Ask HN: How many jobs will AI eliminate? Which ones are safe?
41 comments
> Over the past few weeks, I’ve spoken with several VCs who confidently claim that companies won’t need to hire anyone in the next year or two because AI will handle everything.
An important part of being a VC appears to be the ability to believe anything, no matter how stupid, as long as it is currently on-trend, then to abandon it as if it had never existed when fashions change (for instance, see the blockchain craze, which was shortly followed by a distinct NFT craze that was just the blockchain one resprayed). I would not take what _VCs_ say as any particular guide to reality.
> Even Bill Gates recently said we might all be out of work soon and won’t need to do anything in an "age of abundance."
This would be the same Bill Gates who was claiming in the late 90s that by the early noughties our primary means of interfacing with our computers would be by talking to them, yes? I mean, again, you've got to consider the source.
An important part of being a VC appears to be the ability to believe anything, no matter how stupid, as long as it is currently on-trend, then to abandon it as if it had never existed when fashions change (for instance, see the blockchain craze, which was shortly followed by a distinct NFT craze that was just the blockchain one resprayed). I would not take what _VCs_ say as any particular guide to reality.
> Even Bill Gates recently said we might all be out of work soon and won’t need to do anything in an "age of abundance."
This would be the same Bill Gates who was claiming in the late 90s that by the early noughties our primary means of interfacing with our computers would be by talking to them, yes? I mean, again, you've got to consider the source.
I was sceptical about the VCs to start with but the Bill Gates quote made me really question it all! Thanks for sharing the story about him predicting voice as primary means of interfacing with computers.
In fairness it wasn't _just_ him (or Microsoft); everyone was very, very excited about voice in the late 90s.
Meanwhile, today, almost 30 years after Dragon NaturallySpeaking (which was kind of in the ChatGPT role; it seemed so magic that everyone thought everything was about to change), voice recognition is finally just about good enough that it can be used for unimportant things. Though personally I still wouldn't trust it to control my computer.
Meanwhile, today, almost 30 years after Dragon NaturallySpeaking (which was kind of in the ChatGPT role; it seemed so magic that everyone thought everything was about to change), voice recognition is finally just about good enough that it can be used for unimportant things. Though personally I still wouldn't trust it to control my computer.
Interesting thought experiment. I think accuracy is table stakes for a voice control system, I think the bigger challenge for adoption is discoverablility. It's hard to know what the exact syntax the voice prompt expects and everyone is different. I do wonder if LLMs and their ability to parse natural language will rapidly unlock that type of thing because they're a lot more flexible about how you command them
My personal opinion: It will be similar to how computers affected jobs.
No jobs are safe, but no skilled jobs are immediately on the chopping block either. AI will boost peoples' efficiency, such that it will take one skilled worker to do what it used to take five skilled workers to do. Law briefs, ad copy, scripting, code refactoring, engineering. Everything will use AI as a tool, but AI won't replace most people. People will instruct AIs to do sub-tasks and validate and integrate its responses. Similar to how people use computers today. Some peoples' specialized skills will be rendered obsolete, similar to what happened with secretaries, switchboard operators, elevator operators, proofreaders, etc. with computers.
I think it will change what it means to do many jobs, just like working with computers changed, for instance, accounting, or visual design. Working with specialized AI tools will become an essential part of many jobs, just like working with specialized software tools is today.
Long term, I think AI will eat further into current jobs, but I think they will be replaced with different human jobs. I think there will still be skilled careers and roles to be had, but I don't think any of us can predict what those jobs will look like.
No jobs are safe, but no skilled jobs are immediately on the chopping block either. AI will boost peoples' efficiency, such that it will take one skilled worker to do what it used to take five skilled workers to do. Law briefs, ad copy, scripting, code refactoring, engineering. Everything will use AI as a tool, but AI won't replace most people. People will instruct AIs to do sub-tasks and validate and integrate its responses. Similar to how people use computers today. Some peoples' specialized skills will be rendered obsolete, similar to what happened with secretaries, switchboard operators, elevator operators, proofreaders, etc. with computers.
I think it will change what it means to do many jobs, just like working with computers changed, for instance, accounting, or visual design. Working with specialized AI tools will become an essential part of many jobs, just like working with specialized software tools is today.
Long term, I think AI will eat further into current jobs, but I think they will be replaced with different human jobs. I think there will still be skilled careers and roles to be had, but I don't think any of us can predict what those jobs will look like.
As someone who believes AI will completely eliminate humans I wish I was could only be worried about job security.
But to answer your question, AI will ultimately eliminate every job in the sense that if you believe AI will eventually be able do any task, and do it for cheaper than a human labourer, then there would be no reason for you to ever pay a human labourer.
The best analogy here is probably how the industrial revolution made the horse redundant. It's not that horses can't be used for physically demanding work anymore, it's simply that it's cheaper in practically every imaginable case to use a machine so horses are no longer used.
Of course, machines didn't replace the use of horses in all industries over night and we should expect the same will happen with AI. I'll assume this is what you're asking.
It seems rather obviously to me that knowledge workers will be the easiest to replace with AI – especially knowledge workers who work purely digitally. The hardest work to replace will likely be physical labour since robotics is still very far from biological sophistication required to be competent at many physical jobs. And even when it starts to become competent it's going to take a while to build the robots to replace labourers and do so at a competitive cost.
But to answer your question, AI will ultimately eliminate every job in the sense that if you believe AI will eventually be able do any task, and do it for cheaper than a human labourer, then there would be no reason for you to ever pay a human labourer.
The best analogy here is probably how the industrial revolution made the horse redundant. It's not that horses can't be used for physically demanding work anymore, it's simply that it's cheaper in practically every imaginable case to use a machine so horses are no longer used.
Of course, machines didn't replace the use of horses in all industries over night and we should expect the same will happen with AI. I'll assume this is what you're asking.
It seems rather obviously to me that knowledge workers will be the easiest to replace with AI – especially knowledge workers who work purely digitally. The hardest work to replace will likely be physical labour since robotics is still very far from biological sophistication required to be competent at many physical jobs. And even when it starts to become competent it's going to take a while to build the robots to replace labourers and do so at a competitive cost.
I think it’s already having an effect on labor jobs too, and will continue to ramp.
Not because of robots (but that will come soon enough), but because you can take a picture of leaky plumbing or a home maintenance task and upload it and ask for step by step directions on how to DIY.
I’ve saved thousands doing this already this month alone.
Not because of robots (but that will come soon enough), but because you can take a picture of leaky plumbing or a home maintenance task and upload it and ask for step by step directions on how to DIY.
I’ve saved thousands doing this already this month alone.
Is this really a trend ? It surely is helping a DIY enthusiast to perform a new task, but I doubt it will convert a non manual labour enclined person to do the task.
Also these ready made DIY how-tos work for a certain country only and can hardly be made global. E.g Most of american made how-tos make no sense for someone living in Europe
My wife used ChatGPT to fix a garage door opener while I was out. Uploaded a picture, and followed the instructions. She's not a manual labor inclined person by any stretch and knew nothing about garage door openers.
How many people call a plumber, HVAC, or auto repairman because they don't want to fix it themselves vs don't know how to fix it?
I'm guessing the overlap of those who know how to fix it with those willing to pay is not that large.
That could be because those who really know how to fix it, have a pretty good idea how much expertise and effort the fix requires. They are less willing to be overcharged or have unnecessary options added.
the industrial revolution made the horse redundant
I love this analogy. People hope they'll be like carriage drivers who transitioned into driving cars. Not this time. On the other hand, I'm not sure how this transition actually affected the horses — they weren't killed off or driven to extinction, were they?
I love this analogy. People hope they'll be like carriage drivers who transitioned into driving cars. Not this time. On the other hand, I'm not sure how this transition actually affected the horses — they weren't killed off or driven to extinction, were they?
The horse analogy helped me a lot to understand this better. Over a longish timeline most jobs disappear in their current form & generate other jobs which we aren't even aware of rn
> I’ve spoken with several VCs who confidently claim that companies won’t need to hire anyone in the next year or two because AI will handle everything.
VCs jobs rely on following and hyping trends. You can safely discount most things they say by a lot.
VCs jobs rely on following and hyping trends. You can safely discount most things they say by a lot.
Even if LLMs can't do everything yet, I have a feeling there is a malaise of wanting to do work in the tech industry.
Exactly! And i really want to find out how much of this is pure hype.
No one is going to be safe. But some jobs are safer than the others.
I'd say anything physical is safer than knowledge jobs.
I'd say anything physical is safer than knowledge jobs.
And sales. People ain't gonna sign large contracts with a bot.
At the very least, they'll need contractors to take people out for dinner.
At the very least, they'll need contractors to take people out for dinner.
People ain't gonna sign large contracts with a bot.
True. Bots will sign contracts with other bots.
True. Bots will sign contracts with other bots.
The VC idea is incorrect. AI in some areas where companies compete like sales is just adding to the arms race. It's also rising expectations making products more complex. It's being used to make presentations better which is raising expectations.
Translators jobs are going away but the low end is disappearing. Graph ic artists selling logos now need to offer a bigger branding experience.
Translators jobs are going away but the low end is disappearing. Graph ic artists selling logos now need to offer a bigger branding experience.
Translations of products are now clearly AI without proof reading. It's so bad you have to switch to english to understand what it is supposed to mean. Even big brands like Microsoft cannot translate properly.
So yes, they save on translation short term, but by making your product unprofessional.
So yes, they save on translation short term, but by making your product unprofessional.
> anyone in the next year or two because
If you believe this, just buy otm calls on s&p500 and you won’t have to worry about a job.
If you believe this, just buy otm calls on s&p500 and you won’t have to worry about a job.
Maybe it will out perform their fund too lol
Yes, vcs can be stupid and blind sometimes too. Just like all of us.
Smart companies now will hire the best of the best from a large pool of people laid off by dumb companies.
Fast forward 10 years - dumb companies are drowning in false hopes, bad staff and mess they created for themselves. While smart ones growing and replacing them.
Smart companies now will hire the best of the best from a large pool of people laid off by dumb companies.
Fast forward 10 years - dumb companies are drowning in false hopes, bad staff and mess they created for themselves. While smart ones growing and replacing them.
Very interesting perspective
Most of it is extrapolation. Yes, it'll triple productivity. But from what we've seen with frameworks etc, that just ends up creating jobs. We work more hours today than before the computers came in.
The other problem is the unfortunate naming. Nobody talks about Jetbrains taking coder jobs but it's essentially doing the same thing, allowing people write more lines of code with less typing. How many jobs were eliminated with Zapier or Pipedream? "AI" implies that it replaces a human.
The other problem is the unfortunate naming. Nobody talks about Jetbrains taking coder jobs but it's essentially doing the same thing, allowing people write more lines of code with less typing. How many jobs were eliminated with Zapier or Pipedream? "AI" implies that it replaces a human.
even simple stuff like language translation for corporations, has shed significant jobs already
it went from a large cottege industry, requireing skill and concentration, to skimming the computers work for....wierdness
lots of government stuff is ripe for converting to
web forms, and smaller municipal offices...,that have say, already offshored there call centers, will be looking at putting as much as posdible in a box
and there is good evidence that when government is streamlined, there is more public engagement, and higer revenues for taxes and fee's.Lots of stuff that isn't AI per say,but just good automation that allows people to self serv, is going to eliminate (hopefully) a lot of jobs, that almost no one will miss
Anything that involves dealing with stakeholders and figuring out how an implementation can add business value, including gathering requirements and managing implementations.
But no, AI is not going to completely replace developers.
But no, AI is not going to completely replace developers.
One thing I am having a hard time wrap my head around is that in its current form AI might not replace developers or say sales callers completely but what about 3-4 years from now? Does AI continue to get exponentially better or will we hit some kind of tech ceiling due to inherent limitations of LLMs?
It’s not about sales callers who are the “top of the funnel”. Many B2B solutions require customizations and take time to implement and make work with the clients business. Think cloud migrations and consulting for new products, SalesForce, various CRM and whatever business processes that Oracle does, etc.
You have sometimes dozens of people across multiple departments that you have to coordinate and train.
I can’t say where AI will be. But it might get good enough to know how to do something. It won’t be able to know what to do and what people need.
You have sometimes dozens of people across multiple departments that you have to coordinate and train.
I can’t say where AI will be. But it might get good enough to know how to do something. It won’t be able to know what to do and what people need.
Can you honestly name a single billionaire who is going to be happy to share this supposed 'abundance'? They are going to use their capital to extract as much wealth as possible and they have no intention of letting the average person live out a comfortable life without working.
These threads show off one thing: that the world is full of bullshit jobs, and companies selling bullshit products.
If you work in AEC, youll see how far away we are from "fully replacing" anything.
Are we really going to start erected skyscrapers and bridges, wait for them to collapse, then wait for the gov to slap AI on the wrist?
Get a job in building engineering, mech eng and come back to me. Theres a lot of ML and assistance, but humans are still thinking and collaborating.
If you work in AEC, youll see how far away we are from "fully replacing" anything.
Are we really going to start erected skyscrapers and bridges, wait for them to collapse, then wait for the gov to slap AI on the wrist?
Get a job in building engineering, mech eng and come back to me. Theres a lot of ML and assistance, but humans are still thinking and collaborating.
No jobs will be eliminated. Unless they're aiming to lay people off anyways -- if anything, companies will be slower to hire.
They're already being eliminated... I'm using freelance designers much much less these days because I just don't need them anymore with AI tools.
I've seen others use AI so they don't have to hire engineers to build things (basic things like simple websites, but still).
I've seen others use AI so they don't have to hire engineers to build things (basic things like simple websites, but still).
The problem becomes when you want the logo to match the rest of the design. The bar raises on what a simple website is.
Call centres are the first target for full automation. If I remember correctly they represent 3 million jobs in the US only.
There are no "safe jobs" in capitalism.
There are no "safe jobs" in capitalism.
> "age of abundance"
No such thing will come to pass in the common way you think. This is classic doublespeak. Similar to "You will own nothing and be happy". If you think hard on it, the only way that can be true is by limiting the scope of the number of people involved, and the only way to do that is if a large number of people die, and that is what they mean and intend when they do not explicitly spell out how you get from statement to reality, addressing the complex hidden dependencies.
It is unlikely that any blue-collar job will be impacted aside from the 2nd order effects that occur as displaced workers move to other areas as a necessity for survival.
That won't last long though before the monetary and economic systems break down as they will have more chaotic whipsaws than the labor market.
As for how safe are other jobs. You can expect all white collar low complexity work, normally found in entry level positions on career ladders, and related support staff, to be at extreme risk.
The upper echelon of senior positions, and even mid-level positions are safe, due to the abstract nature of the tasks required, but are confounded by additional costs being imposed by third-parties (i.e. ghost jobs/postings). There's no good way to evaluate you have this degree of experience/knowledge, and worse you lose this type of knowledge if you don't use it regularly.
The career pipeline however is a sequential pipeline. If there is no one going into the pipeline because AI has blocked and replaced the entry-level worker as a surrogate, and there is no one for the mid and senior level people to pass knowledge to because they've all been replaced by an AI it won't happen, and as those mid and senior level people age out you find yourself in a hysteresis problem that cannot be solved in the time period needed. The same as any ponzi scheme. Three stages.
The funding mechanism for almost all companies today ensures this is always a problem for next quarter, creating a cascading failure, or slow motion trainwreck of costs where the employer will be unable to find any labor meeting that criteria for any price by stage 3.
If this happens in critical national dependencies then you get a full blown socio-economic collapse in an environment of ecological overshoot (Malthus/Catton) that is global in nature (via reserve currency). Not pretty, 1/3 of 4bn people might survive such, maybe, and that's as a global population.
We've known how to build resilient systems for quite some time, but resilient systems cost more and in furtherance of globalization optimization has coincided towards brittle single-point-of-failures broadly. It allows the most chaos, which speculators and companies alike use to reap gross profits at everyone elses expense through plausible narratives backed by price fixing.
No such thing will come to pass in the common way you think. This is classic doublespeak. Similar to "You will own nothing and be happy". If you think hard on it, the only way that can be true is by limiting the scope of the number of people involved, and the only way to do that is if a large number of people die, and that is what they mean and intend when they do not explicitly spell out how you get from statement to reality, addressing the complex hidden dependencies.
It is unlikely that any blue-collar job will be impacted aside from the 2nd order effects that occur as displaced workers move to other areas as a necessity for survival.
That won't last long though before the monetary and economic systems break down as they will have more chaotic whipsaws than the labor market.
As for how safe are other jobs. You can expect all white collar low complexity work, normally found in entry level positions on career ladders, and related support staff, to be at extreme risk.
The upper echelon of senior positions, and even mid-level positions are safe, due to the abstract nature of the tasks required, but are confounded by additional costs being imposed by third-parties (i.e. ghost jobs/postings). There's no good way to evaluate you have this degree of experience/knowledge, and worse you lose this type of knowledge if you don't use it regularly.
The career pipeline however is a sequential pipeline. If there is no one going into the pipeline because AI has blocked and replaced the entry-level worker as a surrogate, and there is no one for the mid and senior level people to pass knowledge to because they've all been replaced by an AI it won't happen, and as those mid and senior level people age out you find yourself in a hysteresis problem that cannot be solved in the time period needed. The same as any ponzi scheme. Three stages.
The funding mechanism for almost all companies today ensures this is always a problem for next quarter, creating a cascading failure, or slow motion trainwreck of costs where the employer will be unable to find any labor meeting that criteria for any price by stage 3.
If this happens in critical national dependencies then you get a full blown socio-economic collapse in an environment of ecological overshoot (Malthus/Catton) that is global in nature (via reserve currency). Not pretty, 1/3 of 4bn people might survive such, maybe, and that's as a global population.
We've known how to build resilient systems for quite some time, but resilient systems cost more and in furtherance of globalization optimization has coincided towards brittle single-point-of-failures broadly. It allows the most chaos, which speculators and companies alike use to reap gross profits at everyone elses expense through plausible narratives backed by price fixing.
Even Bill Gates recently said we might all be out of work soon and won’t need to do anything in an "age of abundance."
How much of this is hype vs. reality?
Which jobs are likely to be completely eliminated by AI and which ones are likely to survive or even thrive?