WARNING: Summaries are generated by a large language model and may be inaccurate. We suggest that you use the synopsis, short and long summaries only as a loose guide to the topics discussed. The model may attribute to the speaker or other participants views that they do not in fact hold. It may also attribute to the speaker views expressed by other participants, or vice versa. The raw transcript (see the bottom of the page) is likely to be a more accurate representation of the seminar content, except for errors at the level of individual words during transcription.

AI is disrupting many industries and professions, bringing with it potential liabilities and advantages. Major countries like China are investing in AI and automation, and knowledge workers are the most vulnerable to displacement. Open AI is attempting to own images generated by AI, but potential for deepfake technology has caused restrictions. AI is changing the way humans find meaning in work, and STEM fields may no longer be reliable for middle class lifestyles - liberal arts and fine arts could be more beneficial.

Disruption of industries can lead to a reversal of fortune for companies and nations due to the acceleration of new technology and the decline of old industries. This can have a significant impact on GDP, as the value of assets and infrastructure invested in the industry can be large. Fossil fuel infrastructure, such as drilling sites, pipelines, refineries and fuel pumps, are worth trillions of dollars, but if demand disappears, they can become liabilities. This means the fossil fuel industry must plan for the possibility of stranded assets, as the original advantage of these assets has become a disadvantage.

Technology can create large liabilities, such as the maintenance of fossil fuel infrastructure and decommissioning of nuclear power plants, which can be a costly burden for companies and society. Advantages and disadvantages can quickly change due to new technology, such as Saudi Arabia's newfound fossil fuel wealth or small island nations leveraging maritime trade. China's large population may soon become a disadvantage due to the potential of a phase transition in the availability of cognitive labor, and AI taking away jobs that humans are typically competitive at, leading to a decrease in economic and geopolitical power.

AI and automation are causing a major disruption to the global economy, with implications for labour and export economies such as China. This could lead to a decrease in economic contributions, but could also result in increased productivity and prosperity due to a higher level of labour abundance. This could result in changes to the economy's structure and distribution of resources, and scientists must be careful to consider all possible consequences when predicting the future.

AI is rapidly changing the world, with the Chinese government investing heavily in educating their population to move up the value chain. Automation of cognitive tasks and Nanobots are becoming more widespread, replacing many programming jobs and making military equipment obsolete. Those with the necessary skills and resources will benefit the most from AI, while those without may be left behind. It is important to consider both the fast and slow factors when considering the potential impact of AI.

China has been touted as a leader in AI development, but in the past five years it has lagged behind other countries in creating its own Chinese language datasets. Adam argued that automation and difficult work are not strongly correlated and knowledge workers are more vulnerable to automation than unskilled workers. AI is quickly disrupting professions such as illustration, knowledge workers, and those in STEM, resulting in a swift decline in the need for many professions. AI is becoming an increasingly powerful tool and will continue to disrupt many professions in the near future.

AI is being used to generate images and automate jobs such as anesthesiologists, displacing the need for human intelligence and knowledge. Open AI is attempting to own the images generated, however the potential for deepfake technology to generate pornography has caused restrictions on Dolly 2. Society is likely to push back against automation in the future, and the culture that made countries successful in the agricultural and industrial eras are very different.

Humans have traditionally found meaning in hard work, but this may no longer be the case as cultures shift away from it. This could lead to neurotic behaviour or grinding virtual reward hierarchies, making for a miserable experience. STEM fields have been reliable for middle class lifestyles, but may not produce well-rounded individuals, so a focus on liberal arts, philosophy and fine arts may be more beneficial. Despite jokes that suggest Arts degrees will lead to McDonald's, they may have the last laugh as they can use their skills to be creative and sophisticated in ways STEM fields cannot.

Disruption of industries can lead to a reversal of fortune for companies and nations. This is caused by a pair of feedback loops which accelerate the adoption of new technology and the decline of old industries. This can have a large impact on the GDP of a nation, as the value of the assets and infrastructure invested in the industry can be large relative to the GDP. This can be seen in the United States, where the value of assets and infrastructure can be viewed in various ways.

Fossil fuel infrastructure such as drilling sites, pipelines, refineries and fuel pumps are worth trillions of dollars. However, if demand for these assets disappears, they will become liabilities due to the overhead and operating costs of maintaining them. This means that the fossil fuel industry must plan for the possibility of stranded assets, as the original advantage of these assets will become a disadvantage. This is a reversal of fortune, as the circumstances have changed, making the assets no longer useful.

Technology can often lead to large liabilities, such as the maintenance of fossil fuel infrastructure and decommissioning of nuclear power plants. These can be a costly burden for companies, and when companies are restructured and allowed to go bankrupt, the liabilities fall to the state and become a liability for society. Accounting sets do not always reflect this, as liabilities may not always move from one balance sheet to another. These liabilities can be a major cost for governments and society.

Advantages and disadvantages can quickly change due to new technology. For example, Saudi Arabia did not have any 19th century biofuels, but in the 20th century its phenomenal endowment of fossil fuel wealth transformed the situation. Similarly, small island nations may lack resource endowment, but with the advent of sailing technology and maritime trade, they can quickly turn this into an advantage. Britain is an example of this, having built an empire through maritime trade.

China was once an advantage due to its large population, however, this may soon become a disadvantage due to the potential of a phase transition in the availability of cognitive labor in the next 5-10 years. This could lead to a demographic crisis as the population becomes older and fewer working age people are available to pay for the productivity. This could be further exacerbated by the potential of AI taking away jobs that humans are typically competitive at, leading to a decrease in economic power and geopolitical power.

China has a large population and has been using it to its advantage to become a powerful economy. However, with the rise of advanced robotics, mundane programming jobs may no longer require humans and this could lead to a decrease in China's geopolitical might. This is an example of the ceteris paribus principle which states that when other factors remain the same, a large population may no longer be an advantage. This could have a major effect on how countries view their population and power in the future.

Ai and automation of labour is causing a major disruption to the current paradigm of producing goods and services. This disruption can have a large number of consequences, including the obviation of human labour. Scientists need to be careful to avoid the 'all else equal' trap when considering the future, as reality cannot be put in a laboratory. If this were the only consequence of the disruption, it would be a terribly frightening situation.

AI and automation are causing a disruption in export economies, particularly China, which relies heavily on cheap labour. This will cause a massive shift in the economy and could lead to a decrease in economic contributions. However, AI and automation could also lead to increased productivity and prosperity everywhere, including China, as it could allow for a higher level of labour abundance. This could potentially lead to a change in the economy's structure and distribution of resources.

AI could potentially have a large impact on the world, however it is important to consider the fast moving factors as well as the slow ones. For example, when Britain industrialised India, the terms of trade changed rapidly and India was colonised by the East India Company before they had a chance to build factories. It is unlikely that robots will be widespread in China within five years as it will take time to build factories. Even after human level intelligence is invented, it will take time for automation of cognitive tasks and Nanobots to become widespread. It is important to consider both the fast and slow factors when considering the potential impact of AI.

AI is rapidly changing the way the world works and its effects are being felt in China. The Chinese government is investing heavily in educating their population to move up the value chain and into software and more advanced tech. AI is already replacing many programming jobs and this trend is likely to continue, leaving those without the necessary skills at a disadvantage. This is not limited to China, as AI will soon make military equipment obsolete and robots will be able to manufacture things much faster than humans. As AI advances, those with the necessary skills and resources will be the ones to benefit the most.

China has been touted as a leader in AI development due to its large population and ability to create large datasets. However, this has not been the case in the past five years, as China has been a follower in this time. Although it may be true in some limited domains, large datasets are not necessarily a determining factor. Furthermore, China has lagged behind other countries in creating its own Chinese language datasets.

Adam discussed how automation has changed the way people traditionally viewed professions for knowledge workers. He argued that automation and difficult work are not strongly correlated, and that knowledge workers are more vulnerable to automation than unskilled workers. He gave the example of walking across a room as something humans take for granted, but in reality requires a lot of intelligence. He suggested that until recently, the public may not have fully understood this.

AI is quickly disrupting many professions, such as illustration, knowledge workers, and those in STEM, with the ability to quickly generate images at a low cost. This will result in a swift decline in the need for many professions, such as anesthesiologists and lawyers who do intellectual grunt work. An example of this is a friend from LA who is working on IP for a Chinese games company and is using AI to come up with ideas for the script. AI is becoming an increasingly powerful tool and will continue to disrupt many professions in the near future.

AI is being used to generate images for creative directors and other teams in the industry. This is a displacement of the need for an artist, as the AI can be tweaked to produce the desired images with no need to wait for an artist to complete the task. Open AI is attempting to own the images generated, however it remains to be seen what this will look like in the future. The use of AI in this way is becoming increasingly popular, and it is likely to continue to grow rapidly.

In the digital age, it is becoming increasingly clear that professions that rely on human intelligence and knowledge are no longer safe from automation. This is particularly true for medical professionals such as anesthesiologists and technicians who currently earn a high salary. The potential for deepfake technology to generate pornography has also caused restrictions on Dolly 2, a new version of the original Dolly. Society is likely to push back against automation in the future, and it is uncertain what form this pushback will take. Additionally, the culture that made countries successful in the agricultural era, such as the Protestant work ethic, is very different from the culture that made countries successful in the industrial era.

Humans have traditionally found meaning in hard work and have seen it as a moral virtue. However, in the new era, this may no longer be the case. Cultures that do not place as much emphasis on hard work, such as those in the Middle East, can be seen as having merits, such as celebrating leisure and family. This shift away from hard work could lead to neurotic behaviour, as individuals' source of identity and validation is taken away. Alternatively, people may shift to grinding virtual reward hierarchies, as seen in an episode of Black Mirror, which is a miserable experience.

Stem fields have been a reliable and secure way to ensure a middle class lifestyle for the past few decades, leading to a surge in emphasis on these fields in the education system. However, stem fields do not necessarily produce well-rounded and sophisticated individuals, and a focus on liberal arts, philosophy and the fine arts may be more beneficial in achieving this. Despite jokes that suggest those with Arts degrees will end up working at McDonald's, they may have the last laugh as they can use their skills to be creative and sophisticated in ways that stem fields cannot.

AI is a major concern for the future and transitioning to a post-Singularity Utopia without catastrophic consequences is a challenge. China must find a way to survive the shift away from human labor. Science fiction fans have dreamed of robots doing all the work, but getting there will require good decisions, luck and planning. To mitigate potential harms, we must figure out how to solve collective action problems and start convincing the public of ideas such as universal basic income or other forms of guaranteed prosperity.

yeah so I sent over a note to you with Dana a couple days ago because he wrote this really interesting thing giving a perspective about Ai and and China and power and population and what it made me think of is sort of a pattern that we see when Industries are disrupted which is that um you know portions uh The Fortunes of an industry or especially of an individual company um they can kind of flip around and uh they can they can invert pretty quickly right what's the word I'm working a reversal of Fortune sorry that took me a while to claw out of this Dusty old brain of mine um yeah so reversal of Fortune is something that we see uh pretty commonly when an industry is disrupted it just it's you know it's it's all part of the Dynamics of um uh you know gearing up and equipping and capitalizing with the focus to produce one good or service and then suddenly the you realize that all of that investment um is it it swiftly loses value as consumers and customers just just wanting and stop demanding whatever it was you were producing and start demanding something new um and so the the you know you can you can imagine there are a number of of sort of reinforcing feedback loops in that Dynamic and we've spelled we've kind of um in our uh theoretical framework we've kind of identified the main uh uh causal Pathways there and it forms a pair of flywheels um that causes sort of accelerating adoption sort of a feedback a set of feedback loops that accelerate the adoption of new technology and another set of feedback loops that accelerate the decline of the old Industries based on older technology so you know it's it's not anything that's particularly mind-blowing but what what the result is that for for any given industry if it has you know Assets in the form of equipment um if it's if it has invested in building out infrastructure to support its goods or services um and this can be an industry but also an entire nation um that that that those assets that infrastructure when you're talking about big Industries fundamental things like energy or Transportation those the value of those assets and that infrastructure can be very very large relative to this you know to the country's GDP right so in the United States for example um uh the the the value the current the the the depends on how you want to look at it you know the current value the scrap value the um uh the original investment required there's a number of different ways you can look at at how to you know value assets none of which are

accurately reflected on actual accounting balance sheets um there's a lot of creativity that happens there but at any rate one could imagine that by any reasonable standard the the value of all the stuff all the equipment all of the gear um in the fossil fuel uh uh ecosystem right so the the everything from the drilling sites and and all of the equipment that's there to the pipelines to move things around to the refineries to the you know fuel pumps you know fuel where you pump your fuel at the at the petrol station and all of that and all of the infrastructure that goes with it but this you know for for something like our energy industry that's in the United States that's trillions of dollars that were originally spent building that stuff out and so one thing that's going to happen is that is that uh the Val those assets will cease to become um useful and they will effectively turn into liabilities because maintaining the assets maintaining infrastructure um that isn't free that's not Costless so there's some overhead there's some you know operating expense and additional in addition to the original capital expenditure to keep this stuff going and keep it continuing to be valuable um but you know if if demand disappears um then all of a sudden just holding that stuff and having to pay to keep it up up and running or even to just to pay to decommission it safely and get rid of it these this is this is you know it can be an extremely costly Prospect and the Energy System right now if the fossil fuel system is starting to face that Prospect you know the the the um the writing is on the wall and the fossil fuel industry has already begun to um uh plan to uh deal with this challenge of stranded assets okay so what is this at a fundamental level well it's it's a reversal of Fortune right what the things that give you an advantage under certain circumstances under one set of assumptions under one status quo they become a disadvantage they become a liability to you if circumstances change in in a certain way can I status quo ask about that a little bit I mean it sort of I never thought about them as active liabilities before which is the new part of this story for me so it seems kind of obvious that they might no longer be useful but uh and in that sense you know I suppose you had on your asset sheet something that seems like it's worth a billion dollars because it facilitates future cash flows and now it's worth zero because there is no such future cash flow so I guess in some

sense you lost a lot uh but in a more naive sense it's it wasn't obvious to me that it's somehow like an active drag as opposed to just being something that isn't worth anything anymore your example of maintenance I guess makes sense and at the scale of say fossil fuel infrastructure that could be a large cost indeed uh but that's that's sort of like only necessary as long as you need to eke out some kind of return until you're completely obliterated by the new technology do you have examples of other things where other examples where it's a kind of more severe liability than merely no longer very useful yeah I mean that's a that's a good question I suppose in that particular instance a more severe liability uh in in that you know in the energy sector is uh that you you continue to be responsible for keeping for ensuring that those assets are safe and that can be a very very burdensome responsibility so fossil fuel is not so much but I mean yes substantially but but not catastrophically so I mean yes you have to decommission oil fields in line with environmental law you can't simply abandon pipelines and it's that's not to say this doesn't happen um and you know companies can are deliberately restructured and then allowed to go bankrupt and so forth to kind of escape from from the burden of that but uh you know the the the it depends whether or not you call that a catastrophe is you know is is one thing but if you switch to another part of the energy industry for example I mean look at nuclear power well you can't simply walk away from a nuclear power plant and those assets even if they become useful because they're extremely dangerous and so there's a massive liability attached to decommissioning um uh uh assets that no longer have any any useful any use value that aren't going to generate as you said down any any you know returns any cash flow or anything like that and so they that is absolutely um it is effectively that's why you said effectively I suppose not technically on balance sheets and so forth but effectively what what once we're assets effectively become liabilities and the industries Industries involved often shirk these and they fall to to to the state they fall to governments and so they become they become a liability for society um and so this you know I think probably we have this again the liabilities for society you know things don't move from one balance sheet to another and across scales in any stress accounting sets I don't think it's better to think just

deal with these things and sort of conceptual categories rather than you know in the technical meaning of the assets and uh liabilities but you I think you take my you take my meeting there what seems like a very good idea and gives an industry uh and its host nation advantages under certain circumstances can pivot can can reverse and then and change into um uh disadvantage and that that can happen as a result of new technology as a you know new technology comes along and changes the game now to give it a kind of a completely different example um uh uh and I get I think I mentioned this in the note that I wrote to you Dan also sticking with energy um you can have a reversible a fortune where um uh where you where you once had no advantage and suddenly find yourself having a lot of Advantage so for example Saudi Arabia is is sort of like a quintessential uh uh case right where Saudi Arabia did not have any of the 19th century biofuels that that provided energy for civilization they didn't have whales they didn't have wood because in that many forests didn't have any biofuels of any kind they didn't have large herds of animals they could you know uh make Tallow for uh from you know nothing that very very little in terms of the the biofuel energy endowment that other nations have for heavily forested Nations had you know better endowment of wood and other biomass uh for energy uh but then in the 20th century that changed dramatically and Saudi Arabia's phenomenal endowment of fossil fuel wealth uh transformed that situation completely so that's on that side of things another example that I can think of I don't know if it's a great one but it's another one that comes to mind at sort of the national level is um vegan Island being a small island nations maybe not you know um uh it's possible that has that has some well certainly has a mix of advantages and disadvantages but it's certainly one you can certainly look at that and say well a small island nation is is does it lacks the resource endowment that that a much larger Continental Nation uh might have other things be equal but um uh when sailing technology and Maritime trade became uh sources of um strategic value and uh economic value um well then island nations I'm thinking Britain here um uh seized that and and turned that into a huge advantage and built an Empire obviously out of that so um that those are examples where what were once disadvantages or perhaps that's perhaps simply a lack of Advantage we might want to distinguish

those two things but certainly a lack of Advantage changed dramatically and turned into an advantage um the other way around I'm trying to think of examples where what was once an advantage might turn into a disadvantage I think maybe yeah maybe um I don't know maybe Empires that spread themselves too thin or something like that that then collapsed under the you know the uh the weight of their own overhead kind of kind of thing I don't know if if that if there are examples of that throughout history I'll have to think about it the other way or the other way around where you become hugely disadvantaged well once you you had an advantage that's a good good question and I don't have I don't have any great examples of that other than ins other than the stranded assets of Industries which is what that's sort of the the main thing that we see in our own my team sees in our own work and we look at an industry and see that it's assets we've got our either have or because or are poised to become stranded and that seems like a very very large disadvantage um and uh but maybe that's through some other examples the example that I raised in my email which is population uh so if we take seriously the idea that within five to ten years there might be some kind of phase transition in the availability of cognitive labor because of very Advanced artificial intelligence then that changes the relationship between population and Power in various respects and the example that comes to mind quite forcefully is China which is facing a demographic crisis of kind of terrifying proportions uh as its population becomes older so even absent even absent new features from say AI China's facing a very difficult problem of how to actually pay for its population with its productivity given there will be relatively few working age people compared to old people um going forward but then if you take the fact that China is a not very developed a not very developed economy which still relies largely on Manufacturing and if you take away those jobs as being something that humans are kind of competitive at it's really quite diabolical um and so China goes from a country where it's quite common to to link China's economic power to its geopolitical power and that makes sense right and its economic power is predicated not on high productivity it's a very low productivity country and it's not predicated on high levels of development although you know it is quite a high well-developed place in many places

including Shenzhen there's a lot of high tech going on uh but overall it's not a highly developed economy the the economy in China is Big because there's lots of people there fundamentally um and if that goes from being an advantage to a relative disadvantage because I mean it won't be absolute right but even small percentage changes are probably enough to make big effects if say 10 of the manufacturing population in China or 20 percent is uh is no longer competitive because there's Advanced robotics um or 40 of the programmers in China who are largely doing kind of Fairly mundane programming at least our friends in China are if those jobs become things that can be automated um then the 4.5 million stem graduates that China has every year this is a kind of very impressive statistic for an Australian where we have you know 25 million people in our country right so there are way more so the number of stem graduates every year in China is uh is kind of impressive from the point of view of a future tech industry but if humans just aren't the thing you want in your tech industry anymore then I don't the case for for example China's geopolitical might just kind of collapses now that's just one example China but it's it's kind of a significant one I think so I wonder if this isn't one of the major effects that even as we move into this takeoff right as it becomes clear that this will happen people's thinking around just I mean I think people don't probably think very much about the connection between population and power I mean obviously there's examples like India which are relatively making less use of their population I suppose would be a popular impression compared to China but there's a just an assumption that if you get your ducks in a row then you can exploit a large population to translate that into a powerful economy and therefore a powerful country but that unexamined assumption may go away incredibly rapidly on the time scale of how these things usually go so that was the one I mentioned in my email um so I'm interested in your take on that or if there's other examples similar to that in in in the so I I would I guess my my initial my initial reaction is to is to uh I guess my initial reaction is to evoke a principle that I wrote about some years ago and Daniel will remember this one but I called it the ceteris paribus uh it's it's a I I described it as an informal logical fallacy but it's basically just a a problem um in our it's it's a pitfall it's it's

a it's a it's a a trap that we could sometimes fall into um DCU mistake to make and so my mind goes straight to this because as as enormous as this problem is that you've that you've just described it's also very clear to me that there will be a a large number of you know a significant number um of other uh major changes that are caused by the same thing Ai and automation of Labor and the obviation of human labor um so many other things in our our current you know sort of socialists our current Paradigm as uh you know the the law the overarching system with which we produce um uh goods and services to meet our needs and and everything that's structured around in in relation to that so much of that will be impacted both directly and indirectly by this same technological disruption Ai and I'm thinking narrow AI here not even necessarily in a general artificial intelligence narrow is more than enough to do the disrupting um it seems to me um that that I I I want to be cautious that we when we think through the specific example of impact that you're describing we keep in mind that many other pieces of Machinery are going to be moving around at the same time and so um the the danger of course is drawing conclusions about the future uh that ignore those other changes and that's where you fall into the sort of the all else equal trap right um and this is a trap that scientists you know I I we have to be very very um Vigilant to uh avoid falling into because we're trained basically you fall into it we're trained to understand and appreciate and and work on the value of isolating you know variables and controlling everything else um and holding everything else constant of you know holding all else equal and imposing a center as part of his situation that's what that's what Laboratories are for right and so we're trained to to you know understand the value of that and attempt to to um to reason in those circumstances the problem is that you know when you're talking about the world you know their own reality and that doesn't doesn't you can't put it in laboratory Okay so my initial reaction having episode with that caveat with that caveat in place and understanding that all sorts of other things are going to move around and shift at the same time um yeah I think that this is that this is a it's a it's a it's a terribly frightening uh situation on its face again if that were the only thing that we're going to if that were the only consequence of this disruption it would

be terrifying the idea that that you have a you know a huge population that is who's whose economic contributions um will suddenly be you know nullified they'll be rendered sort of moot um that's terrifying and it's terrifying to any economy but but you know I think quite rightly you make a strong case for it being particularly problematic for China um probably uh all export economies are going to get clobbered by uh uh this disruption in some sense um or in other words they're going to suffer the same it's this same particular effect and we'll talk about you know the other things that might go on here in a minute um but certainly if you are if you're economy at the moment depends on making things and exporting them because you have some sort of advantage that allows you to produce things competitively in other words for a lower cost uh and that you know than other places and so other places are willing to buy them from you well if you're if your primary advantage um that gives you gigantic export economy is labor which seems to be the case in a place like China not the case for every country or even every large country but I think certainly it's one of China's economic advantages is is an abundance of cheap labor well you're going to lose that advantage and Manufacturing again ol SQL manufacturing would move back to places that are importing uh if they can use robots in those places instead of you know having to depend on GPU and labor in China and so that's a massive massive shift and it's a huge you know huge problem that China is going to face um but having said that you know and painted this sort of brother Bleak picture the other thing of course that's going to happen the other the other consequence of all of this of artificial intelligence and the automation of Labor is that it it has the potential to make every locality every location every country certainly um just just massively more productive internally and uh so then that calls into question we'll we'll uh will we continue to have a economy structured at all around the existing um distribution of various resources or will uh you know will we be looking at a situation where labor is so abundant that uh it's possible to provide for prosperity at a you know at an extremely high level everywhere including in China I mean if China had if China had instead of a billion uh human laborers if China had a billion or 10 billion or 100 billion Ai and robot laborers um would it matter that they are could no longer export to

anywhere would it matter maybe it doesn't matter at all so this is the where that you know you have that we have to be escape from the the you know single variable thinking right yeah of course you're right it makes it hard to really think through if one has to keep in mind some infinite number of factors which could change your thinking in any way um I don't know that I really buy The Rosy picture so much though so okay it's true that if China has 10 billion robots uh then it could be fine but you could have said the same thing about India in the industrial revolution right so India a very productive Rich agricultural Nation uh the terms of trade change extremely rapidly because of Britain industrializing you could have said Well India can just build factories and of course they did eventually but that's a kind of very high level observation which in terms of the actual history involved didn't help them one bit I mean very rapidly much more rapidly than they could build those factories they became essentially you know colonized by the East India Company and and then they didn't get to build those factories um so I don't know that I mean some of these factors they're sort of like long-term equilibrium things right where eventually it settles down to behave that way but I don't I'm not sure I would find that comforting if I was a Chinese leader uh while it's true that you know if we get AGI in five years then probably eventually all human beings that survive it uh are going to be very well off assuming the safety is not a problem but in that initial period which could last decades or a hundred years you know it's uh it's very hard to say that just because somebody somewhere has robots and the knowledge is is kind of transferable that uh you'll have robots too so I I guess um it's not clear to me how to identify those fast moving factors versus the slow factors however so I I think the the robotics thing is probably much slower than the other parts of AI as far as I can tell right so it takes time to build factories uh we'll probably have very Advanced AI 10 years before we have you know countries carpeted with three-dimensional factories that go deep into the crust of the earth you know this this will take time probably uh I don't know that I think that the day after human level intelligence is invented a deep mind there will be Nanobots crawling around in the center of the Earth maybe but it seems also reasonable to project that we have automation of cognitive tasks ahead of

um you know in a sort of relatively slow Learners over the period of years which is of course incredibly fast but compared to progress on like purely digital tasks maybe it takes you know a decade to roll out the kind of factories that will completely transform manufacturing so actually what I expect is that the nascent tech industry in China will be wiped out much more much earlier than the manufacturing is um now that's a problem because then you know this is a huge source of jobs for the middle class in China anyway that's kind of China's specific concern but uh I think what I'm referring to with the population from Advanced disadvantages is sort of I mean if you pay attention to what the Chinese leaders are trying to do well obviously they know that they want to educate their population and they spend a lot of effort to do that in order to move up the value chain right to get into software and to get into more advanced Tech and this is a kind of life or death proposition for a country as large as China and as poor as China still is per capita um and what I see is the the sort of core scary thing about uh the AI transition is that essentially that door is going to be shutting their face right I mean it's already pretty clear that you know GitHub co-pilot is still crap in many respects but uh it's on a trajectory to replace a lot of programming jobs in the very near term so that's I think I'm more concerned about that as a source of people as a liability rather than kind of the guys out there building apartment blocks or or making iPhones you know what are some other examples I mean that's one population so that's kind of a maybe an obvious one um but as AI becomes more capable can either of you think of other examples of relative advantages that we wouldn't have to stick to like geopolitical power as the relative quantity that uh we're paying attention to but I think it's an interesting one I guess military equipment could very quickly become obsolete right uh you know once you have very Advanced and fast reacting drones it's not so clear you really need a lot of the other toys that our militaries waste all our tax money um failing to buy I feel like that's probably still a hardware bound it might be slower than the software to be out of the run and by the time you have drones that are very effective in combat you probably also have robots about very effective at manufacturing well I think I think I disagree with that because it's much easier to build things that I mean running around in the

sky is a much simpler thing than walking down a supermarket aisle I think I think sophisticated drones for Warfare will come significantly earlier than household robots for example or manufacturing robots that are very capable that's already the case right I mean the most most uh some of the most advanced robots we have are flying around in the sky um trying to kill people somehow much easier problem than even than sorting I mean sorting stuff in an Amazon Factory is a harder problem than uh shooting somebody from a drone unfortunately yeah one yeah I lost him too Adam we can't hear you yeah I hope you can see the chat nope foreign [Music] on your brain reply because I think um that my understanding is like the more data you have the things like scanning models and stuff like that the more data you have the more AI power China has a big population so maybe they could create large data sets than anyone like yeah that could be a way for them to kind of take lead in the back development yeah this is often touted as one of the advantages China has but I don't buy it I think I mean I think it's clearly get Adam yep we I hear you at least um we were just discussing how in principle China's large population might mean that they can gather large data sets and this is often touted by you know an Eric Schmidt's book or kaifuli who's written about Ai and China quite a bit people are this kind of standard observation about one of the reasons why China might be a a leader in AI um their scale means they they might have big data sets I don't I don't actually buy that uh at all I think it's clearly true that on some problems like identifying cars and video feeds they're motivated to solve them for surveillance reasons and they can collect large data sets and so on but you know we've been in the scaling era for five years now and China has been a follower that entire time if large data sets were going to make a difference it would be in the scaling era and it hasn't made a difference at all I mean they've gathered their own Chinese language data sets like four years behind the ones that open Ai and so on are putting together I just I think it might be true in some limited domains but I don't buy it as a like an overall determining factor and I could be wrong but I think I'm not sold on that yeah I was thinking of it so you mentioned sort of some specifics but I was thinking in terms of damage right you have so much data that you can just generalize to on the data sets um and if you

uh if you got to a point where you're like okay we need more data than exists in the world um so let's let's get uh let's just like take everything that anyone's ever written in a messaging in a message between two people in our country then maybe that's a really big data set that hasn't been used I'm not sure if that has actually been used but um if you have more people they're sending more messages to each other it seems yeah that makes sense what were you trying to say Adam before you cut out oh um well Dan you had mentioned a a couple of minutes ago about the Chinese middle class and Tech and you know stem sort of education and and um training translating into uh good middle class incomes and and basically that that was providing jobs and uh some some robust prosperity for some segment of China and I'm just thinking of just generalizing from that to the entire world um with Western countries the global North in general the middle class and those countries being built around the the professions if you imagine that in air quotes right um but these are sort of the the the the what I have traditionally been regarded as um occupations for knowledge workers and one of the ironies I think I um I wrote an essay about it actually six or eight years ago one of the first ones I put them on my little blog website years ago um where I I doubted that uh there was that there was a what I argued in the in that essay was that the um that Automation and uh what is difficult and thus um um uh compensated economically um are not they're not they're not they aren't very strongly correlated I don't think I think they're probably you know quite orthogonal and the examples are all the professions that are that where where you know you're as a knowledge worker you're actually more vulnerable to automation than a quote unquote unskilled worker um because we don't we haven't we've sort of failed to until very recently um probably the public even still doesn't really fully understand that um uh you know the things that human beings take for granted as being easy easy to do are just formidably difficult um and require a huge amount of intelligence I mean like you said Dan just walking across the room or sorting things in the warehouse requires a huge amount of intelligence compared to things that that you know previously seemed like they were much harder like playing chess right which is easier driving a car or or being a chess Grandmaster whatever you know up until very recently it was pretty obvious to

everybody that playing chess was harder because only a handful of people can do that and everybody can drive how hard is it to drive well it turns out that's completely wrong um and and as you guys know and so uh my concern here is that everywhere across the planet the the uh the the middle class and the professions around which the middle class is built you know not people who own a huge amount of capital and are rent seeking and extremely wealthy as a result of that but people you know who earn a living as knowledge workers reading you know reading x-rays being anesthesiologists doing you know Accounting in people's taxes you know doing the rummy dumb work of of um you know what most lawyers do which is not fireworks in the courtroom but a lot of borings you know uh intellectual grunt work and and um doctors diagnosing things and you know all that sort of stuff uh and then you know pretty much everything in stem apart from actual engineering engineering like getting your hands on things and tinkering and Building Things so much of that is just going to be obliterated pretty quickly right by by AI um I mean the the the the I was just talking to my team today about um Dolly the first one not Dolly too but you know Dolly uh is in is in commercial beta now you can you you're gonna be able to buy access to it right 15 bucks lets you generate 100 or 200 or some odd images with it and so this is going to be an unbelievably Swift disruption of the illustration industry people who draw and illustrate for a living they're gone that that whole that whole that whole career is just going to evaporate over the next two or three years nobody's gonna pay a human being to do that right with a handful of bespoke exceptional artists at the very top of the game but nobody's gonna pay for that service anymore and the thing is nobody's gonna pan a human anesthesia sorry go ahead go ahead yeah have I told you about uh lavender's work collaborator uh and and Dolly no okay yeah this is a perfect example of what you're talking about so she's working on some IP for a Chinese games company and it'll be made into a television show and it's already been made into a a manga a comic um cool and so the the director she's working with who's running that project as a friend of ours from LA uh Chinese very capable Chinese woman and it's a fairly small team and in order just in when they were coming up with ideas for the script she was using uh I think it was you know lavender told her through me about some of these new

pieces of AI where you can generate images or maybe she found out herself it doesn't matter and she was generating some images just to basically just for discussions between her and lavender right where you would sit down you'd talk about some ideas and you could show some images as a kind of stimulus to your creativity and also to present to other members of the team and to the executives and so on and the the images she got back from this generative model was so good uh and she you know she actually downloaded the code and then I mean she was doing a biology PhD before she became like a creative director so she's pretty on board with actually editing the code herself so she downloaded the model and tweaked it a bit and then she started getting out but she really liked and now this is how they're generating images so it's so good that she doesn't she was looking for an artist so this is like a literal example of a displacement so she was looking for an artist to generate these images but she found it so irritating because the artist always wanted to do what they wanted to do not what she wanted to do right they had their own ideas and they would go off and then she'd have to wait until she'd get it back and it's not what she wanted right but with the models you can just keep tweaking the The Prompt right and then like adjust some parameters and and you know she can look at 10 of them and pick one and then look at 10 versions of that and pick one and she doesn't have to wait for some artist to come back and not do exactly what she wants and so now they're not getting an artist she's just going to be generating images and she's on the wait list for Dolly too and when she gets that that's going to be how she makes images and they will make a book where they print out all these images that will be like the art book for this like fairly high profile IP um and if you multiply that by I mean she's maybe a little ahead of other creative directors and being comfortable with that but yeah this is gonna happen so fast people are totally annoying so fast so fast one interesting question is I think with I think if I read it right in the announcement that I saw earlier um open AI is going to own the images that are generated or at least try to claim that right yeah I don't I don't know what that looks like going forward um I mean that's not yeah I I guess if it continues to be to be extremely expensive to train these models then you know maybe um I don't know how you know maybe maybe

they'll end up on the Pirate Bay you know and oh this is open there's open versions of the original Dolly already I I don't I don't see how they can persist with I don't know what the details are around that ownership but if they continue to insist that so people can't use it for other purposes without paying more somebody will just uh open and train their own yeah it's a good point Matt too that they think that's right if they're not safe I might understand yeah I don't know exactly it has been defensive my understanding is that one of the major reasons why um dolly two has been very limited in terms of its who's been able to access it is because it generates it has the potential to generate pornography just you know perfectly more or less perfectly and instantly and you know so extremely offensive potentially um you know if you could instantaneously deepake things and it's yeah it's uh it's it's it's we're gonna run into all these challenges and and awfully bigquery but but at any rate the the idea that the professions are in any way safe at all because it's a human with knowledge and intelligence I think this is very very oh is now having seen what's happened with with you know the latest models this is just spectacularly naive at this point and and uh you know even if it's only only a decade or two until all work is automated you know even if it's relatively Swift and and you know the Tesla bot can come to your house and work on the plumbing um you know presumably that is going to lag a long way behind um the disruption of of people who read medical uh scans for a living and you know make 250 000 a year doing it which is you know this which is what anesthesiologists and technicians and you know other experts in those sorts of areas are are enjoying right now and if what this that begs the question what is how is society going to react I don't see us just taking that line down um probably there's going to be push back to it but I I don't know I don't know how that what form that would take or how it could possibly be successful right so um maybe let's finish with one other example I think might be fun to discuss which is culture so in the Indus I mean the the culture that made you successful as a country was very different in the agricultural era versus the industrial era right so you know Protestant work ethic so-called and these kind of other ideas of populations that are willing to spend all their lives in either the factory or the office and work very hard

and long hours at building capital I mean there's been an immense return on to having that kind of culture over the last 200 years but they may be the cultures that go insane the fastest in the new era right I mean you can see how individuals behave when they're when they have to retire um and they're no longer derived meaning from that word right and and the cultures that that default where individuals Define themselves more by their work and where for example hard work is is seen as a as a an ethic and a moral virtue you know when hard work becomes meaningless because humans don't work humans don't toil and drudge and don't engage in toil and drudgery and and it isn't seen as you know uh meritorious um the Derek cultures that don't celebrate that nearly to the extent that that for example you know the United States or Japan for example quite famously um you know the the the I lived in parts of the world where where that definitely wasn't the case I mean the parts of the Middle East I've lived in were definitely not that way at all and many westerners had this sort of this sort of derogatory very negative perception of the local culture being just lazy and you know unmotivated and blah blah and always and attached all these these this this moral baggage to it this moral unethical um value-laden you know the judgment on that but um you know it there were if you look at it a little I think with a broader perspective you can see there's some merits to to a society that doesn't place as much emphasis on quote unquote hard work and uh defining Yourself by your occupation for example and places more emphasis on Leisure and family and you know socializing and those sorts of things and yeah so this is a really interesting example boy we're going to have a lot of neurotic you know people whose identities uh whose source of identity and source of validation as as hard workers is really going to get exploded right although maybe well maybe we'll just all shift into grinding the next level in World of Warcraft seven you know it'll just just pick up a one-to-one substitution from grinding the social hierarchy to grinding the virtual reward hierarchy episode Black Mirror right oh my God yeah otherwise the treadmills is that the one well yeah I'm probably mixing up a couple of them it's been several years but the you know the the ones where you're grinding up basically points in order to yeah as some sort of currency oh it's miserable the the um uh another example going back to the idea

of stem I mean yes in many countries China and the U.S prominent examples I'm sure it's the same in Australia in many other places there's been a over the last generation there's been a a a a surge in emphasis upon stem Fields as promising places to get training and become an expert as a guarantee or you know to raise your chances of having a middle class upper middle class lifestyle because they're at the profession du jour right they're they're um uh you know they're reliable they have been a reliable secure way to ensure that you're valuable in the workforce over the last you know generation or two 30 30 to 50 years I guess um and so there's been a lot of emphasis on that in our education system but what's ironic about about that is that you aren't trained to necessarily be a a highly social uh uh or generally knowledgeable or sophisticated person with that kind of training you know there this is where more classical education that focuses on Arts liberal arts you know being a classic classical idea of Arts uh Fine Arts which is a little bit different um things like you know philosophy and and so forth if if you're if you if your goal is to become a well-rounded sophisticated person then you know the stem fields are not necessarily what you might choose to focus on in your education um and so if it's possible that you know that because right now we kind of tease people for being all you know you're a philosophy major or you're a you know you're you're an Arts major what are you going to go do with you do with yourself there's they're literal jokes right like what do you say what do you say to a reach recent what what you know what what is it I forget how the joke goes but the punch line is would you like fries with that you know I can't remember the joke but you guys probably know and I'll butcher it if I try to do it but the that's the punch line the idea is that you you graduate with an arts degree and you end up working at McDonald's right but it's possible that that you know that that could have the last laugh here um or the latest laugh maybe is a better way to think about it um because you know stem is going to be worthless sooner than those Arts degrees uh with your arts degree you can you can give a very poetic prompt to GPT that will uh then spit out the business plan for your boss yeah and with an arts degree you can sit in a coffee shop all day and receive your generous Universal basic income check and not go out of your mind with you know neuroticism

or you know I I don't know or you can maybe more swiftly adapt just you know hanging out in in the metaverse or something I I don't know I I it seems a little this is all all quite silly um but uh uh the yeah my overarching concern and I want to I haven't been able to put any real research time into it yet because I haven't been able to get my team to take artificial intelligence seriously enough yet and I want that to change over the course of the next year or so but it's the the the my major concern has always been the transition and maybe we can talk about this next time as if you know maybe we can get into some of the details we've kind of been talking about it here it's easy to it's easy to you know project forward 30 or 50 years to some you know post Singularity Utopia presuming that you know it isn't it's either you know it's either a Utopia or obliteration basically I mean if AI is not aligned you know we're gone uh so presume that it is because it's the only presumption that Affairs thinking about and okay you can fit you can you know pick up nice Fiction and Fantasy about whatever that might look like but the real question is how do we manage to get from here to there without you know just just completely imploding in the process right um that's the real Challenge and it's sort of challenge for for China for example is you know how do they weather this this potentially catastrophic you know apocalyptic storms that they're that they're going to be heading into uh as you know as we shift away from Human labor how are they going to navigate that I mean yeah it could have incredible upsides it could be amazing you don't have to do any you know robots do all the work isn't that what we've all been dreaming about as science fiction fans forever of course it is but getting there is going to be really hard it's going to take a lot of good decisions it's gonna take a lot of you know luck perhaps it can take a lot of good planning um you're probably gonna have to make preparations long ahead of time if we have any chance of of mitigating the worst of the harms that could come so we really do have to get this figured out um yeah and and uh now it's making good points here and we've got a lot of you've got a lot of collective action problems to solve we've got a lot of you know um uh we've got a lot of tough things to start selling the public on I think you know something like a universal basic income or other otherwise Prosperity guaranteed as a right of citizenship is