This should make for interesting conversation at "Career Day" at my daughter's elementary school next week :)
The traditional format of questions: What do you? What was your career path to where you are today? Where did you go to school and what degrees did you get?
My path and current job don't feel relevant to these kids. It won't be their world. What do I say to them?
Everything is up in the air. Career day should be: "Go to chatGPT, go to Midjourney, go to RunwayML, go to Hugging Face, go to Replit, study all of them, and you can make money in months and be properly set up for the future
Any commitment to a 4y thing (or worse, decades) sounds like professional suicide to me right now
A lot of things will be up in the air, but will managers and entrepreneurs go away? I doubt it because humans still hold the power and will want humans held responsible for any decisions even if aided by AI.
I really do not have a good idea what to tell my children, as the value of education is getting lower every day, unless you get into an exclusive club (a.k.a. Ivy league) just to meet people. I don't have a million bucks (https://www.bloomberg.com/news/articles/2023-03-31/how-to-get-your-kid-into-an-ivy-league-college-parents-pay-750k-for-a-shot), so it's going to be the %99 for them. The best I can do it to help them understand what's going on, so they know who's responsible for this and maybe do something about it before it's too late (i.e. now).
Tomas, you provide new ways of seeing the world by revealing previously unrecognized patterns in events and in society. Once seen and understood, seeing these patterns is extremely valuable. This article does than in spades. Thank you so much for your careful research and very thoughtful analysis. I’ve come away from reading your article seeing the world in a new way, which will help me navigate the future.
I really have a problem when we talk about AI replacing Authors and Artists like its nothing. Likewho actually wants that? who does that serve? do we sacrifice creativity to become mindless consumers?
people churning out AI generated content are often doing so for a quick buck. Which I guess the end product is not much different from typical mid level commercial art. But they don't understand that engaging with the process really is the act of creation. I guess I can see a situation where real artists and writers could use tools to speed up their work (since there is a lot of tedious stuff involved) while maintaining their artistic integrity.
But the reason a lot of artists/creatives view AI with such distain is they understand that there's no sense of control to what the AI spits out. It's not their work. It has none of them in it.
But hey, maybe we have a future where those who have the fundament skills in art/writing can put out elevated content. It's not there yet.
I'm not hopeful though. "Good enough" is good enough for the vast majority of people and companies.
Do we want to read novels and listen to music created by AI? half the appeal is to engage with the human behind the work.
1. If in your paragraph you change art for agriculture and artist for farmer, you'll see that the functional/market part of different "arts" (eg illustration or videos for marketing) will be much better off
2. Don't worry about the "artistic" part of art. We like art because of its scarcity. The moment part of art is not scarce anymore, art moves on to something scarce (ie that only humans can do). Consider for example the fact that thousands of years of evolution towards more realistic painting suddenly disappeared after the invention of photography, and creativity suddenly exploded with things like surrealism, cubism, or impressionism.
With the advent of photography yes, we got impressionism, cubism, surrealism. We also got the loss of hundreds of years of the development of very refined artistic methods that only recently have become appreciated again by a narrow band of society and are in the process of being restored. We need to recognize that with change we risk losing things are of great value. We risk throwing the baby out with the bathwater.
I find your comment kind of scary. What was lost was centuries of the development of skills that are part of the reason that we go to museums to see work made during that time. The skills themselves are meaningful. To lose them in society is a very great loss. With all due respect, to make light of them and say because we have photography they are not important indicates to me a lack of understanding. Fortunately, they are in the process of reemerging. And, I’m not sure simply recording them is sufficient because the way they are transferred is from master to student.
I don’t think photography is more valuable than fine art. I think more people use photography than explore fine art. And, as we move down the road with this intense focus on becoming more and more productive if we lose the subjective—-if we lose love, truth and beauty—-I think it will become a very sad and empty world.
Machine-made shoes means there are far fewer cordwainers in the world, but I’d rather live in a world where there are plenty of shoes than one where shoes are expensive so many people have to go without shoes.
I think it doesn’t have to be either or. It can be both and. Both mass produced shoes and hand made artisan shoes for those with the money and the interest.
Tomas you haven't allowed for the upswing in live music that is happening. More and more bands, from the large to the small, the well-known to the extremely local, are touring, playing their music to live audiences. At least here in Aotearoa New Zealand any way!
Would love to see data on this! My current hypothesis is that if it's truly a trend that's happening, the upswing is a fraction of what it used to be. I'd love to be proven wrong!
It’s what I talked about about humans being forced in to bespoke niches. As the mass commercial item becomes cheaper, people will want more idiosyncratic bespoke experiences. We see this in beer too, where there were a ton of local brewers in the US (because nobody could ship beer far), then the beer market became dominated by a few big players and now there is a craft brewer revolution (and hipster types drink stuff like Pabst which use to be consider cheaper more inferior more blue-collar beer just because that’s the hipster thing to do).
I think this clearly leads to the conclusion that somewhere, not so far, in the future (or would it be 'somewhen'?) the balance job creation-job destruction will got persistently in the negative side.
For me it means, that we (the whole humanity) will need to switch to some 'Star Trek'-like 'economy' where means of survival (with adequate quality of life) and growth (intellectual, social, etc.) are not contingent on 'jobs/salary'.
Some 'Singularity' pundits take it as a given that exponential growth would bring solutions faster and faster, bringing a nearly painless transition to an 'abundance' heavenly scenario , but I think a more tortuous, and suffering-ladden for billions of people, path is presently at our door.
Until we have ubiquitous energy supply and energy to mass hardware (food synthesizers, etc.) like Kirk/Picard/et al.. I think we will need to accelerate some huge efforts on UBI (Universal Basic Income) implementation.
That is a socio-political nightmare as complex to unravel as climate change de-worsening initiatives (we can't prevent significant climate change anymore, so we hope that work can be done to make it less harmful), and this AI-boosted clock is also ticking fast...
PS.: there is also the 'crime' business to consider, where sadly many jobless people find their new "way-of-life"
The problem with a Star Trek-like economy is that while we're about to automate the hell out of anything having to do with information processing, we're still fundamentally limited by raw materials, i.e. what we can dig out of the ground. I totally agree that we need to fundamentally rethink the social contract because the intrinsic capital value of human labor is going to be driven down almost to zero, but the intrinsic value of natural resources will only increase. The people who own/control those natural resources will probably resist the redistribution of those resources in a UBI system, so . . . I'm not hopeful for the long-term future. Honestly, the only way the equation gets balanced if there are a hell of a lot fewer humans around trying to consume resources, so that will probably happen eventually, either gracefully or not.
It depends on what you call labor. Humans labor certainly will move to those areas that are tougher to automate. Originally, human labor was largely doing backbreaking stuff that really is better off done by machines. Soon, purely intellectual work will be better handled by AI. Though AI actually currently has trouble handling an idiosyncratic physical world that humans and other animals evolved to handle much easier (I’m thinking of driving in city streets in the rain). So plumbers will be around for a while.
Humans also hold the power. I’m pretty certain that humans will always want other humans responsible for decisions they deem important (so we won’t elect AI lawmakers, though lawmakers will be assisted by AI in the future), and there will be human managers, human entrepreneurs, etc., each with an AI army working for them.
That’s also why I think the general area of risk will still have plenty of human involvement (AI will help, but I don’t imagine humans becoming OK with purely AI determining how much capital banks should hold or which banks to rescue or let go under).
Humans will also be more employed in what I consider the luxury goods industry (which, thanks to a great increase in productivity and wealth, more people will enjoy). So everyone will have personal therapists to figure out the meaning of life but the poor will only have AI ones while the better off will also have human ones. There will also be jobs related to the physical world. I’m sure people will still want to travel, and that will be limited by the cost of transportation, but while maybe fewer people may be needed for the actual transportation part, and while I’m sure you can get AI tour guides, getting a human tour guide would be seen as a luxury good (not everyone will be able to afford the actual real human touch and insight).
I had imagined that AI designers would aim for thinking machines that REMOVED human bias and emotion to provide optimised advice and solutions. The current versions seem to be more about just learning from the version of "us" that is available to them via the internet with all its biases
I mean, “optimized” depends on your values and goals so always may be “biased” based on your point of view. As a crude example, an algorithm to optimize white power certainly would seem biased to some. An algorithm to optimize engagement may also lead to bias (because humans are biased about what they will engage with). And yes, ML trained in “biased” datasets can only be expected to be biased.
Thanks for all your helpful AI comments. I wrote a reply ages ago, but it raised more questions than answers so it felt sensible to just think about things for a bit. I suppose my concept of optimised could best be defined by the principles that Tomas uses in Uncharted Territories and the logical conclusions that those principles lead to. Whether people have the wisdom to choose those logical solutions is another matter. As you say above, humanity still holds the power (for now), but the question is: which specific humans get to make the decisions? History suggests that those with power and wealth will not give it up lightly and that they tend to make decisions based on their own short-term interests rather than for the good of humanity (whatever that means!). If we could harness the Wisdom of Crowds we might have a better chance of making good decisions.
Part of the reason for writing this now is that I recently posted a couple of things about the importance of what and how we learn, both as individuals and as groups. This inevitably brought me back to the AI questions. How will the precocious child that is AI learn and who will be the teachers? It is true that a subservient AI might happily advise on optimising for white power, but if we are really talking about Artificial Intelligence that is smarter than humans then a reasonable starting comparison would be with the most intelligent humans we have now.
An extremely intelligent human would give a balanced view when asked about optimising for white power. She might first ask “what do you mean by white power?”. He might dispassionately point out the difficulties in deciding who is white and who is not and the downsides of the conflict that would ensue and cause harm to “white” people as well as others. They might point out that the optimum solution long term for the wellbeing of white people is to co-operate with all colours of humanity to increase the power of humanity as a whole, so the white slice of the pie would be bigger. An intelligent and wise human would take into account the context of the question
What we have now is machine learning which is just a new and powerful tool. But what are the machines being taught and by whom? An algorithmic tool can just magnify the bias and lack of understanding of the users and might just produce “Artificial Prejudice” or “Artificial Stupidity”. The question is whether anything that is being controlled by humans can truly be called intelligent. Humans sometimes believe and do really stupid things, even highly intelligent ones. We like to feel in control of things and feel threatened by things we don’t understand and can’t control. The entire history of humanity is about us learning about and understanding the world around us so that we can find ways to control it.
We fear AI because we don’t understand it and the way it comes up with its answers. It is a very big leap from there to trusting AI and giving it the power to make important decisions. But the counter argument is: do we understand how human brains work and how they come up with the answers they do? Not really. Trust has to be earned and is easily lost.
"For me it means, that we (the whole humanity) will need to switch to some 'Star Trek'-like 'economy' where means of survival (with adequate quality of life) and growth (intellectual, social, etc.) are not contingent on 'jobs/salary'"
Oh boy, let me introduce you my pal, he is called Karl Marx and has been saying this for 150 years.
With some things added to the mix like "before we move to communism we need an intermediary step of socialism where we wrestle control of the means of production from capitalists. Oh and we should have centralized planning in communism"
The part you're referring to from Marxism makes sense. There are many others that don't.
With all due respect, that's utopianism, and is not supported by any historical or theoretical evidence. Marx himself said:
"The founders of these systems see, indeed, the class antagonisms, as well as the action of the decomposing elements in the prevailing form of society. But the proletariat, as yet in its infancy, offers to them the spectacle of a class without any historical initiative or any independent political movement.
[For the utopians] historical action is to yield to their personal inventive action; historically created conditions of emancipation to fantastic ones; and the gradual, spontaneous class organisation of the proletariat to an organisation of society especially contrived by these inventors. Future history resolves itself, in their eyes, into the propaganda and the practical carrying out of their social plans.
[...]Hence, they reject all political, and especially all revolutionary action; they wish to attain their ends by peaceful means, necessarily doomed to failure, and by the force of example, to pave the way for the new social Gospel."
Dear Tomas, Please address the following issues in your next articles on AI.
1) while the criteria for “ good“ soccer playing he’s quite straightforward, the criteria for “good“ stories or drawings or architecture arr not straightforward. For example trumplican‘s would prefer to have stories in which virtuous white protestants crush evil “ others” . As we have seen in social media, what is good for advertisement is not necessarily good for society. How would we resist a race to the bottom in bad taste.
2 with regard to Bloom, training is different from educating. You can train a composer in musical Theory but what are the conditions that allow a greater proportion of latent geniuses to become a Brahms or Schoenberg?
Humans do have broad agreement on what stories are good or bad, and in case you don't know, I have a book and a TEDx in storytelling. I believe there are objectively better stories than others.
More broadly, your question is: What makes humans happy? And that's a hard one, but I intend to address it.
I'm not sure I understand your difference between training and education.
I have some thoughts on the geniuses though. One of their commonalities? Very good 1-1 tutors.
You could say that there is no objective criteria for what music is considered good, but it turns out that virtually all of the most popular music in human history fit certain patterns.
Tomas, from a practical perspective, what would you do if you currently worked in a profession that’s likely to be automated away in the near future? Let’s say you’re mid-career, kids, mortgage, etc. Do you proactively quit your job and transition into a manual labor-oriented field that pays far less than what you currently make as a white-collar worker? Hold on at your current job as long as you can and hope society figures out a safety net before your job gets eliminated, so you don’t end up on the street? I 100% agree with your conclusions about future job loss, but have no idea how to plan for the future in a world where thinking has no economic value.
Your posts are way too long and too detailed. I don't really know if people in general are interested in that so much detail. And I'm sure they can be summarized.
Surely that's the whole point of Substack? Longer, more detailed articles, with every point of argument thoroughly backed up. People work in a variety of jobs, so going through the consequences of AI for each work area was very interesting and reinforced the main points about latent demand and productivity at the beginning of the article.
Excllent overview of an important and urgent topic, as always. The one issue I have my doubts about, though, is childcare. Would parents feel comfortable leaving their young children in the care of AIs? They might, but I suspect culture and social stigma might hinder or at least slow this development (on a gut level I'm not sure that would be a bad thing, either)
This hinges on a realization: from what I know so far, my hypothesis is that education is mostly worthless, and that we're just paying for the childcare piece. If that is true, and people realize it, they will be much less inclined to pay $$ for a service that is inferior.
Well, note that public schools are paid for indirectly (through taxes). And considering the cost of private childcare, I’d say many parents rather like what they are getting from “free” public schools right now.
This could be true above a certain age. I work in a kindergarten, and most childcare tasks involve taking them to the toilet, helping them eat/get changed, calming them down, taking them outside to run around etc. This takes up the bulk of the time. I can't imagine AI doing any of this. Possibly from ages eight and up? The education chunk that the parents see is things like performing in a play in English (I'm in Japan) or a craft activity, again something I can't see AI doing anytime soon.
Sounds like AI would be very complementary then with childcare. As you say, no AI can do these tasks or will in the coming years. But they could be entertaining / guiding / teaching / encouraging the kids while you're taking care of these tasks?
I was in Semiconductor Manufacturing for 30yrs and the whole time the threat of automation was on our minds. Turned out it was a great boost to productivity and repeatability but had little impact on fixing and maintaining machines and developing processes. Even the threat of oversees competition wasn't that much of an issue since innovation was our strength. AI might take that role but we will have to watch.
We have fewer farmers than before, but what about the food industry itself? Maybe the numbers of jobs in the field of food actually went up (fast food, food companies). The snack industry now is surely much bigger than whatever it was back then, for example. Total demand in calories may not have went up too much, but now people want food that never spoils, portable food, etc.
I mean we don't have 70% of the population working in these jobs, so I think it's safe to assume that there was a net destruction in food jobs, even while the quality and quantity have improved. The examples you give are relevant to quality I'd reckon.
I am disconcerted by the idea of AI raising children. I think we need a lot of research into brain development before we allow such a thing in society. The human component of loving and caring makes a big difference in the way a child‘s brain and personality develops.
But is having one tired adult (not the parent) for 10 children the best way to show love? Or is it a dedicated AI?
Remember, this wouldn’t be a total replacement of humans, but rather a complement for all these times when children can’t have a dedicated human! Sounds strictly better to me than the current system
This should make for interesting conversation at "Career Day" at my daughter's elementary school next week :)
The traditional format of questions: What do you? What was your career path to where you are today? Where did you go to school and what degrees did you get?
My path and current job don't feel relevant to these kids. It won't be their world. What do I say to them?
Everything is up in the air. Career day should be: "Go to chatGPT, go to Midjourney, go to RunwayML, go to Hugging Face, go to Replit, study all of them, and you can make money in months and be properly set up for the future
Any commitment to a 4y thing (or worse, decades) sounds like professional suicide to me right now
A lot of things will be up in the air, but will managers and entrepreneurs go away? I doubt it because humans still hold the power and will want humans held responsible for any decisions even if aided by AI.
Hmmm... True in the short term. Not sure about the long term
In the long run, we're all dead, said Keynes (https://econ.economicshelp.org/2008/10/in-long-run-we-are-all-dead-jm-keynes.html). Unfortunately for us, the long run is probably as long as he assumed, which is more or less a decade. I'm fully expecting we'll end up here sooner or later: https://en.wikipedia.org/wiki/Player_Piano_(novel)
I really do not have a good idea what to tell my children, as the value of education is getting lower every day, unless you get into an exclusive club (a.k.a. Ivy league) just to meet people. I don't have a million bucks (https://www.bloomberg.com/news/articles/2023-03-31/how-to-get-your-kid-into-an-ivy-league-college-parents-pay-750k-for-a-shot), so it's going to be the %99 for them. The best I can do it to help them understand what's going on, so they know who's responsible for this and maybe do something about it before it's too late (i.e. now).
Teach them to build products with AI. Very low cost of entry, very high reward
Tomas, you provide new ways of seeing the world by revealing previously unrecognized patterns in events and in society. Once seen and understood, seeing these patterns is extremely valuable. This article does than in spades. Thank you so much for your careful research and very thoughtful analysis. I’ve come away from reading your article seeing the world in a new way, which will help me navigate the future.
Very glad to hear, this is the goal of UT! I need to add this to my list of testimonials :)
If you are going to use it for your testimonials I suggest you delete the word “often” in the first sentence. It dilutes the meaning.
I deleted”often”
Qué maravilla de artículo. Para leerlo una segunda vez, con mucha calma. ¡Gracias Tomás!
Gracias Pablo! Me alegro!
I really have a problem when we talk about AI replacing Authors and Artists like its nothing. Likewho actually wants that? who does that serve? do we sacrifice creativity to become mindless consumers?
people churning out AI generated content are often doing so for a quick buck. Which I guess the end product is not much different from typical mid level commercial art. But they don't understand that engaging with the process really is the act of creation. I guess I can see a situation where real artists and writers could use tools to speed up their work (since there is a lot of tedious stuff involved) while maintaining their artistic integrity.
But the reason a lot of artists/creatives view AI with such distain is they understand that there's no sense of control to what the AI spits out. It's not their work. It has none of them in it.
But hey, maybe we have a future where those who have the fundament skills in art/writing can put out elevated content. It's not there yet.
I'm not hopeful though. "Good enough" is good enough for the vast majority of people and companies.
Do we want to read novels and listen to music created by AI? half the appeal is to engage with the human behind the work.
2 thoughts:
1. If in your paragraph you change art for agriculture and artist for farmer, you'll see that the functional/market part of different "arts" (eg illustration or videos for marketing) will be much better off
2. Don't worry about the "artistic" part of art. We like art because of its scarcity. The moment part of art is not scarce anymore, art moves on to something scarce (ie that only humans can do). Consider for example the fact that thousands of years of evolution towards more realistic painting suddenly disappeared after the invention of photography, and creativity suddenly exploded with things like surrealism, cubism, or impressionism.
With the advent of photography yes, we got impressionism, cubism, surrealism. We also got the loss of hundreds of years of the development of very refined artistic methods that only recently have become appreciated again by a narrow band of society and are in the process of being restored. We need to recognize that with change we risk losing things are of great value. We risk throwing the baby out with the bathwater.
Now everything is recorded so we wouldn’t have this type of loss
Even if we did, I’d argue losing these techniques pales in comparison to the value of photography!
I find your comment kind of scary. What was lost was centuries of the development of skills that are part of the reason that we go to museums to see work made during that time. The skills themselves are meaningful. To lose them in society is a very great loss. With all due respect, to make light of them and say because we have photography they are not important indicates to me a lack of understanding. Fortunately, they are in the process of reemerging. And, I’m not sure simply recording them is sufficient because the way they are transferred is from master to student.
My comment is not to say they are worthless.
My comment means that photography is more valuable to humans than these techniques.
I guess the reason why I'm saying this is because of the facts: if they were useful, we wouldn't have forgotten them.
I recognize this is a very utilitarian way to think about it though, and as you say it misses the understanding of a more subjective value.
I don’t think photography is more valuable than fine art. I think more people use photography than explore fine art. And, as we move down the road with this intense focus on becoming more and more productive if we lose the subjective—-if we lose love, truth and beauty—-I think it will become a very sad and empty world.
Machine-made shoes means there are far fewer cordwainers in the world, but I’d rather live in a world where there are plenty of shoes than one where shoes are expensive so many people have to go without shoes.
I think it doesn’t have to be either or. It can be both and. Both mass produced shoes and hand made artisan shoes for those with the money and the interest.
Tomas you haven't allowed for the upswing in live music that is happening. More and more bands, from the large to the small, the well-known to the extremely local, are touring, playing their music to live audiences. At least here in Aotearoa New Zealand any way!
Would love to see data on this! My current hypothesis is that if it's truly a trend that's happening, the upswing is a fraction of what it used to be. I'd love to be proven wrong!
It’s what I talked about about humans being forced in to bespoke niches. As the mass commercial item becomes cheaper, people will want more idiosyncratic bespoke experiences. We see this in beer too, where there were a ton of local brewers in the US (because nobody could ship beer far), then the beer market became dominated by a few big players and now there is a craft brewer revolution (and hipster types drink stuff like Pabst which use to be consider cheaper more inferior more blue-collar beer just because that’s the hipster thing to do).
Social status points are a resource that is always in short supply so there is constant competition for them
I have tried to obtain data Tomas. It seems no-one in Aotearoa NZ collects data on this. Shameful I know!
Maybe it’s because the trend is not as big! 😅
Kia ora from a Kiwi in London. And here's to live music and all musicians, not just the famous ones (though Tomas' points do apply unfortunately).
I think this clearly leads to the conclusion that somewhere, not so far, in the future (or would it be 'somewhen'?) the balance job creation-job destruction will got persistently in the negative side.
For me it means, that we (the whole humanity) will need to switch to some 'Star Trek'-like 'economy' where means of survival (with adequate quality of life) and growth (intellectual, social, etc.) are not contingent on 'jobs/salary'.
Some 'Singularity' pundits take it as a given that exponential growth would bring solutions faster and faster, bringing a nearly painless transition to an 'abundance' heavenly scenario , but I think a more tortuous, and suffering-ladden for billions of people, path is presently at our door.
Until we have ubiquitous energy supply and energy to mass hardware (food synthesizers, etc.) like Kirk/Picard/et al.. I think we will need to accelerate some huge efforts on UBI (Universal Basic Income) implementation.
That is a socio-political nightmare as complex to unravel as climate change de-worsening initiatives (we can't prevent significant climate change anymore, so we hope that work can be done to make it less harmful), and this AI-boosted clock is also ticking fast...
PS.: there is also the 'crime' business to consider, where sadly many jobless people find their new "way-of-life"
The problem with a Star Trek-like economy is that while we're about to automate the hell out of anything having to do with information processing, we're still fundamentally limited by raw materials, i.e. what we can dig out of the ground. I totally agree that we need to fundamentally rethink the social contract because the intrinsic capital value of human labor is going to be driven down almost to zero, but the intrinsic value of natural resources will only increase. The people who own/control those natural resources will probably resist the redistribution of those resources in a UBI system, so . . . I'm not hopeful for the long-term future. Honestly, the only way the equation gets balanced if there are a hell of a lot fewer humans around trying to consume resources, so that will probably happen eventually, either gracefully or not.
You will love, love, love an upcoming article!
It depends on what you call labor. Humans labor certainly will move to those areas that are tougher to automate. Originally, human labor was largely doing backbreaking stuff that really is better off done by machines. Soon, purely intellectual work will be better handled by AI. Though AI actually currently has trouble handling an idiosyncratic physical world that humans and other animals evolved to handle much easier (I’m thinking of driving in city streets in the rain). So plumbers will be around for a while.
Humans also hold the power. I’m pretty certain that humans will always want other humans responsible for decisions they deem important (so we won’t elect AI lawmakers, though lawmakers will be assisted by AI in the future), and there will be human managers, human entrepreneurs, etc., each with an AI army working for them.
That’s also why I think the general area of risk will still have plenty of human involvement (AI will help, but I don’t imagine humans becoming OK with purely AI determining how much capital banks should hold or which banks to rescue or let go under).
Humans will also be more employed in what I consider the luxury goods industry (which, thanks to a great increase in productivity and wealth, more people will enjoy). So everyone will have personal therapists to figure out the meaning of life but the poor will only have AI ones while the better off will also have human ones. There will also be jobs related to the physical world. I’m sure people will still want to travel, and that will be limited by the cost of transportation, but while maybe fewer people may be needed for the actual transportation part, and while I’m sure you can get AI tour guides, getting a human tour guide would be seen as a luxury good (not everyone will be able to afford the actual real human touch and insight).
I had imagined that AI designers would aim for thinking machines that REMOVED human bias and emotion to provide optimised advice and solutions. The current versions seem to be more about just learning from the version of "us" that is available to them via the internet with all its biases
I mean, “optimized” depends on your values and goals so always may be “biased” based on your point of view. As a crude example, an algorithm to optimize white power certainly would seem biased to some. An algorithm to optimize engagement may also lead to bias (because humans are biased about what they will engage with). And yes, ML trained in “biased” datasets can only be expected to be biased.
Thanks for all your helpful AI comments. I wrote a reply ages ago, but it raised more questions than answers so it felt sensible to just think about things for a bit. I suppose my concept of optimised could best be defined by the principles that Tomas uses in Uncharted Territories and the logical conclusions that those principles lead to. Whether people have the wisdom to choose those logical solutions is another matter. As you say above, humanity still holds the power (for now), but the question is: which specific humans get to make the decisions? History suggests that those with power and wealth will not give it up lightly and that they tend to make decisions based on their own short-term interests rather than for the good of humanity (whatever that means!). If we could harness the Wisdom of Crowds we might have a better chance of making good decisions.
Part of the reason for writing this now is that I recently posted a couple of things about the importance of what and how we learn, both as individuals and as groups. This inevitably brought me back to the AI questions. How will the precocious child that is AI learn and who will be the teachers? It is true that a subservient AI might happily advise on optimising for white power, but if we are really talking about Artificial Intelligence that is smarter than humans then a reasonable starting comparison would be with the most intelligent humans we have now.
An extremely intelligent human would give a balanced view when asked about optimising for white power. She might first ask “what do you mean by white power?”. He might dispassionately point out the difficulties in deciding who is white and who is not and the downsides of the conflict that would ensue and cause harm to “white” people as well as others. They might point out that the optimum solution long term for the wellbeing of white people is to co-operate with all colours of humanity to increase the power of humanity as a whole, so the white slice of the pie would be bigger. An intelligent and wise human would take into account the context of the question
What we have now is machine learning which is just a new and powerful tool. But what are the machines being taught and by whom? An algorithmic tool can just magnify the bias and lack of understanding of the users and might just produce “Artificial Prejudice” or “Artificial Stupidity”. The question is whether anything that is being controlled by humans can truly be called intelligent. Humans sometimes believe and do really stupid things, even highly intelligent ones. We like to feel in control of things and feel threatened by things we don’t understand and can’t control. The entire history of humanity is about us learning about and understanding the world around us so that we can find ways to control it.
We fear AI because we don’t understand it and the way it comes up with its answers. It is a very big leap from there to trusting AI and giving it the power to make important decisions. But the counter argument is: do we understand how human brains work and how they come up with the answers they do? Not really. Trust has to be earned and is easily lost.
This is the biggest challenge of our time
"For me it means, that we (the whole humanity) will need to switch to some 'Star Trek'-like 'economy' where means of survival (with adequate quality of life) and growth (intellectual, social, etc.) are not contingent on 'jobs/salary'"
Oh boy, let me introduce you my pal, he is called Karl Marx and has been saying this for 150 years.
With some things added to the mix like "before we move to communism we need an intermediary step of socialism where we wrestle control of the means of production from capitalists. Oh and we should have centralized planning in communism"
The part you're referring to from Marxism makes sense. There are many others that don't.
Do you think humanity would naturally and spontaneously reorganize into communism without any kind of struggle? That's optimistic.
I do! With the right plan.
I’ll share it in the coking weeks
With all due respect, that's utopianism, and is not supported by any historical or theoretical evidence. Marx himself said:
"The founders of these systems see, indeed, the class antagonisms, as well as the action of the decomposing elements in the prevailing form of society. But the proletariat, as yet in its infancy, offers to them the spectacle of a class without any historical initiative or any independent political movement.
[For the utopians] historical action is to yield to their personal inventive action; historically created conditions of emancipation to fantastic ones; and the gradual, spontaneous class organisation of the proletariat to an organisation of society especially contrived by these inventors. Future history resolves itself, in their eyes, into the propaganda and the practical carrying out of their social plans.
[...]Hence, they reject all political, and especially all revolutionary action; they wish to attain their ends by peaceful means, necessarily doomed to failure, and by the force of example, to pave the way for the new social Gospel."
Excellent article! Time to incorporate AI into my job!
For sure! Don't wait!
Dear Tomas, Please address the following issues in your next articles on AI.
1) while the criteria for “ good“ soccer playing he’s quite straightforward, the criteria for “good“ stories or drawings or architecture arr not straightforward. For example trumplican‘s would prefer to have stories in which virtuous white protestants crush evil “ others” . As we have seen in social media, what is good for advertisement is not necessarily good for society. How would we resist a race to the bottom in bad taste.
2 with regard to Bloom, training is different from educating. You can train a composer in musical Theory but what are the conditions that allow a greater proportion of latent geniuses to become a Brahms or Schoenberg?
Interesting questions.
Humans do have broad agreement on what stories are good or bad, and in case you don't know, I have a book and a TEDx in storytelling. I believe there are objectively better stories than others.
More broadly, your question is: What makes humans happy? And that's a hard one, but I intend to address it.
I'm not sure I understand your difference between training and education.
I have some thoughts on the geniuses though. One of their commonalities? Very good 1-1 tutors.
You could say that there is no objective criteria for what music is considered good, but it turns out that virtually all of the most popular music in human history fit certain patterns.
Tomas, from a practical perspective, what would you do if you currently worked in a profession that’s likely to be automated away in the near future? Let’s say you’re mid-career, kids, mortgage, etc. Do you proactively quit your job and transition into a manual labor-oriented field that pays far less than what you currently make as a white-collar worker? Hold on at your current job as long as you can and hope society figures out a safety net before your job gets eliminated, so you don’t end up on the street? I 100% agree with your conclusions about future job loss, but have no idea how to plan for the future in a world where thinking has no economic value.
I work not in one, but 2 of them: creator and product person.
My answer is to dive right into AI and master the tools. You can be either the automator or the automated.
Also, save up for an early retirement. If you could transition to a job that pays far less, you should be able to sock a lot away for retirement.
Your posts are way too long and too detailed. I don't really know if people in general are interested in that so much detail. And I'm sure they can be summarized.
Thank you Maria. I hear that consistently. I need to get better at it. Maybe I'll ask ChatGPT to summarize them?
I didn’t find the article too detailed. I thought it was just right.
The more detailed the better. I don’t want fluff.
Maybe a TL’DR would help in the meantime
Surely that's the whole point of Substack? Longer, more detailed articles, with every point of argument thoroughly backed up. People work in a variety of jobs, so going through the consequences of AI for each work area was very interesting and reinforced the main points about latent demand and productivity at the beginning of the article.
Excllent overview of an important and urgent topic, as always. The one issue I have my doubts about, though, is childcare. Would parents feel comfortable leaving their young children in the care of AIs? They might, but I suspect culture and social stigma might hinder or at least slow this development (on a gut level I'm not sure that would be a bad thing, either)
Yes, maybe.
This hinges on a realization: from what I know so far, my hypothesis is that education is mostly worthless, and that we're just paying for the childcare piece. If that is true, and people realize it, they will be much less inclined to pay $$ for a service that is inferior.
Well, note that public schools are paid for indirectly (through taxes). And considering the cost of private childcare, I’d say many parents rather like what they are getting from “free” public schools right now.
This could be true above a certain age. I work in a kindergarten, and most childcare tasks involve taking them to the toilet, helping them eat/get changed, calming them down, taking them outside to run around etc. This takes up the bulk of the time. I can't imagine AI doing any of this. Possibly from ages eight and up? The education chunk that the parents see is things like performing in a play in English (I'm in Japan) or a craft activity, again something I can't see AI doing anytime soon.
Sounds like AI would be very complementary then with childcare. As you say, no AI can do these tasks or will in the coming years. But they could be entertaining / guiding / teaching / encouraging the kids while you're taking care of these tasks?
You think education is worthless *currently* or that it's going to become so in the near future? Seems like an important distinction.
I'll write about this, but I also need to learn some more before I do!
I was in Semiconductor Manufacturing for 30yrs and the whole time the threat of automation was on our minds. Turned out it was a great boost to productivity and repeatability but had little impact on fixing and maintaining machines and developing processes. Even the threat of oversees competition wasn't that much of an issue since innovation was our strength. AI might take that role but we will have to watch.
Hopefully most industries are like that
Elon Musk's solution seems like the only path for people who want to survive post-AI. Get implanted with Neuralink and live the borg lifestyle.
This is the natural evolutionary step for a single hivemind. People prefer security over freedom. They will get the conformity so desired.
Some people prefer security over freedom, some don't
It might be an option!
We have fewer farmers than before, but what about the food industry itself? Maybe the numbers of jobs in the field of food actually went up (fast food, food companies). The snack industry now is surely much bigger than whatever it was back then, for example. Total demand in calories may not have went up too much, but now people want food that never spoils, portable food, etc.
I mean we don't have 70% of the population working in these jobs, so I think it's safe to assume that there was a net destruction in food jobs, even while the quality and quantity have improved. The examples you give are relevant to quality I'd reckon.
I am disconcerted by the idea of AI raising children. I think we need a lot of research into brain development before we allow such a thing in society. The human component of loving and caring makes a big difference in the way a child‘s brain and personality develops.
But is having one tired adult (not the parent) for 10 children the best way to show love? Or is it a dedicated AI?
Remember, this wouldn’t be a total replacement of humans, but rather a complement for all these times when children can’t have a dedicated human! Sounds strictly better to me than the current system
Really interesting read. Thanks for this.