Our gestating Machine God doesn't yet have hands as good as ours, so good this is good advice.
We had a system of overproduction of sentient office equipment, who waste their time on pointless Zoom calls and sending emails that no one reads. That had to end sooner or later, and was already about to collapse on its own. BS job holders are miserable anyway. Let's give these people purpose (closing the gap between activity and tangible output), free them from debt-slavery, and fix all the broken infrastructure around us.
I mean, good? We need more tradespeople than we do office workers - and we need everyone in both fields to be paid substantially more than they presently earn, which will only come about through new business creation, something tradespeople dominate. Software engineers won’t power the green energy transformation so much as skilled electricians, architects, contractors, and builders. More finance people won’t replace aging and inefficient infrastructure so much as more road workers, pipe fitters, plumbers, and wastewater engineers.
I think the current (over)hype around LLMs replacing jobs wholesale is an excellent catalyst for this shift, but I also acknowledge that the pendulum was already naturally swinging that way after decades of over-prioritizing white collar work as the only means of joining the shrinking middle class.
I'm an industrial electrician. I have zero fears of being replaced by any any sort of AI. Maybe by someone younger and smarter, but I have 38 years experience. The trades are a decent living, and lots of people could do worse.
Something that bugs me about all these articles. It’s all bullshit. Almost any future is possible at this point and the data can be analyzed in almost any way. Could be AI, could be the economy, could be vibes. You’ll just get stupider reading all this shit.
I wonder what effects this will have in the long run, as people attuned to white-collar labor and culture begin to assume jobs usually held by people who aren’t...
I generally don’t believe in Panacea’s, and if these jobs were so great and the solution to all our problems, why is it when we live in a country where a large percentage of folks are rural or otherwise have blue collar values, there not pursuing them?
Don’t get me wrong, I’m not saying all blue collar jobs are bad, many are fantastic! But like all things, it’s likely a pyramid. Just like a lot of software engineering jobs are shitty.
If it is offered as a panacea, then the glut of applicants will just drive down wages.
Blue collar jobs are only going to be the go-to for as long as the labor supply is lower than the commercial demand, which I doubt will be long (if this is even the case now). It takes a non-neglible amount of time to train (or certify) for these jobs in much of the western world, which means that any bump in demand is met with a significant lag in supply. Once that supply catches up, the momentum of people saying "bro just go to trade school instead of college, plumbers/electricians/welders/dirt farmers make $100k" is going to dump thousands and thousands of juniors into a market that will very quickly oversaturate. We are witnessing this at an accelerated rate with software development and other tech jobs, largely due to extreme external factors (high interest rates limiting VC cash flow, Trump's hostile domestic economic policies[0], demonstrable downward pressure on employment mobility and wage suppression from big tech[1]).
The american "middle class" is still shrinking, but now it's shrinking faster than ever before, largely because the capital class wants more money, and there is one more stone to bleed. Creating a market of blue collar professionals who will be blackballed from white collar markets due to their educational and work histories (in tandem with the desired outcome of using AI workloads for these jobs in lieu of people) will raise the commercial value of these white collar services, while gatekeeping their entry and stagnating/lowering their compensation. The ladder is being set on fire.
I am UX Researcher, Designer and UI Developer. Up until a week ago (company gave permission to use GPT) per some tests I did (personally) with GPT i thought indeed it can do my job yet as of today and using GPT 5...
Logo & Graphic Creation....
- Does this the best so far, but asking it to edit a graphic it created / you liked it doesn't always do what you asked. So, playing with GPT to get it right vs. taking what it initially gave you (same as going to a stock photo site and finding graphics to manipulate) and opening Photoshop is still same amount of time. So here GPT is just another stock photo site like resource yet can take TIME to generate. Is it pulling from a stock site anyway?
- Web Design
On your first attempt to ask it to create a solid web design it works good yet asking it to make edits to it forget lol. It will go and create totally new design and at times on the first attempt it chooses to create a design with a width of 800px (cutting off the left and right sides of the full design).
Front-End Coding...
It does not host images and or provide an entire zip of a website it coded. My Front-End skills are still needed until it provides a zip and handles images properly!
UX Research...
These skills I feel are safest from AI as it requires interfacing with users!
Mentioned in the article text, but not the title, is the fact that blue collar work typically does not require a college degree. You can start getting paid immediately after high school. Contrasted with a possible alternate path of four to five years of undergrad at a third tier college paying $80k/yr and financed at an 11% APR...
The real secular arc here predating the GenAI rush has been the decreasing ROI of a generic college degree.
Who are these trades going to sell their services to when a large proportion of people employed in white collar work are looking at a prospect of reduced income or loss of jobs?
When I see stories like this all I can think of is that rich people must really hate paying their blue collar workers market value and want to flood the market to lower their wages and bargaining ability. Especially when I hear how hard it is to actually get a proper blue collar job (that isn't just a manual labor with a bit of carpentry job paying below a living wage) and be apprenticed in it from my blue collar friends.
The problem with Blue-collar work under AI is that your white collar AI manager can fork itself and have infinite time to micromanage you.
I foresee people being asked to wear AR glasses and/or work in a digital camera panopticon to be constantly evaluated on performance and compliance with (disconnected and disaffected) policy (imagine if your checkbox-first-results-never corporate compliance officer can design new checkboxes 100x as fast as a human one could have). When the machines can real-time analyze and provide "corrective" feedback or "training" to a pool of juniors who don't feel like they deserve rights or pay and have never even heard of a union, the value of skill in labor drops to nothing, and only people who are able to perform blind obedience will be valuable in the market.
If management is infinite heartless machines chasing profit motive, then every job becomes Amazon Fulfillment on steroids.
The safest are the licensed professions because numbers of new entrants are regulated by cartel and people won't be able to let go of the demand because again, regulations. Simply put, safest jobs will be red-taped sinecures.
People are going to be surprised when the cheap humanoid robots out of China get paired with transformer models for planning and control.
I got a bit caught with my pants down between GPT2, which was so hyped, and GPT3.5 which was also hyped but the delta between reality and Sam's bullshit was much smaller. I've been keeping my eye out on both the hardware and software and we're in the pre-gpt2 days for robotics right now. There's a lot of cute little results that need a few million dollars to get put together.
The next 5 years will make the effect of transformers on office work since 2022 seem like a slight breeze compared to a tornado.
I think it'll be a while (10-20 years?) until we have AGI good enough to at least potentially replace a non-trivial number of workers as opposed to just being a productivity tool. LLMs are clearly just a tool.
We will of course eventually create human-level AGI, but things like emotions and empathy will come later, if at all, which may limit its utility in interacting with humans and therefore usefulness for many jobs.
The major job categories that AGI might be capable of is "professional and business" office jobs and many government jobs, which each account for about 15% of US jobs, but what that leaves is the other 70% which includes not only "blue collar" jobs such as trades, manufacturing, construction, transportation, but also major employers such as healthcare, education, leisure & hospitality, etc.
The future doesn't always turn out as expected though, especially when it comes to sci-fi visions of people in flying cars, living on mars, etc. AGI and robotics may well turn out to be the same - we may imagine humanoid robots and human-like "remote workers" doing many jobs, but it'll probably not be like that. Humanoid workers are probably about as likely as flying cars - problematic and just not that useful. AGI "workers" may well be just agents that are given well defined tasks to execute as opposed to fully autonomous human worker replacements.
This is another recycled version of the "college is a scam" trope that comes around during every recession. I first experienced it firsthand when I graduated smack dab in the middle of the Great Recession of 2008-09. Those who had majored in the so-called "basket weaving" disciplines were doomed, while those in "job-ready" disciplines like comp sci, accounting and finance were also doomed, but marginally less so as they'd bounce back when the S&P500 did.
It's far too simple to say "train to be a plumber" or HVAC, construction, or auto mechanic or whatever people latch on to as "future-proof / recession-proof". The reluctance to take these jobs is not the resistance to doing manual labour, but that lifetime income is limited unless you're the boss.
As with any role in a broadly non-unionized labour market, you only make proper money if you incorporate, strike deals of your own, hire people and pay yourself multiples more than the labour. Otherwise, your time, labour and frankly, bodily integrity is at the mercy of whoever signs your paychecks. And just like every other industry, only a small proportion of workers are cut out to be biz dev or people managers, and as a boss, you need to be able to do both.
Demand for electricians or mechanics will not grow proportionately with growth in electrician training programs, any more than SW Eng jobs did with the growth of bootcamps in the 2010s. The difference was that coding jobs were often paid for on the back of venture capital, or on the back of a giant, profitable corporation. The trades are not like that, you can't be paid more than what the company brings in revenue-wise.
25 comments
[ 4.5 ms ] story [ 45.1 ms ] threadWe had a system of overproduction of sentient office equipment, who waste their time on pointless Zoom calls and sending emails that no one reads. That had to end sooner or later, and was already about to collapse on its own. BS job holders are miserable anyway. Let's give these people purpose (closing the gap between activity and tangible output), free them from debt-slavery, and fix all the broken infrastructure around us.
I think the current (over)hype around LLMs replacing jobs wholesale is an excellent catalyst for this shift, but I also acknowledge that the pendulum was already naturally swinging that way after decades of over-prioritizing white collar work as the only means of joining the shrinking middle class.
Don’t get me wrong, I’m not saying all blue collar jobs are bad, many are fantastic! But like all things, it’s likely a pyramid. Just like a lot of software engineering jobs are shitty.
If it is offered as a panacea, then the glut of applicants will just drive down wages.
When you have a salary, you’re getting paid what you’re getting paid and nothing more and hoping whatever raise might come is better than inflation
The american "middle class" is still shrinking, but now it's shrinking faster than ever before, largely because the capital class wants more money, and there is one more stone to bleed. Creating a market of blue collar professionals who will be blackballed from white collar markets due to their educational and work histories (in tandem with the desired outcome of using AI workloads for these jobs in lieu of people) will raise the commercial value of these white collar services, while gatekeeping their entry and stagnating/lowering their compensation. The ladder is being set on fire.
0. https://news.ycombinator.com/item?id=44180533 1. https://equitablegrowth.org/aftermath-wage-collusion-silicon...
Logo & Graphic Creation....
- Does this the best so far, but asking it to edit a graphic it created / you liked it doesn't always do what you asked. So, playing with GPT to get it right vs. taking what it initially gave you (same as going to a stock photo site and finding graphics to manipulate) and opening Photoshop is still same amount of time. So here GPT is just another stock photo site like resource yet can take TIME to generate. Is it pulling from a stock site anyway?
- Web Design
On your first attempt to ask it to create a solid web design it works good yet asking it to make edits to it forget lol. It will go and create totally new design and at times on the first attempt it chooses to create a design with a width of 800px (cutting off the left and right sides of the full design).
Front-End Coding...
It does not host images and or provide an entire zip of a website it coded. My Front-End skills are still needed until it provides a zip and handles images properly!
UX Research...
These skills I feel are safest from AI as it requires interfacing with users!
looking forward to the Learn2Plumb bootcamps on HN
all these type articles do is manufacture consent for layoffs
The real secular arc here predating the GenAI rush has been the decreasing ROI of a generic college degree.
I foresee people being asked to wear AR glasses and/or work in a digital camera panopticon to be constantly evaluated on performance and compliance with (disconnected and disaffected) policy (imagine if your checkbox-first-results-never corporate compliance officer can design new checkboxes 100x as fast as a human one could have). When the machines can real-time analyze and provide "corrective" feedback or "training" to a pool of juniors who don't feel like they deserve rights or pay and have never even heard of a union, the value of skill in labor drops to nothing, and only people who are able to perform blind obedience will be valuable in the market.
If management is infinite heartless machines chasing profit motive, then every job becomes Amazon Fulfillment on steroids.
No-one will be spared.
I got a bit caught with my pants down between GPT2, which was so hyped, and GPT3.5 which was also hyped but the delta between reality and Sam's bullshit was much smaller. I've been keeping my eye out on both the hardware and software and we're in the pre-gpt2 days for robotics right now. There's a lot of cute little results that need a few million dollars to get put together.
The next 5 years will make the effect of transformers on office work since 2022 seem like a slight breeze compared to a tornado.
We will of course eventually create human-level AGI, but things like emotions and empathy will come later, if at all, which may limit its utility in interacting with humans and therefore usefulness for many jobs.
The major job categories that AGI might be capable of is "professional and business" office jobs and many government jobs, which each account for about 15% of US jobs, but what that leaves is the other 70% which includes not only "blue collar" jobs such as trades, manufacturing, construction, transportation, but also major employers such as healthcare, education, leisure & hospitality, etc.
The future doesn't always turn out as expected though, especially when it comes to sci-fi visions of people in flying cars, living on mars, etc. AGI and robotics may well turn out to be the same - we may imagine humanoid robots and human-like "remote workers" doing many jobs, but it'll probably not be like that. Humanoid workers are probably about as likely as flying cars - problematic and just not that useful. AGI "workers" may well be just agents that are given well defined tasks to execute as opposed to fully autonomous human worker replacements.
It's far too simple to say "train to be a plumber" or HVAC, construction, or auto mechanic or whatever people latch on to as "future-proof / recession-proof". The reluctance to take these jobs is not the resistance to doing manual labour, but that lifetime income is limited unless you're the boss.
As with any role in a broadly non-unionized labour market, you only make proper money if you incorporate, strike deals of your own, hire people and pay yourself multiples more than the labour. Otherwise, your time, labour and frankly, bodily integrity is at the mercy of whoever signs your paychecks. And just like every other industry, only a small proportion of workers are cut out to be biz dev or people managers, and as a boss, you need to be able to do both.
Demand for electricians or mechanics will not grow proportionately with growth in electrician training programs, any more than SW Eng jobs did with the growth of bootcamps in the 2010s. The difference was that coding jobs were often paid for on the back of venture capital, or on the back of a giant, profitable corporation. The trades are not like that, you can't be paid more than what the company brings in revenue-wise.