I follow Goldman Sachs multiple years and their predictions are always changing. I remember in 2020 after March crash they had this prediction (https://fortune.com/2020/03/20/coronavirus-economic-predicti...) which in reality completely opposite happened. There were multiple other similar cases that their prediction was wrong and after sometime they just changed their prediction.
Few people could probably predict the amount of spending from governments all over the world. What would've happened if they just shut everyone down? Their prediction doesn't sound far fetched. Now the government's went on a spending spree and we got massive inflation instead.
I started a project where a guide with art history background was writing the content... no we use GPT4 for 80% of it.
A lot of job are at risk. Most of them no because of full automation, but because less people will be required because of productivity increases "thanks to" AI. And many more job will become boring.
The problem is this productivity gain will hurt a lot of people, and there are probably/possibly only a small minority of people that will fully benefit from it
You will have to proof read anything it writes and you should also fact check it, otherwise you are taking a risk. However, this is very difficult to do if GPT4 is unable to provide references or citations, or any provenance at all.
The human with the art history background at least has some credentials, whereby the output can be trusted, to some degree and explained if necessary.
If you have a model that can read text and extract its meaning; and a model that can generate text, the work required to create a model that can generate text and then query for other texts that support its claims is not very hard.
This is, imo, such an obviously near term capability that its not worth bringing up as a limitation.
> If you have a model that can read text and extract its meaning
That model doesn't exist.
We have models that can read text and then probabilistically create a summary that's the most likely to match the original.
But none of these models "extract meaning" in a way that can be used for fact checking purposes in the way you describe.
What you're describing is basically a world model, and while there's a few folks claiming these LLMs are showing the glimmers of building world models, it's mostly speculation.
I’m not sure I’d consider illustrator and artist jobs to be “generating bullshit” and I’d say at least 90% of them are probably going to be eliminated. You can probably add some other non-bullshit media jobs like audio mastering and video colouring to the list.
Are you actually joking? I think you should go have a look at Midjourney v5 and have a read through some Reddits in the art community where people are already losing their jobs.
Is that where your 90% estimate comes from? Browsing Reddit?
You are the one with the extraordinary claim (and it therefore requires extraordinary evidence).
Well you’ve got financial analyst firms left right and center coming out with big numbers like this that have far more access to resources and data than I ever will. But you could just use your common sense. Most illustrators and artists make their money doing commissions for businesses. Most businesses will opt for the cheapest solution that meets the quality standard they’re after. The quality standard has already been met and it is far cheaper and faster than hiring an illustrator. So tell me how are all these illustrators and artists going to carry on making a living producing art if they’re priced out of the market?
I acknowledged in my first post that a particular quality standard had already been met by AI. I also concede that this quality standard could be sufficient for many businesses. God help us all if it's 90% of businesses.
Sadly I think that’s where it’s heading. We’ve already got this dynamic where the majority of people get their stuff from the same big brands, and then increasing levels of luxury brands for the top 5% or so. I think it’s going to end up like that, where luxury brands who can afford it will brand the human touch as “luxurious craftsmanship” and then the other brands will have no alternative but to use AI because competition is fierce and everything is a race to the bottom.
Even for stuff like movie making, a director or producer would hire an illustrator or team of illustrators and send off a script to get it storyboarded and probably have to wait at least a few weeks. Now they can just go to a prompt engineer and have it done in a single afternoon. There’ll be similar developments in VFX, if you have a look at the latest Unreal Engine 5.2 demo they do motion capture from a woman’s face from a mobile phone camera in a matter of minutes. The actress in the demo says that kind of thing used to take a month to get the same result. When you’ve got a bunch of competing streaming services and studios trying to churn out as much content as possible to keep subscribers, it’s going to be difficult for them to justify the slowness and cost of keeping those human workers compared to the AI.
Well humans have adapted, just as they have adapted to every technological development. There are tons of jobs out there with shortages e.g paramedics, nurses, teachers, engineering, plumbers, electricians, construction etc. People will end up doing these jobs if that's the only way they can make a living. It does majorly, majorly suck for everyone whose dream it was to have one of the creative jobs that are being replaced. I'm not saying it's right, I'm just saying it's probably what is going to happen.
I believe GPT-X will generate tons of ... something ...
How you would force people to READ IT ?
If you know your incoming info is 90% BS, not a guess, not a thought, just knowledge that this is all auto-generated by some multiCPU entity - why would you read it? What are going to gain or to learn?
I have worked with deps around the world of JPM, GS and others and others and there are not many places I have seen where there are so many replaceable and basically not really employable people. They could’ve been replaced with some code 10 years ago; now it’s only easier. Not necessarily AI but AI won’t make it better for them. These are very large numbers who are only there to pump the number of employees under their managers or some other lame metric.
Of course it does not mean they will do anything about it.
I'm assuming by "deps" you really meant "devs", and if so, I'd have to agree. I worked in the investment banking world for some years in a back office IT role and there were mini armies of devs there doing the most mundane, basic CRUD style work for years on end getting paid cushy 6-fig salaries. Not FAANG level, but solid nonetheless.
I would not at all be surprised if these large firms hired consultants from places like McKinsey to come by and analyze the work their IT staff is involved in to see how tools like GPT-4 could be leveraged to "streamline" (aka reduce the headcount of) these orgs, and then dramatically restructured them based on their findings. The only thing keeping many of these folks around is the political clout of their managers and department heads.
> The only thing keeping many of these folks around is the political clout of their managers and department heads.
People love saying that, but why wouldn't an exec take a multi-$million bonus to cut that dead weight? Or Private equity buy it out to gut it and resell it?
That happens as well, but a little while later the ‘organigram’ under these managers will be the same size or larger as, for some reason, more people under you means more money. And every manager has the same thing, so this happens at every level; I know managers only in title who have 2 persons under them and that entire ‘team’ does exactly nothing outside looking busy and taking pay. That way that ‘manager’ who was supposed to be a leaf in the tree and not a manager, makes more than the people under them, so win for them and the managers above them can boast both more people and more teams, so they get a nicer title and make more. At banks.
The actual report doesn't say lost or degraded, but rather 'exposed to automation', and right after that statement, they also mention the historical trend of automation has been moving people around:
>The good news is that worker displacement from automation has historically been offset by creation of new jobs, and the emergence of new occupations following technological innovations accounts for the vast majority of long-run employment growth.
This article is a misrepresentation of Goldman's analysis.
Can anyone here provide evidence of automation moving jobs to equal quality or higher quality employment?
Because what I’ve seen and read does not confirm this, in fact it disproves it. I can find multiple reports showing de industrialisation of the US economy happened mostly due to automation.
It highlights our preposterous assumption that in order to live everyone must have a "job". Why? If people become destitute then some others will gain even more wealth than they need. Is that a great way for humanity to behave?
Doesn't sound that much in global terms compared to the global population. And does it account for all the new jobs that will be created along the way? When I started out there was nothing like a DevOps engineer, social media influencer or Deliveroo driver...
The point being that this happens all the time and is a natural consequence of progress and evolution.
I expect jobs to be lost, but new jobs to be created. For example, prompt engineers. There is already a job board for prompt engineers (https://news.ycombinator.com/item?id=35436168). I've been told they can make around $300k. I wouldn't be surprised to see the quant developers among the hardest hit by AI automation, maybe 10 quants replaced by one high-skill quant who can write the necessary prompts.
Yeah, but how many people who write professionally will have their career prospects significantly reduced. Many fewer writers for press releases, real estate, general PR to reporter stuff. Just like there aren't many travel agents. That will be millions lost.
Eh we've heard that song before but it rarely turns out to be true. Rust belt cities didn't transform with new kinds of jobs and industries. They just crumbled and fell apart.
Absent coordinated action by government to reorient economies as these new technologies take over, I fully expect wealth inequality to only get worse and for more and more white collar folks to be left behind. That's just wealth concentrating capitalism in action, which is inevitable when left to its own devices.
They are saying new jobs will show up. Not that they will be available for those displaced or even in America. Just that jobs will show up. "Jobs" is often callous or disingenuous because when they say it, they know it's not how people generally use it.
This is a common fiction trope that’s never been demonstrated in reality as far as I know. Everyone uses the bank tellers example and ignores multiple reports showing the leading cause of de industrialisation in the US was automation, followed by “free trade”.
It’s all nonsense unless you show evidence, and the evidence points to the contrary. What we have is a move of society towards “service” jobs, but service jobs exist in two different spheres - high wage (IT) and low wage (Starbucks). Having lived in the UK under a so called post industrial society, de industrialisation did exactly what people like Vaclav Smil say it does - it creates a society with no middle class.
> Rust belt cities didn't transform with new kinds of jobs and industries. They just crumbled and fell apart.
The interesting part is that the artistic jobs are mainly in the tier 1 cities e.g New York, LA, London etc. I'm sure there's historical precedent for tier 1 cities having their industries undermined but it'll be interesting to see what happens considering these areas tend to suck up the most government spending (or London does at least in the UK).
Personally, I think the current generation of kids have got their hearts set on being content creators and they're going to be in for a rough ride. The next generation are probably going to grow up amidst the devastation and turn to high demand, high skill blue collar jobs in droves e.g construction, plumbing, electricians, engineers, nursing etc and with a bit of luck we might actually end up fixing some of the gargantuan problems we've got as a society such as the housing crisis. Kids with construction skills probably aren't going to be daunted buying dilapidated houses and making them liveable again or developing new ones.
If I was faced with the prospect of losing my job as a software engineer and instead becoming a prompt "engineer", I would simply opt to drop out of the labor market entirely, or at least out of tech. At that point, there are no more interesting problems worth spending your time on. Just become a yak farmer or something and leave
Writing. I hope that it will be the last thing we automate. As smart as we can make an LLM, can we get it to write original material as good as Ted Chiang? Feels like this would be an AGI-level problem.
What I'm definitely not interested in is telling a computer to make images for me.
Like I said at the time I trained to be a monk transcribing books. I'm not interested in becoming a printing press operator.
> I would simply opt to drop out of the labor market entirely, or at least out of tech.
This is the problem for automation at the limit. Unless we generate an immensely powerful AI that sustains itself entirely then AI will destroy the economy and therefore itself.
The one good thing I see out of the coming AI cambrian explosion is that it'll hit the bracket of top 10% to 1% earners the hardest, while those above and bellow are safer.
This bracket has been in recent decades the enforcers of the status quo, helping widen the wealth gap all over the developed world. Maybe once they get a taste of what everyone is going through they might change sides and actually align with the general population on these issues.
> The one good thing I see out of the coming AI cambrian explosion is that it'll hit the bracket of top 10% to 1% earners the hardest, while those above and bellow are safer.
What are you basing this on? Perhaps I'm missing your exact meaning of "AI cambrian explosion" but right now I think it's arguable that artists and writers are the most threatened by this wave of generative AI - neither of which are usually considered top paying positions.
I'm mostly trying to extrapolate from current code and text generation abilities, so I know any prediction has a high chance of striking out.
Legal professionals that work on boilerplate could see simple contract generation being taken away from them for instance, and even run-of-the-mill developers could see themselves competing against much cheaper app/website generators.
Even if you can beat the AI, you might be forced to make your rates more competitive since the new baseline solution anyone can get for $10 would be a lot better than before.
> Legal professionals that work on boilerplate could see simple contract generation being taken away from them for instance, and even run-of-the-mill developers could see themselves competing against much cheaper app/website generators.
I don’t think people will use AI in its current (hallucinating) state if there is money or lives on the line.
61 comments
[ 3.1 ms ] story [ 128 ms ] threadA lot of job are at risk. Most of them no because of full automation, but because less people will be required because of productivity increases "thanks to" AI. And many more job will become boring.
The problem is this productivity gain will hurt a lot of people, and there are probably/possibly only a small minority of people that will fully benefit from it
The human with the art history background at least has some credentials, whereby the output can be trusted, to some degree and explained if necessary.
This is, imo, such an obviously near term capability that its not worth bringing up as a limitation.
That model doesn't exist.
We have models that can read text and then probabilistically create a summary that's the most likely to match the original.
But none of these models "extract meaning" in a way that can be used for fact checking purposes in the way you describe.
What you're describing is basically a world model, and while there's a few folks claiming these LLMs are showing the glimmers of building world models, it's mostly speculation.
I asked 3.5 to do it for you.
> The paragraph that explains that models cannot be used for fact checking is:
> "But none of these models "extract meaning" in a way that can be used for fact checking purposes in the way you describe."
It’s not hard to imagine a very simple tool to try and find citations for claims. Or perhaps, derive claims from a source with citations.
If i know for sure it is 100% GPT-X generated bs, do you think I'll read it?
I acknowledged in my first post that a particular quality standard had already been met by AI. I also concede that this quality standard could be sufficient for many businesses. God help us all if it's 90% of businesses.
Even for stuff like movie making, a director or producer would hire an illustrator or team of illustrators and send off a script to get it storyboarded and probably have to wait at least a few weeks. Now they can just go to a prompt engineer and have it done in a single afternoon. There’ll be similar developments in VFX, if you have a look at the latest Unreal Engine 5.2 demo they do motion capture from a woman’s face from a mobile phone camera in a matter of minutes. The actress in the demo says that kind of thing used to take a month to get the same result. When you’ve got a bunch of competing streaming services and studios trying to churn out as much content as possible to keep subscribers, it’s going to be difficult for them to justify the slowness and cost of keeping those human workers compared to the AI.
I believe GPT-X will generate tons of ... something ...
How you would force people to READ IT ?
If you know your incoming info is 90% BS, not a guess, not a thought, just knowledge that this is all auto-generated by some multiCPU entity - why would you read it? What are going to gain or to learn?
Of course it does not mean they will do anything about it.
I would not at all be surprised if these large firms hired consultants from places like McKinsey to come by and analyze the work their IT staff is involved in to see how tools like GPT-4 could be leveraged to "streamline" (aka reduce the headcount of) these orgs, and then dramatically restructured them based on their findings. The only thing keeping many of these folks around is the political clout of their managers and department heads.
People love saying that, but why wouldn't an exec take a multi-$million bonus to cut that dead weight? Or Private equity buy it out to gut it and resell it?
>The good news is that worker displacement from automation has historically been offset by creation of new jobs, and the emergence of new occupations following technological innovations accounts for the vast majority of long-run employment growth.
This article is a misrepresentation of Goldman's analysis.
Because what I’ve seen and read does not confirm this, in fact it disproves it. I can find multiple reports showing de industrialisation of the US economy happened mostly due to automation.
The point being that this happens all the time and is a natural consequence of progress and evolution.
Absent coordinated action by government to reorient economies as these new technologies take over, I fully expect wealth inequality to only get worse and for more and more white collar folks to be left behind. That's just wealth concentrating capitalism in action, which is inevitable when left to its own devices.
It’s all nonsense unless you show evidence, and the evidence points to the contrary. What we have is a move of society towards “service” jobs, but service jobs exist in two different spheres - high wage (IT) and low wage (Starbucks). Having lived in the UK under a so called post industrial society, de industrialisation did exactly what people like Vaclav Smil say it does - it creates a society with no middle class.
The interesting part is that the artistic jobs are mainly in the tier 1 cities e.g New York, LA, London etc. I'm sure there's historical precedent for tier 1 cities having their industries undermined but it'll be interesting to see what happens considering these areas tend to suck up the most government spending (or London does at least in the UK).
Personally, I think the current generation of kids have got their hearts set on being content creators and they're going to be in for a rough ride. The next generation are probably going to grow up amidst the devastation and turn to high demand, high skill blue collar jobs in droves e.g construction, plumbing, electricians, engineers, nursing etc and with a bit of luck we might actually end up fixing some of the gargantuan problems we've got as a society such as the housing crisis. Kids with construction skills probably aren't going to be daunted buying dilapidated houses and making them liveable again or developing new ones.
Same reaction, to the incredulity of the HN crowd. “Why would you quit” etc
What I'm definitely not interested in is telling a computer to make images for me.
Like I said at the time I trained to be a monk transcribing books. I'm not interested in becoming a printing press operator.
This is the problem for automation at the limit. Unless we generate an immensely powerful AI that sustains itself entirely then AI will destroy the economy and therefore itself.
This bracket has been in recent decades the enforcers of the status quo, helping widen the wealth gap all over the developed world. Maybe once they get a taste of what everyone is going through they might change sides and actually align with the general population on these issues.
What are you basing this on? Perhaps I'm missing your exact meaning of "AI cambrian explosion" but right now I think it's arguable that artists and writers are the most threatened by this wave of generative AI - neither of which are usually considered top paying positions.
Legal professionals that work on boilerplate could see simple contract generation being taken away from them for instance, and even run-of-the-mill developers could see themselves competing against much cheaper app/website generators.
Even if you can beat the AI, you might be forced to make your rates more competitive since the new baseline solution anyone can get for $10 would be a lot better than before.
I don’t think people will use AI in its current (hallucinating) state if there is money or lives on the line.