill developed through years of training, education, and experience. Professions establish standards of practice and certify that their members are qualified to serve the needs of society.
--US Army, ADP 6-22, Army Leadership and the Profession
Software does not have explicit standards of practice or a required body of certifications for its practitioners. Software, as an employment, is whatever an employer wants or is willing to accept at any given moment.
Hillel Wayne wrote a really excellent well researched article about software engineering compared to other forms of engineering a few years ago. It necessarily goes into a lot of this stuff but what emerges is that these definitions are regulatory and historical debris.
You don't need medical board or a bar to certify or constrain professional behavior and those organizations may not constrain as much as you assume anyway. The reasons we do have those institutions is as much a matter of historical accident as any modern need for it, and the reasons we don't have it for other high impact fields is pretty much an accident too.
And in a lot of the world software engineering does have professional certification and standards of practice. They still ship adware and racist loan denial algorithms too.
Every part of your existing job is transforming before your eyes.
If you cling to definitions and static roles, you're going to have a bad time.
Systems design, architecture, database fundamentals, a bit of SRE. That matters. Not coding and syntax and toy problem solving with data structures and algorithms.
My own experience is that coding, syntax, and some kinds of toy problem solving are important skills to use AI coding tools most effectively. I routinely get PRs these days whose descriptions seem plausible but are clearly wrong from a glance at the code, and I've had times when I'm unable to communicate to someone why a PR is wrong because they're not able to form a sufficiently detailed mental model of what it's doing.
I'm not going to let the army tell me what my job is.
And lets be pirates and hackers instead.
https://www.computer.org/education/bodies-of-knowledge/softw...
But by that definition, soldiering can never be a "profession" as a soldier is always regulated by the government they serve, and there is no professional association regulating mercenaries. But I would be very hesitant to suggest that the US Marines are not professional soldiers. They are. The definitions are messy.
As they should. Industries should be self regulating. While these industries form their own standards of practice they do so with the blessing of government, typically under state government supervision/authorization in the US, and are held to account by government. A lawyer or doctor who violates their standards of practice could very likely be guilty of criminal law and find themselves in prison depending upon the violation even when no other criminal actions occurred.
Even bus drivers are held to higher standards of practice than software developers.
Because while there are some software developers that might be in a place to cause people physical harm, ALL bus drivers are ALWAYS in a position to cause great and immediate harm to others on the road.
It's the legal liability that SWEs don't want. e.g. You wouldn't be able to just slap "This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE." on your software and walk away from it when lives are at stake. A so-called Profession Software Engineer would have to accept liability for their work product, and those of the juniors working under them.
I started down the road to becoming a Professional Engineer while I was in college as I studied Electrical Engineering and was more interested in hardware/power engineering. I took, and passed, the Engineering Intern Test in my state (which is the preliminary exam you take before you can start your journey to become a PE). Then life happened and job/career opportunities took me in a different direction/industry, and I've been doing software instead. So, I never became a licensed Professional Engineer. I'm still a licensed engineering intern, though. I'm still bound by the legal and ethical standards that it entails and could lose my license for violations.
There are certainly things where you need to have an officially licensed engineer actually stamp the plans. That doesn't mean everything about the plans was done by someone who holds that stamp, just that they were reviewed and signed off by the person holding the license. Lots of people may work on those plans. They're still doing engineering work. And that's just for the kinds of things which actually require a licensed "engineer".
There's still tons of jobs which is often still called "engineering" which often don't involve anyone who has a stamp. Few electrical engineers, chemical engineers, mechanical engineers, or lots of others actually have any kind of engineering license. They still get degrees in things like "Bachelor of Science of Electrical Engineering" or "Master of Science in Mechanical Engineering". Their job roles usually have "engineer" in it. And yet it seems like the job people only really criticize as "you're not an engineer, engineers are licensed" is SWE.
I always come back with "you're not an engineer, engineers wear stripey hats"
I guess search engines count for something if you shovel in enough coal ;)
Many states require you to pass a test and get a license/certification before working in fast food.
That is not a requirement in the SWE world.
The sheer lack of accreditation and standards-enforcing body is what makes sofware engineering a non-profession.
Yes, several: the ACS, ACM, IEEE.
So you’re wrong.
What was your last employer that asked for your license number of member ID from any of those organizations? I am guessing it has never happened.
Normally when people try this argument they are hoping a for a dog whistle to shift ground because they don't want to be regulated.
People refused phone calls, refused to watch TV, refused to buy computers, refused to move their workflows to software, refused emails or browing the web, refused online shopping, refused smartphones, refused social media.
And there's nothing wrong with any of this. Ultimately everyone decides which stage of technology they are comfortable with. Everything until then is comfortable and convenient, and everything newer is corrupting society and rotting our brains.
Saying that this revolution is another one of those is true in a general sense, but misleading in the specific sense. Just as saying that adding 1c heat to some body of water is "more of the same" can be true in a general sense, but misleading when the water's basis temperature is 99.5c.
That is to say, periodic or gradual change induces large changes at inflection points.
It feels similar to the progression from post-WW2 technological utopianism that arose with the adoption and deployment of massive advancements in materials and mechanical, and electrical technology, to the cold-war dread of nuclear and ballistic technology.
I think the 90s saw the end of the cold war which lifted a bunch of that dread. The optimism of those days has been worn down over time.
What I remember back then is at least the pretense being sold that the technology would improve people's lives by giving them more leisure and enjoyment. That the productivity gains would be reflected in some improvement in quality of life. I think the main difference now is that this expectation has been lost by most people.
[0] https://futurism.com/artificial-intelligence/gen-z-turning-a...
You really think this is an apt comparison? I don't think the creators of any of those technologies were warning society that their own creations might wipe out humanity.
They are only doing this now IMO so that they can be the ones writing the regulations to keep us all "safe", just like we need identity verification to "save the children". They want to keep competition at bay so they can have a $2T IPO, and unregulated, "unsafe", Chinese, open-weight, and local AI threatens that IPO.
I disagree with "sizable". We haven't seen near this level of opposition in the past (I'm old enough to know), mostly because moving from say landlines to mobile phones, or postal mail to email, or radio to TV, didn't threaten the jobs of large segments of the population. Past opposition tended to be more rooted in culture or even religion (TV as a vehicle for "ungodly" programming, etc.)
I run a wide-screen 1080i Sony Display as my main TV. Power consumption sucks, though.
And now you want to automate what little most workers do?
I find your Amish example is striking and accurate.
As technical qualms were solved and many other capability issues with AI have been solved, the continual heel digging and goal posting by group 2 has started to just feel more akin to a religious stance than something based on any specific fact. There is no version of AI that they want to see, they will oppose it on moral(-ish) grounds on sight, etc.
AI has real issues, but like you, observing this re-adjustment in realtime has been really interesting, especially as it seems to come from people I share 99% overlap in interest and politics with, but who hate AI like they do.
It also feels akin to de-radicalizing MAGA or something. The rejection or idea that they are wrong causes a whole worldview to collapse in something akin to ego death, so instead of facing that possibility they just insist they are right.
They've stopped technological progress. I tell you since AI started taking over, my experience is that nearly all forward progress (in software at least) has come to a halt. We've been standing in place, maybe sliding backwards a bit.
Here's the thing about AI: intelligence isn't a new thing, and it isn't a tool. The Amish can tell you that having a brain doesn't meant that you don't need a plow. In some ways they're much smarter than the AI crowd, all of who loudly keep insisting that brains will make plows and tractors and combine harvesters obsolete. You're so convinced that intelligence is the ultimate replacement for tools that you've stopped climbing the tech tree of tools...
It's like the tech industry itself no longer believes in technology.
I'm not even talking about holographic game consoles or FTL travel. Why can't we finally have a good container orchestrator that's not kubernetes for example?
Easy, we can't because the LLMs know a lot about kubernetes but wouldn't know anything about a new orchestrator, so they'll always suggest k8s, and the button-pushers will just say "yes, do it, no mistakes". Maybe if Anthropic decides they need to create a new orchestrator, they would bias Claude heavily enough for it to actually get used, but that's about the only way I see.
"Why don't we have any new programming languages" or "Why don't we have any new web frameworks" or "Why don't we have any new operating systems" or "Why don't we have any new container orchestrators" might be the same kinds of questions as "why is the design for washing machines still basically the same as it was 50 years ago?" or "why are we still using English, where are all the new conlangs to replace it?" "why are we still using email/SMTP instead of inventing a better system" "why haven't we invented a new calendar?"
All it will lead to is yet more people that are out of a job. Right now at least a driving license (more so if it is a commercial license) is a source of income to many.
What's the market demand? I'm confident there are plenty of alternatives being churned out today but nobody's looking or asking for any. If only because they settled on k8s, they have it on their CV, they have certifications, and switching is a huge investment and risk.
Right now to me it feels like the innovation is in using LLMs to (finally?) clean up existing code, find vulnerabilities, do maintenance, optimize, and if necessary rewrite.
Where before humans learned and built better tools, now you keep paying forever to get re-trained models. What makes it a regressive feedback loop is that the people who used to learn and push the frontier are now using AI to repackage others' work as their own. Eventually what will be left to train on but copies of copies of copies?
Also, you might think that you are observing Amish (or AImish) from a distance. It is same for them. They watch the modern world rot with uncontrolled and out-of-sync evolution. You are being observed.
There are over 2 Billion people without Internet still and they survive.
https://www.weforum.org/stories/technological-innovation/2-5...
At my age of several decades, I lived several decades before the Internet even caught on and managed very well as did most people.
Questioning whether people might potentially not fare well by choosing to not use AI is what is truly puzzling to me about your comment.
I read it as: faring well in a work/industry/employment situation where expectations for velocity and quality pivot on the fact that 99% of your peers use AI, and 90% of those use it to do poor quality work quickly.
Now your boss is surprised your email took more than 10s to produce, and wonders why you are concerned about what it contains. None of your teammates are.
Small thing: many/most of the people outsiders identify as Amish actually aren't. There are a variety of group living similar lifestyles, give or take. Collectively they are now described as the "Plain People", which links to both their apparently simple ways and mode of dress.
I say that as someone that uses it everyday for technical things only, for research it is great. i literally have 30 sessions in parallel going with an LLM. that said, I struggle to see the need for AI in anything else. I build it into services that I build but that scope is nowhere what meta would have you believe. It doesn't need to be making decisions for you. You don't need to offload decisions to it. It is upon you to validate and research the findings and the finding generation process.
Please read this:
https://archive.nytimes.com/www.nytimes.com/books/97/05/18/r...
What makes the comparison wild? The Amish are thoughtful and deliberate about what technologies they reject, adopt, and accept with constraints. Many of the people I know who have concerns about relying on LLMs are similarly thoughtful and deliberate.
"If our world survives, the next great challenge to watch out for will come - you heard it here first - when the curves of research and development in artificial intelligence, molecular biology and robotics all converge. Oboy. It will be amazing and unpredictable, and even the biggest of brass, let us devoutly hope, are going to be caught flat-footed. It is certainly something for all good Luddites to look forward to if, God willing, we should live so long."
I just looked it up and Thomas Pynchon remains alive.... how wonderful.
Thank you for this
But in this case Amish is not really an insult of course. I guess they are a hard-working people. Self-sufficient. And allow their young to take a sabbatical from the primitive life to see if it’s really something that they want to dedicate their life to.
On the other hand, of course there won’t be some Amish split. You can’t just pick and choose modernity. You can’t form a subculture of No AI. That’s why people are critical of AI; we live in a society.
[1] Being ahistorical and using Luddite as an insult also helps.
Ai is stripping people away from their identity. There’s tons and tons of hatred for it… even here on HN.
societal ostracization (e.g. boycotts)
sabotaging data centers, data gathering infra (flock cameras), web infrastructure
These are just methods within the current societal norms and context, if not I could think of 10 more.
For example, it is not beyond the realm of imagination to think that large swathes of people could stop using technology (as in phones & computers) altogether, and return to a more traditional physical communal state.
It is not beyond the realm of imagination that countries, after suffering from cyber attacks, could start decoupling themselves from the global internet, and wall themselves off (similar to China), and develop their own unique culture around technology.
It is not beyond the realm of imagination that tags like Anti-AI be mandated to be attached to products similarly to kosher, halal, etc.
But if someone was say, working on housing approvals and refused to use AI because of religious exemptions to do their jobs faster, I would hope any court would strike it down.
It's easy to find egregious corner cases of philosophical workflows where an LLM was inappropriately used to replace thinking, where the whole goal was thinking (which is the case here).
But we know most token use is software development + back office paperwork. A big chunk of it is productivity improvements for things we all need (housing, medical paperwork, insurance etc.), so to me, that's where my argument lies.
The bottleneck for housing is construction and landlords. The bottleneck for medicine is the number of doctors and the inelasticity of demand. The bottleneck for education is the number of teachers. Etc.