6 ms·

To me looks like, if we're not collectively careful, civilisation will soon be on a path to an evolutionary dead-end. Anything that can replace a deeply experi

To me looks like, if we're not collectively careful, civilisation will soon be on a path to an evolutionary dead-end. Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack, meaning that only the owners of capital will be left, and they too will soon fade as the economy falls off a cliff and money has no value, because the only value that money has is the value of a human backing that, with thought, with ideas, with human output. Whether you like it or not, "Economic output" is just a different phrase for "Human output that is valuable". When all human output is valued at the fractions of a penny per month of work, there is no future.

p-e-w 5mo ago
> Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Nope, just knowledge workers. We’re decades away from automating many manual labor professions, even “unskilled” ones. Turns out brains just aren’t as special as we thought.
- lanfeust6 5mo ago
  The major blocker for manual labor automation in that fashion is cheap energy. China is ahead of the pack with the States' weight behind aggressive expansion of solar tech, and still can't do that.
- theshackleford 5mo ago
  > Nope, just knowledge workers. Nope, just a specific kind. Those who developed and cultivated only a very specific skill set at the expense of all others. I used to think being a generalist, and having persued technical roles with a people facing element was to my detriment, but it’s turned out to be the best decision I ever made.
  - tempest_ 5mo ago
    I had the opposite thought. Being a generalist was very useful to me 5 years ago. Now AI models have made everyone a generalist. That wide but not terribly deep skillset was immediately devalued by the AI models. You can argue that the models fuck up 20 percent of the time, or that they make poor code but there is a massive part for the industry that is totally fine with that and I think people ignore it to their detriment.
    - theshackleford 5mo ago
      For context: I came from hardware, Linux, networking, telecoms, and datacenter infrastructure, not software development. I always wanted to go deep, but in practice my brain dragged me across many instead, which unintentionally made me very broad. I kept ending up in organisations where I was pushed back into such roles because its apparently where my "value" is. I give that context because unlike a lot of you, I’m not a world class FAANG engineer and never will be. It is from this context all of my thoughts on AI flow. I work with people who are trying to use AI to produce work involving entire markets, roles, skillsets and technologies they don't even know exist, let alone understand. > I had the opposite thought. Until I recently got pulled back deep into engineering despite not being hands on for close to a decade, so did I. I was pulled in not because of any pure technical capability but instead because it's been recognised the team requires more. The skills I thought served only to help me stay employed in any role in the most basic roles are increasingly turning out to be things other's do not have and are becoming increasingly important. These are skills I always assumed crucial to “baseline competency” for everyone, but yet where a significant amount of them do not, and these individuals are now finding themselves in positions where they are less useful than me as a result. Many of them can not simply be acquired from AI either, and require years of active growth and practice. > Being a generalist was very useful to me 5 years ago. Now AI models have made everyone a generalist. I think they could, but have not. Not at a scale required for me to have significant concern. AI works as well as the context you can provide, and you don't know what you don’t know. If the context is shallow, so to will be the output, even when it looks convincing and that “looks convincing” part I believe is the most dangerous part. As an example; I've been (recently) attached to an engineering team, despite last holding that title pre-2015, after AI assisted work contributed to a multi million dollar contract loss. A customer experienced an outage, it was "fixed" and everyone moved on. A month later another outage occurred of a greater scale. A huge amount of time was wasted doubling down on the original AI finding, because the actual root cause had not been identified or understood, because it had been "fixed". Turns out AI had identified and "fixed" a symptom, not a root cause. I was able to identify and resolve the real issue because I had wider operational and infrastructure context the team lacked, but the damage was done. Trust was gone, the client lost, and layoffs will follow. Those layoffs will be “because of AI,” but not any "10x'ing" of productivity. Instead it will be because plausible but wrong work made it into production and hid a very real problem as a result. That’s the issue with AI as I see it now. It generates answers that survive initial scrutiny while completely missing wider context leading to cases where more impactful but hidden problems are introduced. > That wide but not terribly deep skillset was immediately devalued by the AI models. Perhaps “generalist” was the wrong word here. Most "engineers" I have worked with are extremely deep in their area and surprisingly limited outside it. Even with AI, they struggle to move beyond their specialty because they lack broader foundations underneath not just modern infrastructure, but a range of areas equally important to the health of a business. My advantage has never been being the best engineer in the room, I knew early in my career I’d never compete with the engineer who can patch our kernel before upstream does, despite wishing I could. What ended up mattering instead was becoming the "95% guy" across infrastructure, networking, systems, operations, business, customer success, and people management that allows me to work with people/organisations and ultimately connect dots in a way even the best engineers I have worked with can not. AI can help you develop skills in areas you don't have, but starting with most of it in areas in which people have exactly none, and where people seem extremely resistant to developing it with or without AI, has me significantly further ahead in the curve. Ironically, at least in my experience so far, AI has made that more valuable, not less. > they make poor code I consider this to be the least important part. We have testing, review, and process for that. I believe (and have instructed juniors as such) that the real value of valuable technical people has never been producing rockstar code, or being a clone of Linus. It's in having a deep foundational understanding of the building blocks underpinning the now endless layers of abstraction, understanding consequences, tradeoffs, failure patterns, business impact, customer communication, customer wants and needs, and ultimately I guess to sum it up, organisational reality. This feels more important than ever when they can generate plausible looking technical output instantly that they may be able to validate, but equally produce plausible output in a huge range of areas they absolutely can not, but for which their successes in code have led them to believe they can. Because they underestimate what they don't know and in fact often assume they know far more than they do with no real basis for such a belief. On the whole, I think I would end my thoughts like this. For years I lived with the stress that "rockstar" engineers would lead to me eventually becoming irrelevant, much in the same way I might fear AI. So far, being 95% across customers, leadership, sales, support, engineering, and business strategy without losing the technical depth underneath it has meant this fear was unfounded and in fact put me ahead of them. I believe I am not isolated in this, and that in fact we will see more of it. All else aside, my roles as of the last decade often require me to be in the room and working with humans. AI has not changed this, and there is no current indication it will. The requirement will be that I continue to remain in the room, only now with AI. This is for many reasons including regulatory, because portions of what I do involve systems that if mishandled could lead to more than just a loss of profit. There may be less of these roles, but as it stands I see nothing to indicate they will not exist.
- techblueberry 5mo ago
  > Nope, just knowledge workers. We’re decades away from automating many manual labor professions, even “unskilled” ones. How do you figure? We’ve already automated away way more manual labor jobs than we currently have.
- esafak 5mo ago
  I think the manual professions are not far behind. It's just that the development is being done in China, so most readers here are not aware of it.
  - HDBaseT 5mo ago
    There is no robot electricians in China jumping up into ceiling crawlspaces running cable with proper adherence to the plethora of law and regulation.
    - esafak 5mo ago
      If that's kind of job left for us humans, I weep...
darkstarsys 5mo ago
Owners of capital, yes, but also owners of the means of production (which now means AI companies). See https://blog.oberbrunner.com/blog/ai-risks-taxonomy/#economic https://blog.oberbrunner.com/blog/ai-risks-taxonomy/#economi...
- trumpdong 5mo ago
  That's what capital means.
juleiie 5mo ago
That sounds more like a problem of close minded narrow focus on economic output instead of culture, virtue and spiritual traditions. AI is fundamentally an equivalent to slave economy. Cheap, plentiful workforce. This time ethically neutral. You either get Greece or Rome. I’d prefer Greece but it will probably be Rome. From the past we can predict the future.
- techblueberry 5mo ago
  > That sounds more like a problem of close minded narrow focus on economic output instead of culture, virtue and spiritual traditions. I’m starting to be more sensitive to the argument that without god, people are unable to have a strong moral foundation. Not for the people expressing creativity in how they fuck, but as a check on those in power.
  - juleiie 5mo ago
    There has been said so much about it in the past and I always believed it to be true, as an atheist. I like to think that one of the symptoms is politics becoming really absolutist, idealistic and cultish. You do not debate followers of a different religion. But many topics really becoming kind of a mini religions. I don’t know for sure though, there are arguments against it too and other factors. I think substantial amount of people really need some kind of subjective spiritual experience to their life and maybe ignoring that need breeds some maladaptive tendencies
  - goosejuice 5mo ago
    Most people already have god.
  - jplusequalt 5mo ago
    >i’m starting to be more sensitive to the argument that without god, people are unable to have a strong moral foundation. Not for the people expressing creativity in how they fuck, but as a check on those in power. If this were true, why did the medieval peasant have less rights and autonomy in society than we do now?
ilovecake1984 5mo ago
This is so blinkered and egocentric. Just because LLMs are good at translating English to code, doesn’t imply they are good at many other jobs. Coding isn’t that hard, it’s just not enjoyable to most people. The enjoyment has always been the barrier to entry.
- thin_carapace 5mo ago
  the grandparent commentor predicts 'that which replaces a sweng can replace anything'. llms sure do replace certain language related tasks, sometimes, when correctly piloted. however the majority of the world do not work language related jobs. perhaps if robotics firms bridge the gap to reality using some novel architecture this prediction could come a bit more true? until then it does seem blinkered to assume a set of weights could build a house. hard agree on the last statement. programming is language. if you're literate you can code.
- ChrisLTD 5mo ago
  Coding isn’t that hard to you. Coding effectively is hard AND painful for the majority of folks. And I’d venture to guess they don’t want to be responsible for debugging LLM output either. Just like I don’t want to fix a car, or do my own plumbing.
trumpdong 5mo ago
No lelanthran, software engineering and plumbing are not the same job. No lelanthran, LLMs can't be plumbers.
- lelanthran 5mo ago
  > No lelanthran, software engineering and plumbing are not the same job. No lelanthran, LLMs can't be plumbers. Who said that? More to the point, how many plumbers does society need?
  - hbcdbff 5mo ago
    > Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Direct quote
    - lelanthran 5mo ago
      >> Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack > Direct quote And, in your (and GP's mind), that means the same thing as "LLMs can replace plumbers"? After all, I said: >>>> When all human output is valued at the fractions of a penny per month of work, there is no future. I mean, I know it's fashionable to not read the article, but are we all really responding without even reading the comments? Are two paragraphs well beyond the attention span of the readers here? Okay, lets go with that asinine comeback: What do you think happens when the only work left for humans to do involves 100% physical labour and 0% thought? How many plumbers does a society need? Electricians? Even in construction, you can automate almost everything away with cranes and similar. Now imagine that all the doctors, all the office workers, all the warehouse workers, all the bankers, lawyers, teachers, ... basically any job that requires thought ... all those people are now joining the legions of plumbers. That sort of 1000x increase in supply will drive prices to pennies. The LLM doesn't need to replace plumbers directly; all it needs to do is replace everyone else, and the value of plumbers approach zero anyway.
      - trumpdong 5mo ago
hax0ron3 5mo ago
>Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Well, except for roles where being human is an inherent part of the value for customers: bartender, prostitute, certain kinds of boutique sales, professional athlete, stage actor, etc. And for roles that have to be human for legal reasons. Of course such roles make up a small part of the entire job market.
jesse_dot_id 5mo ago
I haven't seen an ounce of evidence that a LLM can replace a deeply experienced engineer. The harness turns coding into sculpting, which happens to be how I coded pre-LLM, but much slower, so I feel very much empowered by them. But I'm still definitely guiding the entire project. I haven't once felt like it wasn't required for me to be in the loop. The people and companies leaning into fully autonomous agents are high on their own supply, in my opinion. They're kicking back in their beach chairs as their pipelines spit out Stüssy S after Stüssy S, with massive architectural flaws and attack surfaces, just lighting gobs of money on fire. Yesterday, I created a fully functional POC for something really cool, it took me all day as I reshaped the agent's rough boilerplate ideas into usable components, and I never once hit a session limit on my $100 month Claude sub. I spent the majority of my time thinking about how I needed to prompt to turn what was in my head into working and secure code. You can't just give the agent a vague idea and expect anything less than a dumpster app. It's probably enough to fool C Suite people into believing the AI apocalypse is coming, which is the crux of the problem, and what is fueling what is certainly a gigantic bubble — but when it comes down to it, shitty software is shitty software. To fix it, you have to know what good software looks and feels like under the hood, and why it's shitty or feels bad to use. It might start up, but mutability, staying up, remaining stable, and remaining secure are very different stories. The clock is ticking on everything that has been developed by a LLM with a novice user behind the prompts.

p-e-w 5mo ago

> Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Nope, just knowledge workers. We’re decades away from automating many manual labor professions, even “unskilled” ones. Turns out brains just aren’t as special as we thought.

lanfeust6 5mo ago
The major blocker for manual labor automation in that fashion is cheap energy. China is ahead of the pack with the States' weight behind aggressive expansion of solar tech, and still can't do that.
theshackleford 5mo ago
> Nope, just knowledge workers. Nope, just a specific kind. Those who developed and cultivated only a very specific skill set at the expense of all others. I used to think being a generalist, and having persued technical roles with a people facing element was to my detriment, but it’s turned out to be the best decision I ever made.
- tempest_ 5mo ago
  I had the opposite thought. Being a generalist was very useful to me 5 years ago. Now AI models have made everyone a generalist. That wide but not terribly deep skillset was immediately devalued by the AI models. You can argue that the models fuck up 20 percent of the time, or that they make poor code but there is a massive part for the industry that is totally fine with that and I think people ignore it to their detriment.
  - theshackleford 5mo ago
    For context: I came from hardware, Linux, networking, telecoms, and datacenter infrastructure, not software development. I always wanted to go deep, but in practice my brain dragged me across many instead, which unintentionally made me very broad. I kept ending up in organisations where I was pushed back into such roles because its apparently where my "value" is. I give that context because unlike a lot of you, I’m not a world class FAANG engineer and never will be. It is from this context all of my thoughts on AI flow. I work with people who are trying to use AI to produce work involving entire markets, roles, skillsets and technologies they don't even know exist, let alone understand. > I had the opposite thought. Until I recently got pulled back deep into engineering despite not being hands on for close to a decade, so did I. I was pulled in not because of any pure technical capability but instead because it's been recognised the team requires more. The skills I thought served only to help me stay employed in any role in the most basic roles are increasingly turning out to be things other's do not have and are becoming increasingly important. These are skills I always assumed crucial to “baseline competency” for everyone, but yet where a significant amount of them do not, and these individuals are now finding themselves in positions where they are less useful than me as a result. Many of them can not simply be acquired from AI either, and require years of active growth and practice. > Being a generalist was very useful to me 5 years ago. Now AI models have made everyone a generalist. I think they could, but have not. Not at a scale required for me to have significant concern. AI works as well as the context you can provide, and you don't know what you don’t know. If the context is shallow, so to will be the output, even when it looks convincing and that “looks convincing” part I believe is the most dangerous part. As an example; I've been (recently) attached to an engineering team, despite last holding that title pre-2015, after AI assisted work contributed to a multi million dollar contract loss. A customer experienced an outage, it was "fixed" and everyone moved on. A month later another outage occurred of a greater scale. A huge amount of time was wasted doubling down on the original AI finding, because the actual root cause had not been identified or understood, because it had been "fixed". Turns out AI had identified and "fixed" a symptom, not a root cause. I was able to identify and resolve the real issue because I had wider operational and infrastructure context the team lacked, but the damage was done. Trust was gone, the client lost, and layoffs will follow. Those layoffs will be “because of AI,” but not any "10x'ing" of productivity. Instead it will be because plausible but wrong work made it into production and hid a very real problem as a result. That’s the issue with AI as I see it now. It generates answers that survive initial scrutiny while completely missing wider context leading to cases where more impactful but hidden problems are introduced. > That wide but not terribly deep skillset was immediately devalued by the AI models. Perhaps “generalist” was the wrong word here. Most "engineers" I have worked with are extremely deep in their area and surprisingly limited outside it. Even with AI, they struggle to move beyond their specialty because they lack broader foundations underneath not just modern infrastructure, but a range of areas equally important to the health of a business. My advantage has never been being the best engineer in the room, I knew early in my career I’d never compete with the engineer who can patch our kernel before upstream does, despite wishing I could. What ended up mattering instead was becoming the "95% guy" across infrastructure, networking, systems, operations, business, customer success, and people management that allows me to work with people/organisations and ultimately connect dots in a way even the best engineers I have worked with can not. AI can help you develop skills in areas you don't have, but starting with most of it in areas in which people have exactly none, and where people seem extremely resistant to developing it with or without AI, has me significantly further ahead in the curve. Ironically, at least in my experience so far, AI has made that more valuable, not less. > they make poor code I consider this to be the least important part. We have testing, review, and process for that. I believe (and have instructed juniors as such) that the real value of valuable technical people has never been producing rockstar code, or being a clone of Linus. It's in having a deep foundational understanding of the building blocks underpinning the now endless layers of abstraction, understanding consequences, tradeoffs, failure patterns, business impact, customer communication, customer wants and needs, and ultimately I guess to sum it up, organisational reality. This feels more important than ever when they can generate plausible looking technical output instantly that they may be able to validate, but equally produce plausible output in a huge range of areas they absolutely can not, but for which their successes in code have led them to believe they can. Because they underestimate what they don't know and in fact often assume they know far more than they do with no real basis for such a belief. On the whole, I think I would end my thoughts like this. For years I lived with the stress that "rockstar" engineers would lead to me eventually becoming irrelevant, much in the same way I might fear AI. So far, being 95% across customers, leadership, sales, support, engineering, and business strategy without losing the technical depth underneath it has meant this fear was unfounded and in fact put me ahead of them. I believe I am not isolated in this, and that in fact we will see more of it. All else aside, my roles as of the last decade often require me to be in the room and working with humans. AI has not changed this, and there is no current indication it will. The requirement will be that I continue to remain in the room, only now with AI. This is for many reasons including regulatory, because portions of what I do involve systems that if mishandled could lead to more than just a loss of profit. There may be less of these roles, but as it stands I see nothing to indicate they will not exist.
techblueberry 5mo ago
> Nope, just knowledge workers. We’re decades away from automating many manual labor professions, even “unskilled” ones. How do you figure? We’ve already automated away way more manual labor jobs than we currently have.
esafak 5mo ago
I think the manual professions are not far behind. It's just that the development is being done in China, so most readers here are not aware of it.
- HDBaseT 5mo ago
  There is no robot electricians in China jumping up into ceiling crawlspaces running cable with proper adherence to the plethora of law and regulation.
  - esafak 5mo ago
    If that's kind of job left for us humans, I weep...

darkstarsys 5mo ago

Owners of capital, yes, but also owners of the means of production (which now means AI companies). See https://blog.oberbrunner.com/blog/ai-risks-taxonomy/#economic https://blog.oberbrunner.com/blog/ai-risks-taxonomy/#economi...

trumpdong 5mo ago
That's what capital means.

juleiie 5mo ago

That sounds more like a problem of close minded narrow focus on economic output instead of culture, virtue and spiritual traditions. AI is fundamentally an equivalent to slave economy. Cheap, plentiful workforce. This time ethically neutral. You either get Greece or Rome. I’d prefer Greece but it will probably be Rome. From the past we can predict the future.

techblueberry 5mo ago
> That sounds more like a problem of close minded narrow focus on economic output instead of culture, virtue and spiritual traditions. I’m starting to be more sensitive to the argument that without god, people are unable to have a strong moral foundation. Not for the people expressing creativity in how they fuck, but as a check on those in power.
- juleiie 5mo ago
  There has been said so much about it in the past and I always believed it to be true, as an atheist. I like to think that one of the symptoms is politics becoming really absolutist, idealistic and cultish. You do not debate followers of a different religion. But many topics really becoming kind of a mini religions. I don’t know for sure though, there are arguments against it too and other factors. I think substantial amount of people really need some kind of subjective spiritual experience to their life and maybe ignoring that need breeds some maladaptive tendencies
- goosejuice 5mo ago
  Most people already have god.
- jplusequalt 5mo ago
  >i’m starting to be more sensitive to the argument that without god, people are unable to have a strong moral foundation. Not for the people expressing creativity in how they fuck, but as a check on those in power. If this were true, why did the medieval peasant have less rights and autonomy in society than we do now?

ilovecake1984 5mo ago

This is so blinkered and egocentric. Just because LLMs are good at translating English to code, doesn’t imply they are good at many other jobs. Coding isn’t that hard, it’s just not enjoyable to most people. The enjoyment has always been the barrier to entry.

thin_carapace 5mo ago
the grandparent commentor predicts 'that which replaces a sweng can replace anything'. llms sure do replace certain language related tasks, sometimes, when correctly piloted. however the majority of the world do not work language related jobs. perhaps if robotics firms bridge the gap to reality using some novel architecture this prediction could come a bit more true? until then it does seem blinkered to assume a set of weights could build a house. hard agree on the last statement. programming is language. if you're literate you can code.
ChrisLTD 5mo ago
Coding isn’t that hard to you. Coding effectively is hard AND painful for the majority of folks. And I’d venture to guess they don’t want to be responsible for debugging LLM output either. Just like I don’t want to fix a car, or do my own plumbing.

trumpdong 5mo ago

No lelanthran, software engineering and plumbing are not the same job. No lelanthran, LLMs can't be plumbers.

lelanthran 5mo ago
> No lelanthran, software engineering and plumbing are not the same job. No lelanthran, LLMs can't be plumbers. Who said that? More to the point, how many plumbers does society need?
- hbcdbff 5mo ago
  > Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Direct quote
  - lelanthran 5mo ago
    >> Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack > Direct quote And, in your (and GP's mind), that means the same thing as "LLMs can replace plumbers"? After all, I said: >>>> When all human output is valued at the fractions of a penny per month of work, there is no future. I mean, I know it's fashionable to not read the article, but are we all really responding without even reading the comments? Are two paragraphs well beyond the attention span of the readers here? Okay, lets go with that asinine comeback: What do you think happens when the only work left for humans to do involves 100% physical labour and 0% thought? How many plumbers does a society need? Electricians? Even in construction, you can automate almost everything away with cranes and similar. Now imagine that all the doctors, all the office workers, all the warehouse workers, all the bankers, lawyers, teachers, ... basically any job that requires thought ... all those people are now joining the legions of plumbers. That sort of 1000x increase in supply will drive prices to pennies. The LLM doesn't need to replace plumbers directly; all it needs to do is replace everyone else, and the value of plumbers approach zero anyway.
    - trumpdong 5mo ago

hax0ron3 5mo ago

>Anything that can replace a deeply experienced s/ware engineer can replace anyone in the employment stack Well, except for roles where being human is an inherent part of the value for customers: bartender, prostitute, certain kinds of boutique sales, professional athlete, stage actor, etc. And for roles that have to be human for legal reasons. Of course such roles make up a small part of the entire job market.

jesse_dot_id 5mo ago

I haven't seen an ounce of evidence that a LLM can replace a deeply experienced engineer. The harness turns coding into sculpting, which happens to be how I coded pre-LLM, but much slower, so I feel very much empowered by them. But I'm still definitely guiding the entire project. I haven't once felt like it wasn't required for me to be in the loop. The people and companies leaning into fully autonomous agents are high on their own supply, in my opinion. They're kicking back in their beach chairs as their pipelines spit out Stüssy S after Stüssy S, with massive architectural flaws and attack surfaces, just lighting gobs of money on fire. Yesterday, I created a fully functional POC for something really cool, it took me all day as I reshaped the agent's rough boilerplate ideas into usable components, and I never once hit a session limit on my $100 month Claude sub. I spent the majority of my time thinking about how I needed to prompt to turn what was in my head into working and secure code. You can't just give the agent a vague idea and expect anything less than a dumpster app. It's probably enough to fool C Suite people into believing the AI apocalypse is coming, which is the crux of the problem, and what is fueling what is certainly a gigantic bubble — but when it comes down to it, shitty software is shitty software. To fix it, you have to know what good software looks and feels like under the hood, and why it's shitty or feels bad to use. It might start up, but mutability, staying up, remaining stable, and remaining secure are very different stories. The clock is ticking on everything that has been developed by a LLM with a novice user behind the prompts.