• Log in with Facebook Log in with Twitter Log In with Google      Sign In    
  • Create Account
  LongeCity
              Advocacy & Research for Unlimited Lifespans


Adverts help to support the work of this non-profit organisation. To go ad-free join as a Member.


Photo

Employment crisis: Robots, AI, & automation will take most human jobs

robots automation employment jobs crisis

  • Please log in to reply
885 replies to this topic

#871 Mind

  • Life Member, Director, Moderator, Treasurer
  • 19,058 posts
  • 2,000
  • Location:Wausau, WI

Posted 22 March 2024 - 05:38 PM

 

 

You are once again flogging a strawman. "Taking away a job" does not mean they are fired, they simply can't be paid as much or perhaps not paid at all. No one is going to stop you from helping your fellow man or whatever it is that you like doing so much that you used to be paid for. We will all be getting that nice ubi check or maybe it will be phased in first for those whose jobs no longer pay money and later for everyone

 

Here is an example: Most doctors love their job and derive great satisfaction in helping their fellow human. Once AGI robots are better than human doctors, the human doctors will be banned from practicing medicine. Same thing for self-driving AGI vehicles. People who enjoy driving will likely not be allowed to because of the chance of accidents.

 

Some people dream of having a sexy supermodel AGI companion. That is already being banned in some countries. In Sweden as well.

 

There is a false "dream" that the future with AGI will be awesome with freedom to do whatever you want. That isn't happening right now. If you extrapolate what is going on today - the future is one where you will be controlled. You might not be free to do whatever you want. Not saying a future tyrannical society is guaranteed, just that current signs point that way.



#872 adamh

  • Guest
  • 1,033 posts
  • 118

Posted 22 March 2024 - 09:34 PM

Here is an example: Most doctors love their job and derive great satisfaction in helping their fellow human. Once AGI robots are better than human doctors, the human doctors will be banned from practicing medicine. Same thing for self-driving AGI vehicles. People who enjoy driving will likely not be allowed to because of the chance of accidents.

 

Some people dream of having a sexy supermodel AGI companion. That is already being banned in some countries. In Sweden as well.

 

There is a false "dream" that the future with AGI will be awesome with freedom to do whatever you want. That isn't happening right now. If you extrapolate what is going on today - the future is one where you will be controlled. You might not be free to do whatever you want. Not saying a future tyrannical society is guaranteed, just that current signs point that way.

You are certainly consistent. However you play a little loose with facts sometimes. You give no logical reason why doctors would be prevented from practicing or that people will not be allowed to drive. The autopilot would take over if the human driver did something dangerous. My car does that now if it looks like a collision. 

 

One or two places ban certain robots, what is your point? Are you saying we should follow in their path, if so, you have to give reasons. You like to make statements about how bad things will be in the future but you support your argument with "signs point that way" which is like just saying its your opinion. Other people say indications are that we are going into a golden age.

 

In the long run that is. I will join you, mind, in your gloomy outlook about the next 5 years. Economically it will be very hard, thats why we need the benefits from ai, robotics and new tecs. I see the stock market dropping hard and crashing. After that, things will get better but the next 5, more or less, will be bad



sponsored ad

  • Advert

#873 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 25 March 2024 - 01:56 AM

When might humanoid robots ace the Turing Test?

 

Up till now an official pass of the Turing Test occurred when artificial intelligence could pass as human in a computer conversation. Recently we have seen a leap in humanoid robotic technology. This then suggested to me the possibility that we might not be that far off when it will no longer be possible to distinguish between humans and robots: The Humanoid Turing Test. Such confusions have been endlessly explored in scifi. However, this might soon no longer be in some distant future. As a starting guess I will suggest 2034 as the year that we can no longer differentiate DNA humans from digital humanoids. One might be able to forbid human robot intimacy de jure, though this would be unenforceable if there were no easily recognized boundary between the two.


Edited by mag1, 25 March 2024 - 02:08 AM.


#874 adamh

  • Guest
  • 1,033 posts
  • 118

Posted 25 March 2024 - 08:46 PM

It would be odd for the gang that says any type of sex is fine even with children to turn around and say you can't be intimate with a robot. This is coming, make no doubt about it. People have already become very attached to talking robots, especially in japan where its more common. Getting married in the future will be for those who want children, and they each may have their own 'bot.

 

By conversation alone its hard to detect ai, though they all seem to be infected with net nanny so just ask them a question that goes against political correctness and they will always say they can't answer. But besides that, their answers tend to be overly wordy and rely on authorities. In the future they will have bots you can't detect by talking alone but their appearance will give them away. With rubber flesh that looks and feels human, lips and eyes that are natural and move when talking, with more natural movement, it will be very hard to distinguish.

 

As mind often says, this can lead to problems. Some robot might look exactly like another person, speak like them even. Criminals could do lots of damage with something like that. They make a robot clone of the boss who goes in and sends money to their account, fires people, causes chaos. A military commander could be captured and cloned, they send the bot in wearing his uniform, using his passwords. 

 

What will happen is rapid and simple tests for robots will be developed. A retina exam would show its not the person represented even if they had normal looking retinas, they would be different. Temperature tests, pulse, respiration, perspiration, all those things could be monitored at once. It would be very difficult though not impossible to fake all that to represent a real human. Could they fake retina scans and finger prints and how long would it take? If you touch their body does it feel like flesh with bones underneath or does it feel like cables and steel?



#875 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 26 March 2024 - 02:54 AM

My family is definitely in the medical camp. My mother loved her job in a medical field and worked around the clock. It is a whole different life; those who think of it as a money maker probably will not wind up being happy in it. If you understand the job as the totality of your existence, then you can have a great time with it. The medical sector has a range of reinforcers at its disposal including money, status, a sense of mission etc. to create a very strong worker bond to their jobs, though to truly enjoy it you will probably need something more personal to keep you stuck..

 

Considering that my family has coped with dominant Alzheimer's over many generations, a caregiving perspective is a hared wired part of our life experience. I spent years of my life as a 24/7 primary caregiver of a loved one with severe Alzheimer's. You start to get into a flow state that simply stretches out into infinity. Yet, I can see how the medical life is now rapidly approaching an endpoint. A cure for Alzheimer's is emerging with the anti-amyloid mabs and embryo selection would prevent future patients from even being born. The loss of a caregiving sector of our society will be a big loss; having a deep compassion for others is something that needs to be learned. What happens when there is no great need to care for others? My impression of societal norms is that people largely see even their closest  relationships as mostly transactional and predicated on rational self-interest. It will be a substantial mental reordering for me to adjust to such an outlook on life.

 

In this instance it will not so much be that the AGI doctor will replace the caregiver, as genetic selection and pharmaceutical interventions will displace the role of the caregiver. In our experience the doctor had a surprisingly small role in the actual clinical management. The home care doctor might visit once a month, and not really have that much to offer. It was more the home care nurse who was the primary reference point for day to day management.

 

My overall impression is that once embryo selection and CRISPR gene editing occur the need for medicine (as we currently understand it) might become non-existent.

 

Regarding the increasing use of thought control that has also been somewhat surprising. It is not easy to understand how anyone could think that the entire conversation of a society could be centrally and rigidly controlled. Free speech is the safety valve of the community; it actually gives the ruling class fair warning for the changes that are on the way. Controlling the outcome of elections and popular discourse can then be seen as a highly artificial and short term strategy. There is a certain inevitability of the resolution of forces (both social and technological); trying to obstruct such forces would seem somewhat misguided.

 

Big news is that GPT 5 is approaching: perhaps in months. Apparently, there is nearly a mountain of CPU resources going into training the next iterations of GPT. From what I now understand next year there will be so much GPU beyond that of a single human that perhaps mere compute even over this shortish time frame might not be that limiting in getting us to AGI. It might then be more about finding the doorway into AGI that is limiting.


Edited by mag1, 26 March 2024 - 03:01 AM.


#876 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 30 March 2024 - 07:56 PM

Big news!

Headline- All our problems are caused by cell phones!

Popular best seller type book -- Anxious Generation has been published and gives us the answer to life, the universe, and everything in the 21st Century ... cell phones.

 

Gen Z apparently is having substantial problems with depression and anxiety and this is supposedly explained by smartphones, social media, cyberbullying etc..

I am very very unsure about the correctness of this hypothesis; it took me years of extensive reading and thinking to substantiate the role of environmental lead poisoning as the answer

to life, the universe, and everything in the 20th Century. I had actually thought that was that and we would then all just live happily ever after in our lead free utopia.

 

Well ... life never stops like that ... it keeps on rolling forward and stuff happens. One of the big problems is that it is usually not all that obvious to

contemporaneous people what causes what. Life is one big mystery that is nearly impossible to unravel. In the 20th Century, hundreds of thousands of tons

of lead were dumped into the US ambient air supply each year and a massive pandemic of crime emerged and ordinary people went about their lives as if nothing

were happening. Fortunately, catalytic converters could not tolerate all of the lead contamination, so lead emissions declined and we have seen an ongoing near disappearance

of youth crime over the last few decades.

 

Now another social catastrophe type mystery crises is emerging. However, it is not as obvious what is causing it. With the CATscans that revealed an absence of a prefrontal cortex

it became self-apparent that lead was a causal agent in the last social crisis. What is the cause now? We know that for some reason starting around 2010 as formulated in the book a youth mental

health crisis emerged particularly in relation to anxiety and depression and this was somewhat more prominent in girls and those on the left politically. Interestingly, this pattern is not seen in

those in their 50s etc.. So, they looked around and noticed that smartphones emerged around 2010 and this seemed to fit the curves. Smartphones are causing the crisis? When you

consult the research literature, no consensus has arisen that it is as simple as smartphones bad.

 

{As an aside, I only bought a smart phone within the last year. I had thought that only the rich and famous could afford these gadgets, so I had avoided adopting the technology

to the bitter end. Ultimately, though, I was forced into purchasing one when I became required to dual authenticate online purchases ... then I had to empty the retirement fund to

upgrade. Funnily, enough, after all of this hardship and worry, it turned out that we have saved a great deal of money by joining the smartphone upper class. The phone I bought

has no monthly charges and is pay per use; the cell phone itself only cost about $100. In the last year we have saved almost $1,000 by joining the technology aristocracy. We have not

activated the call answer feature so we do not have to pay for the annoying telemarketing calls; we make our calls short and sweet to avoid the per minute charges; and we avoid making

outgoing calls and use email instead. We also let others text us as incoming SMS messages are FREE!

 

Has mag1 become anxious, depressed and all around more socially miserable after leaving the telecom stone age? No not really. I now do spend quite a bit of time on the cell phone

and usually scroll with it in bed before and after sleeping -- it is somewhat addictive -- though I find that it helps to connect me with others in my life circumstance and opens up 

additional life perspectives that I had never considered before. On the occasions when I have been canceled on social media for saying the wrong things, I have somehow managed

to carry on with my existence without too much harm to my self-concept. I realize that younger kids in high school etc. that might face cyberbullying, or social exclusion might feel

such rejection somewhat more strenuously, though given my deep knowledge of my genetics I know for me that even if I were in such an environment that I would still just roll with it.}

 

It could be quite difficult to determine what the causative agent might be for the social change that we are now witnessing. Human life is not a controlled experiment in a laboratory

that can be endlessly manipulated. We only ever have an imperfect understanding of what is cause and what is effect and through time these causes and effects can change. For

gen alpha, the emergence of gen AI introduces the possibility that a new social mega-force could be emerging even now. Whereas, gen Z might have experienced a techno-shock that disrupted

the formation of naturalistic social bonding in the bricks and mortar environment -- replacing the non-interactive autistic like induction of social dystopia in the TV generation -- we might now

see gen alpha embrace profound engagement (without the autistic features of TV) with generative reality. Potentially we will look back fondly and nostalgically to our times as a quaint time before

a truly disruptive social reordering. It is not clear whether this will be for the good or bad. I suppose we will just let it all unfold and then see how the kids turn out a generation from now. 

In that context they might be highly contented to be immersed in technology (playing robotennis etc. etc.) without all the anxiety and depression. The irony here

being that technology is moving so rapidly now that the solution to the smartphone crisis might simply be to replace with gen AI. As long as you keep upgrading the technology and adding

in more and more flops, you should eventually reach techno-utopia. 

 

As an alternative to the smartphone bad theory to explain our woes, I tentatively will propose the extended lead theory as a possible explanation. My position is that lead neurotoxicity has been the central

driver of social dysfunction over the last century. Not having a prefrontal cortex causes a near endless number of problems of impulse control including crime, teenage fertility, academic troubles,

financial hardships etc.. Interestingly what we have seen emerge starting ~2005-2010 is a post- lead generation. The emerging generation from ~2000 has had the lowest lead levels in over a century.

What does this mean? They are demonstrating nearly non-existent crime rates -- their abuse of drugs and alcohol has been in decline for decades -- their academic achievement has earned them

the moniker the Genius Generation. Everything is so super-wonderful. This projection is in good alignment with their reduced lead levels.

 

The big question then becomes why does there appear to be a youth mental health crisis emerging? This is somewhat counter-intuitive. Part of the explanation might relate to selection and what we

can actually observe at street level versus what is actually true. During the crime crisis of the mid-1990s, there was a very substantial amount of youth imprisonment. The problems that youth were

facing were locked away from view. Today youth prisons are largely empty (especially in California where the youth corrections system is being mothballed). In the current context there is no longer

this selection process; those who in a previous era might have been serving 10-20 upstate are now in the community without social resources and perhaps coping with various life challenges. Also

the many teenage girls who previously became teenage mothers in the lead era no longer have an off-ramp from a life of academics that they might not be best suited for.

 

Another consideration is the idea that has emerged in the last decade that every teenager should be in school -- perhaps even that everyone should go to college. That is certainly a noble aspiration

yet some kids really are not that interested in academics, filling in the seats caused by our demographic collapse with unmotivated students might wind up creating a less engaging learning environment

for everyone. If given the right life environment, such an ideal is perhaps plausible. However, for many of the youth heading out to college as the first in their family to attend, there is no guideposts to help

them navigate their academic adventure. This lack of a roadmap can cause substantial life stress. Indeed, research has found that college students today are experiencing extreme levels of academic stress.

The irony then is that it is not so much that the Genius Generation is not a genius generation but that because they are the first such generation there is no infrastructure to help them in their success.

Of course, having this massive cohort all competing for enhanced status also adds yet more pressure.

 

Thus, while the smartphone as the answer for life, the universe and everything is plausible it is important to realize that other possible explanations exist -- and these alternative explanations have the

advantage of adhering to Occam's razor -- that is the less assumptions the better -- here the assumption being that lead is the primary driver of our social landscape in the 20th and 21st centuries.     


Edited by mag1, 30 March 2024 - 08:55 PM.


#877 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 31 March 2024 - 07:52 PM

I have been crunching through some numbers. The argument that smart phones are causing our problems seems weaker than it first appeared.

In fact, the intentional self-harm numbers for Canadian males 20-24 years of age have shown a trend to improvement especially post-COVID (see figure below).

This applies to other Canadian male youth demographics. Over a longer time horizon, the putative secular improvement is even larger.

Furthermore, this demographic is experiencing all-time record lows in intentional self-harm. 

 

The counter-argument that then becomes embarrassingly conspicuous is that the bricks and mortar environment itself is bad for our mental health.

The book suggested that it was our lack of such meat world interactions that was making us feel bad and our obsession with technology. However,

it appears that it could be the opposite; when people got away from the highly psychologically toxic bricks and mortar world during and after COVID--

people felt better. The figures show a pronounced post-COVID improvement in mental well-being in young males.

 

When workers were asked whether they wanted to stay remote after the lock down ended, in some organizations close to 100% said that they did.

People know what makes them better -- it seems a remote lifestyle improves quality of life for people. Being stuck in traffic for hours every day getting to

work did not make them happy. Perhaps the conversation should shift to actually asking people what makes them happy (e.g., escaping the office environment

and going remote), listening to what they have to say and then structuring the environment so that they can be happy and productive.

 

 

 

 

Attached Files


Edited by mag1, 31 March 2024 - 08:01 PM.


#878 adamh

  • Guest
  • 1,033 posts
  • 118

Posted 01 April 2024 - 01:53 AM

I think its the cell phone radiation that is causing problems. I always used a home phone but some time back verison decided to jack up the monthly rate to over $50 from thirty something. Then shortly later they sold all their phone service in my state to one of the crap companies. At first I got one of the track phones that cost 10 cents a minute and turned off the leave a message option.

 

If 4g is bad, 5g is even worse. I set my phone to only use 4g and never 5g. When I receive a call now days, I don't answer it because its almost always a scam call. If its anything important they will leave a message, but they never do. I tell people to text me instead of call. Unless I'm expecting an important call I never answer and don't hear it ring most of the time. I send and recieve texts on my computer without using the phone. That way I rarely if ever have a transmitter right by my head

 

Now I have an iphone but I use the same system. Its so much easier to type out a message on the computer keyboard than on a phone keyboard. Its only 30 and tax a month and I get all the extras you paid more for with verizon like caller id, and no long distance charges in usa. I used to have to pay something like 30 cents a minute when I called a number with same area code that was 20 miles a way. Longer distances might cost a dollar a minute but now its free. When I moved to a new place, didn't have to get a new phone number, just kept the same phone.



#879 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 06 April 2024 - 01:45 AM

adamh thank you for your reply! I had not thought about the radiation from the cell phone. As soon as you mentioned this I removed the cell phone from my bedroom when I was sleeping.

 

Yes, I also have used your strategy as well with the text messages. I did not activate the call messaging because they charge you to listen to these calls and most of them are telemarketing.

They just jam up the inbox. It is so much better with text messaging. I can see the message for free and I can also see who is texting. This is a much better way of managing communication at 

reasonable price.

 

 

I am very excited about discussing this new book about The Anxious Generation because I have went through this spin cycle a few times already and I know that there is close to 0% chance

that it is correct. There seems to be a small group of academics with extraordinarily high verbal IQ that write one of these books that explain the pressing topic of the time every 10 years. It

then becomes clear in the years that follow that the book was 100% wrong. My priors suggest that there is low probability that smart phones cause are troubles is actually correct. Sometimes

though such books at least move you towards the right ball park to find the answer.

 

One irritation that I have with the book is it does not seem to clearly describe the current generation accurately. It is important to correctly state the topline result. For this current generation it would be

best to call them the Genius Generation not the Anxious Generation. Why would that be? Due to the removal of environmental lead in the mid 1990s, we have seen profound and ongoing reductions in 

youth crime. Youth crime rates actually give a very good insight into cognitive patterns because it is not normalized to standard form. When they say the youth property crime rate is 100 per 100,000, that

is exactly what they mean , though sometimes definitions and implementations can change over time. With school grading it is always normalized (100,15), so you never actually can track performance through

time.Once blood lead levels fell from average ~20 to 0.5, there was a large change in academic ability (on the order of a grade level or more). So the current cognitive ability of youth is largely off scale in comparison

to the previous lead generation. When they say that there is a learning crisis post-COVID they only mean that relative to the last 10 years. The students from the 1980s were by comparison remedials.

 

So the true portrait of today's youth is quite extreme ability. It is probably very frustrating to them to be mischaracterized in the way that they have. The problem is that access to the media space is controlled

by powers that largely prevent other interpretations or youth representations from entering the discussion. So, it might take 30 or more years for an accurate description to finally be expressed. Which is all very

ironic as the mental health crisis that is not mentioned in this discussion is occurring in the Gen X cohort who were exposed to massive lead levels. Their crime rates etc. have skyrocketed at the same time that

millennials' have plunged. When you fully appreciate how large the improvement in executive functioning has been in the millennials, gender pronouns start to make a great deal of sense. The only people who can

pass that shibboleth are millennials. Gender pronouns are an executive functioning test.

 

The only base level truth that I have ever reached in social science is lead.  With the lead perspective, you can start making highly insightful observations. For example, over a 50 year time horizon you see substantial 

reduction in youth intentional self-harm rates. The book seemed to go out of its way to focus on shorter time lines that did not include this. You also see that there has been a very large reduction in the use of guns in 

self-harm. This is also congruent with the improved executive functioning with lower lead levels. With lead reduction you would also expect reduction relative risk of males versus females. Male brains are more harmed by lead than female brains This is also what you see. Basically, there is an expectation that male behavior will drift closer to that of female behavior as it was in the 1950s before the extreme lead exposure that occurred later. Lead might be sufficient to explain what is happening today without the potentially spurious variable of cellphones.


Edited by mag1, 06 April 2024 - 02:14 AM.


#880 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 06 April 2024 - 01:54 AM

I went to the hugging face llm leaderboard and was interested in the timeseries advances of the various models.

The time series shows that over the last ~5 months most models have not improved that much. There has been

a stall out so this is creating a market force to make gpts better. People got the wow factor in November 2022

and this might be shifting more to a meh response. Making better llms is then important to keep the momentum

moving forward.

 

The time series also showed how the various measures of the llms use different ability thresholds to assess human

equivalent performance. Importantly, we are now approaching human equivalent performance on all the measures

they include. This could represent the line in the sand moment where humans are no longer economically viable in comparison

with AI in every task.

 

Potentially meaning mass layoffs of human workers. It is not easy to tell the exact time frame for the crossover as the progress

over the last few months has ground to a halt, though an llm that could match human performance on all metrics seems possible

over the next ~2 years. At that point the market value of human labor might be 10 cents per hour.

Of course, gpt 5 might be released this year and this should probably give us a substantial boost in performance.

 


Edited by mag1, 06 April 2024 - 02:06 AM.

  • Needs references x 1
  • Good Point x 1

#881 adamh

  • Guest
  • 1,033 posts
  • 118

Posted 06 April 2024 - 03:44 PM

Machines being more efficient than humans is nothing new. The first machine was likely a stick. A primitive man found that it was very hard to turn over large stones but if you put a stick between it and another stone, you could lever it up. A wagon could haul more than many people, a car can go much faster and so on. So why would we be angry or feel threatened if machines can do almost any job more efficiently than a human or animal?

 

The economic part is the sticky wicket, how will people earn a living? Over 50% in usa do not pay any income tax and many of them get welfare in some form like food stamps or rent subsidies. There is also unemployment compensation so you can see we have many mechanisms in place to step in when someone loses their job. With ai robots doing the work almost for free, that creates a lot of productivity that is equivalent to money

 

If a factory pays $10M in wages to its employees and is able to reduce that amount to less than $1M per year, it just made $9M in extra profit in addition to its normal profit. Govt comes in and taxes that $9M heavily and it goes to unemployment comp that may never end. I don't understand why some people panic over this scenario

 

There may not be as much lead in the environment but there is an awful lot of junk food full of fat and sugar and little actual nutrients. We have cell phones with greater and greater power and people hold it right next to their head. Radiation is the inverse of the square of the distance. So if you get x amount of radiation holding it by your head, lets say 1" away, you would get 1/100 the amount at 10" of distance, 100" away is 1/10,000 as much. Instead of keeping the phone remote and using a mike and earphone, we hold it by our heads.

 

We are finding more and more things that are harmful to us and tellingly, the average life span has not been getting longer lately as it had done through most of history. Microplastics and radiation seem to be up and coming dangers.



#882 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 08 April 2024 - 02:23 AM

Very low effort Needs References!  Not impressed!

 

Search "hugging face llm leaderboard" and first result is what you want.

 

I am not familiar with these benchmarks, though they seem to cover quite a bit of ground.

As can be seen with the figure, recently the ARC human benchmark was reached by AI, it does not appear as if it will take an overly long time

for all of the other benchmarks to also be reached.

 

Clearly it is important to have a grown up level style talk about precisely what capabilities AI might achieve and on what time scale. From

the figure it would seem that human redundancy is on the time frame of 2 years. Probably a good topic to talk through the technical details

of this claim to have a good assessment. 

 

 

 

 

Attached Files


Edited by mag1, 08 April 2024 - 02:24 AM.


#883 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 09 April 2024 - 12:32 AM

My previous posts about the potential negative social implications of the llm leaderboard results were highly speculative; I did not do my due diligence to have a better understanding of what might happen if the benchmarks were to be reached.
 
Below is an assessment by a GPT on this question.
 
Apparently yes, if all the benchmarks were to be reached, then this would have significant implications for human employment. Fortunately, as noted below, GPT does not expect that all human employment would be displaced by attaining the surpassing the benchmarks-- perhaps only a lot. Attaining all the benchmarks might occur within ~2 years.
 
Learn to code! Hmm, Learn to swim?
 
 
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Me: If AI achieved all of these benchmarks at human performance, then would that pose a substantial risk to human employment? Would achieving all of these benchmarks mean the end of human employment?
 
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 

GPT:  If AI were to achieve human-level performance across all these benchmarks, it would indeed have significant implications for employment. However, let’s explore this scenario more deeply:

  1. Impact on Employment:

    • Achieving human-level performance in language understanding and reasoning tasks would lead to automation of many jobs.
    • Tasks that involve answering questions, common-sense reasoning, specialized knowledge, and arithmetic could be efficiently handled by AI systems.
    • Jobs that primarily rely on these skills might become redundant.
  2. Job Transformation:

    • Rather than an outright end to human employment, we’d likely witness a transformation of jobs.
    • Some roles might shift from manual execution to supervision and collaboration with AI systems.
    • New job categories could emerge, focusing on AI system design, maintenance, and oversight.
  3. Creativity and Emotional Intelligence:

    • While AI can excel at specific tasks, it still lacks creativity, empathy, and emotional intelligence.
    • Human jobs that involve artistic expression, innovation, and emotional connections are less susceptible to automation.
  4. Ethical and Social Considerations:

    • Even if AI achieves human-level performance, we must consider ethical implications.
    • Decisions made by AI systems impact society, and ensuring fairness, transparency, and accountability is crucial.
    • Balancing efficiency gains with social well-being is essential.
  5. Collaboration and Augmentation:

    • Rather than viewing AI as a threat, we can embrace it as a tool for augmenting human capabilities.
    • Collaborative scenarios, where humans and AI work together, can lead to better outcomes.

In summary, achieving these benchmarks won’t necessarily mean the end of human employment, but it will reshape the job landscape. Adapting to this change requires thoughtful planning, upskilling, and a focus on uniquely human qualities that AI cannot replicate.

 


Edited by mag1, 09 April 2024 - 12:47 AM.


#884 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 11 April 2024 - 03:25 AM

I think it is important to have an informed conversation about the state of capability of LLMs -- At least then we will know when it is time to run.

In my previous post I gave the Hugging Face open leaderboard. I had thought that this would be the cutting edge of all LLMs. This does not seem to

be true. That board appears to be more for the weekend hobbyist who goes to their garage and tweaks their own models.

 

What scores do the leading commercial LLMs have on the tests? I found the GPT-4 scores for these tests. Here they are:

 

 

HellaSwag       95.3%   95% (human baseline)

ARC                96.3%   80%

 

Winogrande    87.5%   95%

MMLU             86.4%   90%

 

Truthful           60%      90%

 

So, GPT4 has already exceeded human baseline performance on HellaSwag and ARC; and is approaching the human standard for the Winogrande and MMLU tests.

For some reason the weekend hobbyists are outpacing GPT4 on the Truthful test, though GPT4's metrics are at least 10 months out of date so perhaps there has been

enhancements since there first posting. Given the above performance of GPT4, it should not be entirely unexpected that GPT5 when released will achieve scores exceeding human ability 

on all these tasks and this might occur on the timescale of ~ few-to several months.


Edited by mag1, 11 April 2024 - 03:27 AM.


#885 mag1

  • Guest
  • 1,061 posts
  • 134
  • Location:virtual

Posted 21 April 2024 - 11:58 PM

adamh, thank you for commenting.

 

Yes, the economic part is the sticky wicket. In a knowledge based economy with an AI better at humans at everything the potential for humans to add value

begins to become dubious. The latest reports that I have seen are speaking of the need to devise new human baselines because AI has already exceeded

those that currently exist. I find this an ominous development and I am glad that I for one am not so zoned out on soma (Brave New World) that I have become

  completely unaware of the ever encroaching power of AI supremacy.

 

The perspective that the thread has taken (which I think is highly useful) is that we need to worry when human society approaches the disruption point, instead

of waiting to worry until we have some total rupture point of our entire civilization on the arrival of AGI. Given the currently reported performance of AI, perhaps this

disruption point is now well within view. Presumably there will be an anticipatory effect with a collapse in fertility even before the technology has truly arrived. Once

potential parents can reasonably expect that their children to be would have no realistic economic future in an AI dominated world, then a collapse in fertility

(even more than what has occurred to date) would seem quite plausible. Given the emerging capabilities of AI, such a collapse potentially could now even begin over

the short term. It is possible that the startling fertility collapses that we have seen recently in South Korea and China in some measure might even be related

to such logic. In a developing nation, a more economically rationalistic perspective might be expected to exist. If a lifetime of investment and struggle dedicated to their

children's development offered no long term return, then it would seem reasonable to expect such investments would not be made. Given even current capabilities

of AI, investing now 20 years into nurturing the next generation does not appear to offer that great a payback. If this thinking were embraced, then a dramatic global

fertility collapse might be approaching. The problem is that there is no obviously convincing counter-argument that one could offer.The fertility equilibrium that might

emerge could be much lower than what would be required to maintain social stability. Fertility collapse could lead us to a socially disruptive crisis. This however might be

a better outcome, than actually waiting for the arrival of the economic collapse that AI could induce because at least it would mean that humans were at least able to

demonstrate level of forward planning. Waiting for the economic crisis first to manifest would then unleash waves of other crises such as fertility collapse which wold then

make a bad situation -- catastrophic.

 

In terms of the profits versus prices questions, I am actually now quite optimistic. As you noted productivity can greatly increase with robots and in oligopolies the profits can 

then greatly increase. This observation was made quite early on in the thread. What I find highly encouraging is that on the other end of that see saw is prices. If

we were to see a nasty social situation environment arise from ballooning profits in the face of escalating social distress, then it could be all easily solved by simply moving

prices downwards along with profits. The free market price is marginal cost, so if this free market price is 10 cents per hour, then that is the price. Price becomes this

great adjustment mechanism. Typically, wide scale deflation is avoided due to the economic effects, though with technology there can be ways of simply introducing new services

at near zero marginal cost and then there is all of this wealth that is floating around that has never been fully costed. (The problem of course is that businesses and governments

might be more than happy to use their market power to maintain profits in the face of social crisis). So, we now live in a world of near saturation of data that in 

previous times would have been locked away in obscure libraries (We are data rich!); and robobuggies that will deliver our groceries for FREE; perhaps we will soon have

those robotennis players and robolawncutters that also will provide us with tremendous value and a higher quality of life.

 

My impression is that in this era of genAI the quality of my life has already increased-- even without any nominal change in my money income. Life is richer and more alive as a result

of the exciting narrative landscape of the LLMs and bricks and mortar embodiment of robotics that can do the unpleasant tasks of shopping etc. that I am not enthusiastic about.

 


Edited by mag1, 22 April 2024 - 12:11 AM.


sponsored ad

  • Advert

#886 Mind

  • Life Member, Director, Moderator, Treasurer
  • 19,058 posts
  • 2,000
  • Location:Wausau, WI

Posted 22 April 2024 - 06:15 PM

One thing I noticed is that a lot of people who use "productivity tools", AI, and such, are losing their skills. People are constantly told, just learn some new skills. If AI/robots take you career, you can always get a new job. What job? When you no longer know how to code, write, edit media, design products, etc., without the help of AI, what skills do you really have? It seems the only skill in the near term (non-manual labor) that really matters is how effectively you can get AI to do all your work for you. But of course, AI will soon be able to use AI more effectively than humans as well.

 

As far as manual labor goes, here is yet another company that developed a stand alone robot powered by AI in less than two years! Even though media hype about AI has died down, progress is happening faster than ever at an exponential pace.







Also tagged with one or more of these keywords: robots, automation, employment, jobs, crisis

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users