World Economic Forum Survey: Net Loss of 14 million jobs to AI and robotics by 2027.

[Uttamattamakin]

Summary

According to a survey of around 800 businesses by the World Economic Forum 69 million new jobs will be created, and 83 million destroyed worldwide.  A 2% decline in employment between now and 2027 due to the advent of AI and advanced robotics.  The WEF are the ones behind the famous Davos meeting, and the infamous "you will own nothing and love it" video from a fgew years back.   This is backed up by other studies by the likes of Goldman Sachs.

 

Quotes

Quote

Like every innovation, artificial intelligence (AI) fuels fears of job destruction. While the ChatGPT program is arousing the interest of the general public and awakening the appetite of investors, a World Economic Forum survey published on Sunday, April 30, estimated that 14 million jobs will be lost worldwide on balance over the next five years (69 million created and 83 million destroyed), i.e. 2% of total employment. It anticipates that nearly a quarter of employees will change jobs, mainly due to the advent of new technologies.

The authors of the study based their estimates on information collected from 803 companies worldwide employing 11.3 million people. The most significant job creation is in the education sector (+10%), and agriculture (especially machine operators), while job losses are mainly in administrative tasks, such as accounting services.

 

Quote

These estimates are fragile and uncertain, as the adoption of new technologies is so unpredictable. In its report, the World Economic Forum noted that "businesses introducing automation into their operations at a slower pace than previously anticipated" compared to 2020. The authors refered to a two-speed automation. "But while expectations of the displacement of physical and manual work by machines has decreased, reasoning, communicating, and coordinating – all traits with a comparative advantage for humans – are expected to be more automatable in the future," they wrote.

 

My thoughts

While it is true that AI and automation will create jobs they will not create jobs that the same people who loose the existing jobs can do.  They will also create fewer jobs than before.  In the past it has been true that advances in technology lead to more jobs created.  If you were employed at the buggy whip factory it was a lot easier to get retrained to work at the auto assembly line, than to become a computer scientist if one was a data entry worker before.   

 

Again a reminder this section of the post are just my thoughts but the idea below has been around for a while and I think it is the correct one (https://www.theguardian.com/technology/2023/feb/18/the-ai-industrial-revolution-puts-middle-class-workers-under-threat-this-time)

As this technology rolls out and we get more sophisticated AI, my feeling is we will see a hollowing out of the economy.  There will be people at the very top who run all the technology and make sure the back end infrastructure of it keep working.  Then there will be people at the bottom who do the most manual labor that cannot be automated.   I say at the bottom but some  of that labor that cannot be automated fully could be everything from traditionally made artisanal products to mining in difficult places, to surgery.   Of course the this hollowing out may be global with some countries having more of that "top" end and others having the un automateable manual labor. 

So what about the vast majority of people in the middle here?  This would mean a great many people would not be needed for labor, in a world where AI and robotics are making products in a better and more efficient manner.  How will we solve the core economic problem of deciding who gets what for those people?  

One option will be Universal Basic Income.  Governments will tax productive entities to distribute payments to all citizens of their country enough to cover basic needs.  Then if someone can work for more money they can build wealth on that base.    In the hollowed out labor market we could have a upper class of more technical workers who keep the automation working.  Then a middle class of skilled manual labor (to include even doctors giving hands on medical care), and a great many people who will be free to not work unless they want to.  

So something between this. 

 And this. 
 

Hopefully NOT this

https://www.facebook.com/watch/?v=10153920524981479 This is a link to the original infamous video from the WEF on Facebook.   Not easy to find.... for such a famous video. 

 

Sources

Le Monde "World Economic Forum estimates 14 million net job losses worldwide by 2027 " https://www.lemonde.fr/en/economy/article/2023/05/03/world-economic-forum-estimates-14-million-net-job-losses-worldwide-by-2027_6025227_19.html#:~:text=World Economy-,World Economic Forum estimates 14 million net job losses worldwide,over the next five years. 

[Monkey Dust]

4 years seems like a very tight timescale considering how slow many businesses are to invest, especially with higher interest rates. That, and for many, many roles A.I & robotics aren't ready for primetime yet. 10 years maybe, but not 4.

 

The way demographics are going, we need to replace some labour with automation, as the worker shortage is only set to get worse.

[Uttamattamakin]

37 minutes ago, Monkey Dust said:

4 years seems like a very tight timescale considering how slow many businesses are to invest, especially with higher interest rates. That, and for many, many roles A.I & robotics aren't ready for primetime yet. 10 years maybe, but not 4.

Depends on the business and their source of capital.  Small to medium sized operations  will be slower on the uptake.  Some of them may even try to sell the fact they employ real people as a selling point.   This will work to an extent.  Big corporate entities with access to equity markets, VC capital or they are listed and can sell stock they will be faster to change.  

 

37 minutes ago, Monkey Dust said:

The way demographics are going, we need to replace some labour with automation, as the worker shortage is only set to get worse.

This is also a concern.  Especially as we move to a world where people can live longer.  As average life expectancy increases we get more and more elderly people who will need goods and services.  Having a lot of money in an account is not the same as there being goods and services to buy with that money.   

 

What facscinates me is how this will replace various parts of even tech.  The "learn to code bro" saying of the 2010's won't apply when the AI can code.   You know.  Like when or if GPT-4 can write me the code for itself. 

# Import the necessary libraries
import torch
import transformers

# Define the model parameters
model_name = "gpt4-large"
vocab_size = 50000
n_embd = 2048
n_layer = 48
n_head = 16
n_ctx = 2048

# Create the model class
class GPT4Large(transformers.PreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.transformer = transformers.GPT2Model(config)
        self.lm_head = torch.nn.Linear(config.n_embd, config.vocab_size, bias=False)

    def forward(self, input_ids, attention_mask=None):
        # Get the hidden states from the transformer
        transformer_outputs = self.transformer(
            input_ids,
            attention_mask=attention_mask,
            return_dict=True
        )
        hidden_states = transformer_outputs.last_hidden_state

        # Get the logits from the lm head
        lm_logits = self.lm_head(hidden_states)

        # Return the outputs
        return transformers.GPT2LMHeadModelOutput(
            loss=None,
            logits=lm_logits,
            past_key_values=None,
            hidden_states=None,
            attentions=None
        )

# Load the model from a pretrained checkpoint
model = GPT4Large.from_pretrained(model_name)

# Generate some text using the model
input_text = "Hello, this is GPT-4 Large speaking."
input_ids = torch.tensor([transformers.GPT2Tokenizer.from_pretrained(model_name).encode(input_text)])
output_ids = model.generate(input_ids, max_length=50)
output_text = transformers.GPT2Tokenizer.from_pretrained(model_name).decode(output_ids[0])
print(output_text)

Just for fun I'll try a little to get this to actually run in Anaconda. 
 

[TetraSky]

Would love to see an AI do my job... of fixing up old ass machine in an industrial setting. I think I'm safe for a while... until they get bodies and fine motor controls. 

[Uttamattamakin]

27 minutes ago, TetraSky said:

Would love to see an AI do my job... of fixing up old ass machine in an industrial setting. I think I'm safe for a while... until they get bodies and fine motor controls. 

If the job market hollows out your skilled manual labor would be one of the last to go.  

 

On the code. 

To get this to work I need to install pythorch and so far Intalling this is taking way WAY too long to be worth it.  Like only in the last minute has it finally solved the environment and gotten to 6% solved.  WTH? 

[Brooksie359]

3 hours ago, Monkey Dust said:

4 years seems like a very tight timescale considering how slow many businesses are to invest, especially with higher interest rates. That, and for many, many roles A.I & robotics aren't ready for primetime yet. 10 years maybe, but not 4.

 

The way demographics are going, we need to replace some labour with automation, as the worker shortage is only set to get worse.

We don't have a worker shortage. It's more that people don't want to do a crap job that doesn't pay well. If you pay well enough you will find people to do that work. Personally I do think it would be an interesting idea if they did have a universal basic income that was funded by companies that use automation and AI to replace jobs. Basically you would have to pay a certain amount for each job your replaced with a robot or AI. Obviously this is just an idea and not really a well thought out plan but it does seem like we could do that if enough things get replaced by AI and robots. Also I think people don't realize that if enough people can't find jobs and don't have any money to spend that is bad for businesses too. I mean who is going to buy things if nobody but a select few have money. 

[freeagent]

I don't think our feelings or opinions matter. We are just a bunch of mouths to feed, livestock. If this does go through, they will just turn off our economy and start a new one. Our Masters have this all figured out.

[Taf the Ghost]

Multi-year projections about Birth/Death models for jobs... yeah, those models are worthless. Thus, this is simply some sort of fearmongering. I can guess the political angle these stories point to, but that's beside the point.

 

The Great AI Job Apocalypse is mostly just cover for tech companies to finally start actually eliminating the... 25%(?) dead weight they've been carrying. I remember looking at Google's Employee numbers a couple of years and their Revenue changes. Yeah, they've mostly gotten negative ROI on employees for quite while. Though that's a bit of a Google specific problem. (They have 2 products that make money, everything else loses money, so that makes sense.)

 

AI Tech brings disruption to a number of places, but like the Ubers of the world... it's going to be Net Nothing Very Useful. At least in economic terms. It also accelerates the reckoning that Computers talking to Computers aren't generally economic activity.  There's going to be some carnage (background art comes to mind) in some professional spaces, but, until someone can replace Medical Insurance Billing with it, it's not going to be a catastrophe for job numbers.  At the current state of the technology, you need just as many people to manage it as to previously do the tasks you'd want it to do.   This is looking a lot like a replay of 3D Printing, which actually lead to more manufacturing jobs because the Testing & R&D pipelines shrank and more designs could be tested & thus produced.

 

AI Tech matters, especially with regards to nearly instantons interventions into Human communication, but the jobs effect isn't close to what the doom prophets are saying. It won't be as bad as the early industrial period (where entire industries rose and fell in 2 years), but it will free up some resources that had to go to very mindless production. That will cost a lot of "make-work" jobs, but they'll show up somewhere else. (i.e. it'll be a great avenue for governments to expand.)

[wseaton]

I've worked enterprise IT since the mid 90s. Worked all industries: finance, Healthcare, manufacturing, etc.

 

Far as I'm concerned we have too many people sitting in cubes and too many software developers not creating shit. Meanwhile our economy is lacking skilled trades of all sorts.

 

I'm sick of the term A.I. All these things are is just large data set algorithms. They emulate crap and produce crap. Some are good enough to create a web site from a basic description. So what.

 

How many ChatGPT prompts does it take to build an affordable home, grow some food, rebuild an engine, or even change a diaper? A.I is software that makes lots of 1s and 0s in patterns we feel is important. If ChatGPT or its siblings puts you out of a job chances are your vocation was pretty thin to begin with.  Fast food places cant hire enough burger flippers, but software development and other data pushing jobs are predicted to slump badly. Tells you sonething. My cousin just got a masters in accounting and I'm laughing at him. He might as well got a degree in refurbishing typewriter ribbons. 

 

Computers manipulate 1s and 0s much better than people. So now we're amazed that, well, computers manipulate 1s and 0s better than us. 

 

I'm more curious what AI is going to do to a doctor visit, or replace a chemist. Until then I'm not impressed. We are just seeing the apex of the data pushing age.

.

 

[Brooksie359]

5 hours ago, Taf the Ghost said:

Multi-year projections about Birth/Death models for jobs... yeah, those models are worthless. Thus, this is simply some sort of fearmongering. I can guess the political angle these stories point to, but that's beside the point.

 

The Great AI Job Apocalypse is mostly just cover for tech companies to finally start actually eliminating the... 25%(?) dead weight they've been carrying. I remember looking at Google's Employee numbers a couple of years and their Revenue changes. Yeah, they've mostly gotten negative ROI on employees for quite while. Though that's a bit of a Google specific problem. (They have 2 products that make money, everything else loses money, so that makes sense.)

 

AI Tech brings disruption to a number of places, but like the Ubers of the world... it's going to be Net Nothing Very Useful. At least in economic terms. It also accelerates the reckoning that Computers talking to Computers aren't generally economic activity.  There's going to be some carnage (background art comes to mind) in some professional spaces, but, until someone can replace Medical Insurance Billing with it, it's not going to be a catastrophe for job numbers.  At the current state of the technology, you need just as many people to manage it as to previously do the tasks you'd want it to do.   This is looking a lot like a replay of 3D Printing, which actually lead to more manufacturing jobs because the Testing & R&D pipelines shrank and more designs could be tested & thus produced.

 

AI Tech matters, especially with regards to nearly instantons interventions into Human communication, but the jobs effect isn't close to what the doom prophets are saying. It won't be as bad as the early industrial period (where entire industries rose and fell in 2 years), but it will free up some resources that had to go to very mindless production. That will cost a lot of "make-work" jobs, but they'll show up somewhere else. (i.e. it'll be a great avenue for governments to expand.)

Honestly we have gotten way more efficient with computers which would lead you to believe that we would replace a bunch of jobs but it seems to me that we just do more stuff. There are always things to do it's just that unless there is a job shortage you won't incentivize people to do new things. I doubt we will ever truly have a job issue other than maybe during the transition period which was the case during most big transitions but it didn't cause the end of the world. I do know one thing and that is way more money could be spent on infrastructure like fixing roads and bridges and upgrading electrical grids and those things aren't going to be done by AI anytime soon. 

[Taf the Ghost]

18 minutes ago, Brooksie359 said:

Honestly we have gotten way more efficient with computers which would lead you to believe that we would replace a bunch of jobs but it seems to me that we just do more stuff. There are always things to do it's just that unless there is a job shortage you won't incentivize people to do new things. I doubt we will ever truly have a job issue other than maybe during the transition period which was the case during most big transitions but it didn't cause the end of the world. I do know one thing and that is way more money could be spent on infrastructure like fixing roads and bridges and upgrading electrical grids and those things aren't going to be done by AI anytime soon. 

I remember seeing reports of studies in the early 1970s, I believe, talking about productivity gains that were coming from new technology and how people would work less. (This was in the States, to note). The productivity gains happened and work hours & employment numbers went up.

 

I really need to dig out the reference, but I came across a report about an industrial production (textiles I think) that sprang up around 1904, employed a bunch of people and was closed by around 1908 because technology had moved that fast that the entire factory line was out dated.  That's what happens in actual "disruption" of industries. AI is simply more advanced automation and automation of tasks that are already semi-automated. It truly lacks the disruptive abilities they're preaching it'll have because AI has a deep Data Acquisition problem.  That said, they're going to insert it everywhere and blow up random industries because VPs love Buzzwords more than oxygen. My sympathies for those effected.

[CarlBar]

7 hours ago, Brooksie359 said:

We don't have a worker shortage. It's more that people don't want to do a crap job that doesn't pay well. If you pay well enough you will find people to do that work. Personally I do think it would be an interesting idea if they did have a universal basic income that was funded by companies that use automation and AI to replace jobs. Basically you would have to pay a certain amount for each job your replaced with a robot or AI. Obviously this is just an idea and not really a well thought out plan but it does seem like we could do that if enough things get replaced by AI and robots. Also I think people don't realize that if enough people can't find jobs and don't have any money to spend that is bad for businesses too. I mean who is going to buy things if nobody but a select few have money. 

 

The population of the developed world, and even recently China, has been shrinking on average for a while now. The number of employable bodies is in fact falling. That said lack of fulfilment in a given type of job is absolutely an issue IMO. We might all joke about not enjoying work, and it's true to a point, but no one sticks around in a job they actively dislike if they can possibly avoid it. And if they try to stick it out for too long they're likely to setoff mental health issues that will eventually force them out anyway.

 

4 hours ago, Taf the Ghost said:

Multi-year projections about Birth/Death models for jobs... yeah, those models are worthless. Thus, this is simply some sort of fearmongering. I can guess the political angle these stories point to, but that's beside the point.

 

The Great AI Job Apocalypse is mostly just cover for tech companies to finally start actually eliminating the... 25%(?) dead weight they've been carrying. I remember looking at Google's Employee numbers a couple of years and their Revenue changes. Yeah, they've mostly gotten negative ROI on employees for quite while. Though that's a bit of a Google specific problem. (They have 2 products that make money, everything else loses money, so that makes sense.)

 

AI Tech brings disruption to a number of places, but like the Ubers of the world... it's going to be Net Nothing Very Useful. At least in economic terms. It also accelerates the reckoning that Computers talking to Computers aren't generally economic activity.  There's going to be some carnage (background art comes to mind) in some professional spaces, but, until someone can replace Medical Insurance Billing with it, it's not going to be a catastrophe for job numbers.  At the current state of the technology, you need just as many people to manage it as to previously do the tasks you'd want it to do.   This is looking a lot like a replay of 3D Printing, which actually lead to more manufacturing jobs because the Testing & R&D pipelines shrank and more designs could be tested & thus produced.

 

AI Tech matters, especially with regards to nearly instantons interventions into Human communication, but the jobs effect isn't close to what the doom prophets are saying. It won't be as bad as the early industrial period (where entire industries rose and fell in 2 years), but it will free up some resources that had to go to very mindless production. That will cost a lot of "make-work" jobs, but they'll show up somewhere else. (i.e. it'll be a great avenue for governments to expand.)

 

Amen for the most part. I like when talking about large space based infrastructure stuff to link an online copy of a book. "Colonies in Space" it's a mostly still solid look at the technical aspects of building and maintaining the resources necessary to build a large scale object, (like a colony or a network of solar power satellites). But the original thinking that produced the underlying research the book is a layman's breakdown of was very flawed. They were looking for a solution to meeting the energy needs of a predicted over 10 billion global population by the year 2000. In reality societal changes produced a lower population whilst technological advances addressed many other potentiol issues.

 

But that doesn't mean there isn't going to be an issue eventually as we find ways to automate more and more things. All jobs are ultimately either in military/defence related stuff or in something that facilitates the production, transport and/or sale of good and services to other human beings. A population of a given size has a finite amount of stuff it can consume, and there's only so much you can do to expand that. Get sufficient automation and you will run out of jobs to employ people in. That said where a good ways off that IMO. In fact i'd argue the rise of social media/youtube influencers is ultimately an expression of one counteracting force in this. Namely that when it comes to arts and media people can consume from a very large number of different sources with each person consuming from a very small subset of that very large number of sources.

[Brooksie359]

1 hour ago, CarlBar said:

 

The population of the developed world, and even recently China, has been shrinking on average for a while now. The number of employable bodies is in fact falling. That said lack of fulfilment in a given type of job is absolutely an issue IMO. We might all joke about not enjoying work, and it's true to a point, but no one sticks around in a job they actively dislike if they can possibly avoid it. And if they try to stick it out for too long they're likely to setoff mental health issues that will eventually force them out anyway.

 

 

Amen for the most part. I like when talking about large space based infrastructure stuff to link an online copy of a book. "Colonies in Space" it's a mostly still solid look at the technical aspects of building and maintaining the resources necessary to build a large scale object, (like a colony or a network of solar power satellites). But the original thinking that produced the underlying research the book is a layman's breakdown of was very flawed. They were looking for a solution to meeting the energy needs of a predicted over 10 billion global population by the year 2000. In reality societal changes produced a lower population whilst technological advances addressed many other potentiol issues.

 

But that doesn't mean there isn't going to be an issue eventually as we find ways to automate more and more things. All jobs are ultimately either in military/defence related stuff or in something that facilitates the production, transport and/or sale of good and services to other human beings. A population of a given size has a finite amount of stuff it can consume, and there's only so much you can do to expand that. Get sufficient automation and you will run out of jobs to employ people in. That said where a good ways off that IMO. In fact i'd argue the rise of social media/youtube influencers is ultimately an expression of one counteracting force in this. Namely that when it comes to arts and media people can consume from a very large number of different sources with each person consuming from a very small subset of that very large number of sources.

The problem is that automating basically most production related tasks would still require you to have people you can sell the produced products too. I don't think it's really fair to say the reason why employment hasn't been an issue is due to population decreasing especially when I don't believe that is even the case for the US. Yes birth rates are down but you have to take into account immigration has largely made up for that for the US at least. Also the problem isn't necessarily having an unfun job but working an unfulfilling job and not having enough time outside of it to enjoy yourself. I know plenty of people who are happy working jobs they don't necessarily like because they make enough money to find fulfillment outside of work. Yes ideally we would all work jobs that we enjoy but unfortunately that is idealistic and not going happen always which is why you should try and find fulfillment where you can and not necessarily relying on your job to find it. 

[Shreyas1]

5 hours ago, wseaton said:

I'm more curious what AI is going to do to a doctor visit, or replace a chemist. Until then I'm not impressed. We are just seeing the apex of the data pushing age.

.

 

It could reduce the amount of time that a doctor spends charting, allowing them to see patients more. It could also be used by insurance companies to more efficiently deny coverage claims.
Legality wise though, I doubt that anyone would want an ai to have the same level of liability over a patient as a doctor. 

 

For a chemist or any other scientist it could suggest research topics after reading a lot of literature. 

[Doobeedoo]

15 hours ago, TetraSky said:

Would love to see an AI do my job... of fixing up old ass machine in an industrial setting. I think I'm safe for a while... until they get bodies and fine motor controls. 

Well... there were some brain operations with robots done already they sure have much more fine control.

[Doobeedoo]

There are already few jobs I could see them almost fully automatized. At least simple ones that would be IT based. Really even for a while I wondered why this and that like parts of one job are not already.

But you always need human monitoring, intervention anyway.

[SansVarnic]

Every year there is a new brand of doom and gloom. I am still waiting for the world's end from Y2k... 24 years later... nothing.

 

*edit: I will add that this same type of news has been around since the early to mid 80's with some hints back in the 50's. Job creation has been and always will be created by the public as demand for product or services grow. Nothing will change this.  This is just a bunch mongering and need to disrupt those that worry needlessly. There will always be jobs as long as economy are not disrupted. Automating jobs has been a thing for years now, since the 90's and yet there are still new jobs being made to for those displaced.

 

Ever watch the movie Willy Wonka? At the end the dads job was automated but womp womp he got a new job maintaining the automation. Ironic...

 

Leason from this: Dont let naysayers ruin your day with stupidity.

Edited May 7, 2023 by SansVarnic

[CarlBar]

4 hours ago, Brooksie359 said:

The problem is that automating basically most production related tasks would still require you to have people you can sell the produced products too. I don't think it's really fair to say the reason why employment hasn't been an issue is due to population decreasing especially when I don't believe that is even the case for the US. Yes birth rates are down but you have to take into account immigration has largely made up for that for the US at least. Also the problem isn't necessarily having an unfun job but working an unfulfilling job and not having enough time outside of it to enjoy yourself. I know plenty of people who are happy working jobs they don't necessarily like because they make enough money to find fulfillment outside of work. Yes ideally we would all work jobs that we enjoy but unfortunately that is idealistic and not going happen always which is why you should try and find fulfillment where you can and not necessarily relying on your job to find it. 

 

 

Parts of Europe, AFAIK Japan, and as i noted, recently China have seen actual population decline. (And this is ignoring the fact that many developed nations are seeing an aging population meaning the % of post retirement population is in general on the rise). The US isn't the entire world.

 

Also there's a huge difference generally between a job you dislike and one that is actively detrimental to your mental health. I'm touching more on the latter with those comments.

[Uttamattamakin]

18 minutes ago, SansVarnic said:

Every year there is a new brand of doom and gloom. I am still waiting the world's end from Y2k... 24 years later... nothing.

I wouldn't call this "doom and gloom" myself.    To me like Y2K this is change that we need to manage and adjust for.  Y2K was a nothing because we made sure to take action to be sure it was nothing.  

Gotta be on the lookout for that killer asteroid to have a prayer of diverting it. 

 

3 hours ago, Brooksie359 said:

The problem is that automating basically most production related tasks would still require you to have people you can sell the produced products too.

This is a big sticking point.  Automate away the jobs and don't provide something like a Universal Basic Income or ... another version of that a "Universal Dividend"  then you get a society where no one has the money to buy the products produced by automation.  

The flip side of that is the elderly will need services and goods to buy.  A shrinking population means an aging population.  So older people will need to buy goods and services.  Care for these elderly people (who are people my age and younger down to say early 20's now) will be one of the last areas to be automated.  

 

3 hours ago, Shreyas1 said:

It could reduce the amount of time that a doctor spends charting, allowing them to see patients more. It could also be used by insurance companies to more efficiently deny coverage claims.
Legality wise though, I doubt that anyone would want an ai to have the same level of liability over a patient as a doctor. 

What a "AI" like GPT can replace are people who  do medical billing and coding (Medical coding is translating what the doctor does into the codes that insurance companies use so the doctor can get paid).   IT can replace a lot of the office and clerical support staff involved.  This means far fewer people working in that area of healthcare.   A large part of many enterprises from medical to education, business... are just massive administrative and clerical offices that don't perform the function that produces the revenue for the business.  

This is the part of the workplace that will hollow out.  You'll see such support staff leaving before robots can replace the hands on workers. 

 

3 hours ago, Shreyas1 said:

 

For a chemist or any other scientist it could suggest research topics after reading a lot of literature. 

I've tried this with GPT-3 and GPT-4.  The problem with this is these models are not anything like an AGI.  As has been pointed out.  These models churn the garbage that has been published into new patterns of the same old garbage.  

 

If I asked GPT4 to come up with a novel theory of space-Time it could not.  It can refer me to String Theory or Loop Quantum gravity or other things published but if asked to be creative it won't really.  

 

So your theoretical physicist or chemist could be fine. The chemist or physicist who just works on small variations of already established work might be in trouble.  A big problem for the scientist who dare to think different is where will they work?  Replace the labs that just do work that is incrementally better.  Automate the research that could be thought of by prompting an AI, then carried out by a robotic system... where would the next Alexander Flemming work?  Who would leave the cheese sandwich to get moldy that would lead to finding the next  penicillin?   

Replace the patent clerks with GPT and where would the next Einstein work? 

 

See what I mean. 

 

A big problem I see is going to be how do we pick the winners in this hollowed out job market.   If you are a skilled or semi skilled laborer and everything between you and management is automated then there is no way to get up that ladder.  If you are in a creative part of STEM ... if there are no jobs to work while you coming up with the next insanely great idea ... then you wind up leaving STEM.  IF you are not instantly a super star then you have no job.  Most people are not going to be super stars.  (For example where I work it seems that a full time faculty member is just going to be replaced by hiring on more part time adjuncts.  I can see a lot of teaching basic undergraduate STEM being automated  and adjunctified.)

The answer may just wind up being we pick winners based on preconceptions of who will have good ideas.  If we pick wrong we miss out on great ideas. 
 

[SansVarnic]

1 minute ago, Uttamattamakin said:

I wouldn't call this "doom and gloom" myself.    To me like Y2K this is change that we need to manage and adjust for.  Y2K was a nothing because we made sure to take action to be sure it was nothing.  

Gotta be on the lookout for that killer asteroid to have a prayer of diverting it. 

Its called doom and gloom for a reason. This is lost on people today.

Y2K was more than just pc crashes... but regardless it was an example of how hyped things get, just like this article. That was my point.

 

The lesser point I was making was this topic is not really newsworthy for the same reason as above.

[Uttamattamakin]

24 minutes ago, CarlBar said:

Also there's a huge difference generally between a job you dislike and one that is actively detrimental to your mental health. I'm touching more on the latter with those comments.

Mental and physical health.  As I said to someone IRL.   Working my job is a lot like working at Chernobyl.  It's fine, the people are great, the students that show up and try hard make it worth while.  Problem is they are in a bad situation (needing to review all of 6th-12th grade math in 16 or 32 weeks).  No matter how great the people are Chernobyl is still Chernobyl.  

This brought to mind a job one of my cousins has/had.  He was a fireman in a midsize midwestern city for a little over 10 years.  Had he not ranked up and out of front line fire fighting he'd be forced to retire.  After 10-15 years or so they get retired with a full pension because the job is so rough

These sort of ... skilled service jobs will be in that tier that may never bee automated away.  How do you automate away fire fighting or police?    Someone may try.  

A funny set of videos you might want to check out.  This is from a 90's cartoon called Bots Master.  I think it rated well but the toys did not sell enough.  Who knows what unintended consequences of really good AI and robotics could lead to.  

 

6 minutes ago, SansVarnic said:

Its called doom and gloom for a reason. This is lost on people today.

Y2K was more than just pc crashes... but regardless it was an example of how hyped things get, just like this article. That was my point.

 

The lesser point I was making was this topic is not really newsworthy for the same reason as above.

I can get why you might think that but I see a lot of news talking about AI replacing a lot of jobs.   It's a discussion that has gone on for a long time, and is in a lot of news sources. 00

 

Like CNN. https://www.cnn.com/2023/04/30/business/automation-jobs-world-economic-forum/index.html 

 

I think we who are discussing this can be excused for thinking tech items published by CNN and in LeMonde are news.:) 

[Uttamattamakin]

1 hour ago, Doobeedoo said:

There are already few jobs I could see them almost fully automatized. At least simple ones that would be IT based. Really even for a while I wondered why this and that like parts of one job are not already.

But you always need human monitoring, intervention anyway.

This leads to a question.  Are the people who say ... proof read what the AI comes up with going to make as much as those who do the writing? 

 

For example,  This summer I may have a hand in writing a solutions manual for a calc based physics textbook.  The person who is going to do most of the writing is going to get 12% royalties.  I am going to proof read and spot check for errors for about 3%.  If everyone is a proof reader, then everyone is getting a pay cut... that is if they even keep their job at all. 

This kind of thing should lead anyone who has to prepare young people for the world to come to rethink what we are preparing them for.  The young ones I teach now are going to be fine for a while.  5-10 years from now... what will be the point of a degree that is less than an MS or MA?  It'll be like either get an advanced degree or get certified in a trade that is very fiddly and hard to automate.  Either of those things are difficult and a great many people just can't do that kind of work. 

[Doobeedoo]

12 minutes ago, Uttamattamakin said:

This leads to a question.  Are the people who say ... proof read what the AI comes up with going to make as much as those who do the writing? 

 

For example,  This summer I may have a hand in writing a solutions manual for a calc based physics textbook.  The person who is going to do most of the writing is going to get 12% royalties.  I am going to proof read and spot check for errors for about 3%.  If everyone is a proof reader, then everyone is getting a pay cut... that is if they even keep their job at all. 

This kind of thing should lead anyone who has to prepare young people for the world to come to rethink what we are preparing them for.  The young ones I teach now are going to be fine for a while.  5-10 years from now... what will be the point of a degree that is less than an MS or MA?  It'll be like either get an advanced degree or get certified in a trade that is very fiddly and hard to automate.  Either of those things are difficult and a great many people just can't do that kind of work. 

Right, though even if some jobs will be very automatized, person needs to be knowledgeable and specialized for that job anyway. It also depends how much work they have in the end too for that job to be paid well. Even decade+ back when I was schooling, things we were being thought each year changed literally how some stuff advances so you need to learn new stuff etc to be relevant, yet that still wasn't enough in it's own. It is what it is.

[Uttamattamakin]

38 minutes ago, Doobeedoo said:

Right, though even if some jobs will be very automatized, person needs to be knowledgeable and specialized for that job anyway. It also depends how much work they have in the end too for that job to be paid well.

 

A big systematic change is how work is valued will have to change.  Good employers will, IMHO, see the few humans they need as valuable and should pay them enough to live well on (over and above any possible basic income).  

Like these proof reading humans who just check the AI's output to be sure it makes sense will need to know how to write code themselves to do that.  They'll have to be highly educated and highly skilled.  Even if they don't literally type it all out the bit they do will be of high importance. 

38 minutes ago, Doobeedoo said:

Even decade+ back when I was schooling, things we were being thought each year changed literally how some stuff advances so you need to learn new stuff etc to be relevant, yet that still wasn't enough in it's own. It is what it is.

Yeah this is true.  It's just there are certain things we could always rely on being important.  Knowing how to read and write analytically was a thing we could bet safely on.  Needing to know basic to intermediate algebra we could bet on that.  Needing to have basic science skills we could rely on.  

Now I'll bet right now I could ask bing this. 

 

Please write me out the Kerr Newman metric for a Black hole. Then write me a short haiku about black holes. Then generate an image based on the math and the poem.

Quote

The Kerr-Newman metric is a mathematical formula that describes the spacetime around a rotating, charged black hole¹. It is given by:

 

Quote

A haiku about black holes is:

Black holes are so dark
They swallow light and matter
But they can still spin

 

Quote

I mean that's pretty darn good already. 

Edited May 7, 2023 by Uttamattamakin
I put the responses in separate quotes so that people are more likely to look at them. Some won't expand a single quote.

[Fasterthannothing]

Honestly I won't sugarcoat this but the people who will lose their jobs already aren't doing their jobs. Literally no one wants to work anymore. I know tons of business owners who have to hire and fire people in the same week because they don't even show up for work anymore.