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.

Automation is changing the way industries operate, allowing machines to take over more and more manual labor. Deep learning algorithms mean machines can process data without human intervention, potentially leading to a state of abundance never seen before. Monsanto is a prime example of this disruption, with their genetically modified crops that are immune to herbicides, increasing productivity but not without controversy. Technology is disrupting labor markets and changing what it means to exchange effort for money, raising questions about the future of work.

Automation and robotics are increasingly being seen as disruptive forces in labor markets, with the potential to replace human work. The potential impact of this is greater than other disruptions such as energy technology, but the journey to a highly automated world is uncertain. Technology has been predicted to end the need for human labor in the past, but the productivity gains from IT between 2000 and 2010 have been slower than expected. This has caused the speaker to think more formally about the potential for automation.

Economists are debating the impact of automation on productivity and labour. Some, such as Kurzweil, believe automation will reduce many white collar jobs, while others, such as Brennolfson and McAfee, think new jobs will be created. Automation raises questions about the future of jobs and employment, as it is expected to replace many human tasks in an economic context. The exchange of money is a mechanism for claiming values in the economy, and the social contract of exchanging effort for money allows people to claim what they need or want. It remains to be seen how automation will affect the labour market and productivity.

Machines have the potential to revolutionize human life by replacing manual labor with automated processes. This could lead to a state of abundance never seen before. Plants already provide the basis of primary production, and in the future, industrial activity may be automated through precise fermentation processes. Deep learning is a system where the machine processes input without the need for human intervention, similar to Tesla's training loop. This could drastically improve efficiency and provide a level of abundance never before seen.

Automation is rapidly changing the way industries operate, with machines performing experiments and assays to design new experiments. This creates a feedback loop managed by people, but increasingly less human input is required. Automation is pushing the human role out of the process, and is happening across multiple levels simultaneously, similar to internet economies of scale where companies change their business models based on what becomes commoditized.

Monsanto has disrupted the agricultural market with their introduction of genetically modified crops that are immune to herbicides. This allows for hyper-productive farming, as farmers can blanket fields with the crops and eliminate weeds and other plants. Monsanto's business model involves owning the IP behind the plants, as well as the Roundup fertilizer they manufacture, allowing them to seize a lot of the value from the operation. This new practice has increased productivity by some measures, although not everyone is happy with this form of agriculture. Derek Thompson, an American writer living in Taiwan, has written about this and other topics related to tech strategy, such as aggregators and the disruption of the old advertising based industries.

Automation is likely to have a major impact on labor markets, with robotics, machine intelligence, and other technologies disrupting the work of humans. This is a bigger change than other disruptions, such as energy technology, and it needs to be addressed. My team has avoided talking about it, but it is becoming increasingly clear that this is a game changer. We may need to develop new tools and approaches to get a handle on it.

The speaker is discussing the potential for automation and robotics to replace human labor. They suggest that it is difficult to predict what the journey to a highly automated world will look like, whether it will follow a s-curve or something different. They also note that technology has been predicted to end the need for humans in labor in the past, but it has never quite happened yet. They suggest that there has been a delay in productivity gains from IT between 2000 and 2010, and they are unsure if this is correct. They are now starting to think more formally about the potential for automation.

There is debate among economists as to whether this time is different in terms of automation and its impact on productivity and labour. Some, like Kurzweil, believe automation will reduce a large number of white collar jobs, while others, such as Brennolfson and McAfee, believe there will be new jobs created. The key question is which forms of activity constitute value creation and which ones should be formalized in the contractual economy. It remains to be seen how automation will affect the labour market and productivity.

Humans do a variety of tasks in an economic context to create value and receive money in exchange. Automation is expected to be a high profile example of this, with autonomous vehicles being a potential 'killer app'. The exchange of money is a mechanism for making claims against other values in the economy, and the social contract of exchanging effort for money allows people to claim what they need or want. Automation raises questions about what will happen to jobs and employment.

Machines have the potential to drastically transform the conditions of human life, from one of scarcity to one of abundance. Through the use of machine labor and the ability of machines to self-power, the process of material provision can become much more efficient. This could lead to a state of abundance that has never been seen before and it is worth exploring how quickly this could happen and what the steps along the way look like.

Plants already provide the basis of true primary production, where they produce carbohydrates and molecules from sunlight and water. In the future, industrial activity may no longer require direct supervision or input from humans, instead taking the form of more automated and precise fermentation processes. This is similar to the way Tesla's training loop works, where a fleet of devices gather data which is processed by the machine and managed by engineers. Deep learning is a next level system, where the machine processes input without the need for human intervention.

Machine learning is becoming increasingly popular in many industries, such as Tesla and protein engineering. This involves machines performing experiments and assays, feeding back into a learning process and designing new experiments. This creates a feedback loop that is managed by people, but increasingly the people are outside of it. An example of this is farms, where plants grow autonomously and are then organised and managed by farmers. Networks process, collect, gather, refine and manufacture the outputs of farms. This system of organisation is becoming more automated, requiring less and less human input.

Automation is pushing the human role to the margins of the process, but this is not necessarily happening sequentially. It could be that automation is surging across multiple levels simultaneously. This could be compared to internet economies of scale and how companies change their business models based on what becomes commoditized. This is because commoditization is hard to arrange without automation. Therefore, automation is pushing the human role out of the process and is happening across multiple levels simultaneously.

Derek Thompson is an American writer living in Taiwan who writes about tech strategy on his blog, Strategy. He is a successor to Clay Christensen and often writes about aggregators as a framework for understanding the internet economy and how Google and Facebook have disrupted the old advertising based industries. He also mentions Monsanto's role in disrupting agriculture, with their new business model and technology such as glyphosate fertilizer, which crops can be bred to be resistant to.

Monsanto's introduction of genetically modified crops that are immune to herbicides allowed for hyper-productive farming, as farmers could blanket fields with the crops and eliminate weeds and other plants. This new practice disrupted the existing agricultural market and created opportunities for new business models. Monsanto's business model involves owning the IP behind the plants, as well as the Roundup fertilizer they manufacture, allowing them to seize a lot of the value from the operation. Although not everyone is happy with this form of agriculture, it has increased productivity by some measures.

the topic that i'd suggested and again this is all quite selfish of me because it's things i'm thinking about at the moment for work um but uh uh i'm interested in trying to understand better um the general picture of what the near to middle term of automation might mean for for labor for labor labor markets i suppose this you could look it through it through it you can look at that to an economic lens um which we often do in our work but it is obviously bigger than that and um i'm i'm wondering if uh if i have the right sort of general take on uh on the on on the technological disruption of labor by the convergence of you know robotics uh machine intelligence you know etc and whether there's whether this is likely to be a you know a disruption that sort of fits into the usual disruption box or whether there's something going on with labor in particular that's that's that's sort of maybe qualitatively different maybe maybe sort of fundamentally different perhaps um anyway i just i'm i'm if it feels as as i think through the implications here a little bit it feels like uh this is a bigger kind of of change a bigger sort of phase change or sea change or something just feels like a bigger deal to me even then um these other quite quite fundamental sectors of the economy energy for example being disrupted by a new energy technology the idea that that that technology could disrupt the work that humans do human labor that seems to me to be something quite really quite extraordinary i don't i mean perhaps even sort of um uh well i don't know about unprecedented but but but certainly in in in in an extreme in the cat in the bucket of of extraordinary as far as disruptions go and um my team has really avoided talking about this or thinking of spending too much time on it but i i'm starting to get the feeling that we can't really dodge this this elephant in the room for that much longer we're going to have to address it because it keeps it keeps creeping into to um other conversations and we keep sort of brushing it aside or brushing it on the under the rug but i think we're i think we have sort of we've done that because my my team and i can can sense that this is a big game changer and um uh uh it it it may involve it may involve um it may involve developing an approach or or tools or something like that that we haven't got yet um to even get a handle on it so anyway um i thought i'd uh take the advantage of this sort of um you know that having some good brains here to um

uh to see what you guys think and um and maybe uh you know get your get your general input your general feel for for this uh you know situation maybe get a sense of what we think a realistic timeline is um what what could what what forms of human labor could plausibly be automated with foreseeable advancements in the state of the art that's relevant so in robotics and in uh uh for lack of a better word the software behind the intelligence that drives um uh machines that can drive automation and um uh and and machine labor in in in general if those two aren't synonymous and um uh and and if we want i mean i don't know how much time we've got today but if we have a um the opportunity maybe today maybe next time as well we could we could we could perhaps start thinking about what the steps between here and a um uh and a a highly automated world and by that i mean a world in which there's there's you know there has been large uh scale widespread deployment of this technology and and and what that you know what what does it i i think the temptation is to look to leap way ahead and say okay what does the world look like when machines are doing all the work etc etc but again the harder part much harder part is what does the what does the journey from here to there look like and is it going to follow something like this you know the s-curve the sigmoids sort of curve that we see in other distractions or something different going to happen here anyway um i thought i would just get this uh you know all of that out on the table and see what we what we make of it um because i'm starting to think about this not just casually but more more formally um yeah because i just i think i think we can i think there's only so long we can ignore this elephant in the room before we're really gonna have to tackle it head-on yeah cool it's an interesting debate there's uh this very active economic literature on both sides of this one i guess saying well people have been saying since the last nights that technology would end the need for humans in labor and it never seems to quite happen and this time will just be like that that's the status quo uh this time isn't different take uh and those people point to the fact that even i'm not sure this is quite correct but i think it's right that you might have expected much higher productivity gains from i.t uh back around 2000 or between 2000 2010 but it took a long time to show up in the productivity figures apparently maybe getting this wrong um

so although the internet and computers definitely did something uh it's not so clear they were as transformative to productivity and that is the ratio of output to input of workers and people point to even though there's been a lot of improvement in the technology in say deep learning or even areas of robotics that are starting to be touched by it people are pointing to the fact that it doesn't seem to show up dramatically in the figures yet there's evidence that they're right that this time is no different and workers will be fine uh on the other side you have well of course there are people like kurzweil and others who would plot us at some point towards the singularity at which stage humans are not much good for anything but within the economics profession i guess two books that come to mind immediately you must be familiar with these atoms is brennolfson i don't know and um andrew mcafee's books mcafee right uh race against the machine and second machine age i thought there was one with a title that had something like light at the end of the tunnel but i couldn't quite find it googling now i thought there was enough an earlier one that i'd looked at before these two but those those people would say something like no this time is different this is [Music] automation is going to reduce a large number of white collar jobs as well as the previous waves of automation dealing with more blue collar jobs and i i don't actually know which one i believe myself i guess um you could always point to the possibility that there will be new jobs right the the characteristics of jobs will just change you know there will be tick tock tutors instead of factory workers so i guess if you imagine a highly automated world are you are you in are we trying to think about a path towards a world where humans really just their labor just isn't useful or that it's just transformed into kinds of labor we want other humans to do for us and not things that we currently think of as the primary activities for human labor yeah i think that's a that's a really key and insightful question right there's it and what it points to is is some set of some nomenclatures some set of definitions about what forms of uh what forms of activity constitute value creation which ones deserve the title labor which ones we can or might or should you know these are all different questions but camera can which ones we could or should um uh formalize within you know you know the contractual economy in other in other words which

ones should we organize jobs around um or occupations or careers or employment what what forms of um activity that are in some way productive in some way you know create value by um uh uh through even through human time and effort um and uh which which tasks are if any are are severable from that and can simply be done by i guess what you might think of as capital in other words what you might think of as as machines or equipment or or um something something maybe i i think that's a reasonably good bucket we'll stick with that one for now um that uh can can produce the again i'm not i don't have a great uh lexicon for all of this stuff yet but they could produce sort of base the basic uh material um goods and perhaps also services that that people require to have some sort some semblance of modern quality of life so i'm thinking um the provision of food of raw materials um there are they take the form of some sort of finished goods that are manufactured um the infrastructure and the provision of water and and other uh you know electricity and other utility services i'm thinking um uh things like transportation uh actually probably that's one of the first ones that will come if autonomous vehicles are sort of an initial very high profile example you know sort of a killer app of um automation uh as as as i think many expect may happen so this is this this is all to say that there's there are a lot of different things that humans do in an economic context that today are jobs that we get paid for there's a contract we put in the time and effort we you know uh do some sort of productive activity some set of tasks uh it rearranges the stuff of the world the materials the information the combination thereof in some sort of useful way and we are you know through a contract we get we are we make an exchange money is a mechanism for the claims that we get to make as a result of that conflict against the other value elsewhere in the system in the economy and so you know we work we get paid and then we can take that money as a as a claim against um other things that we need or want and so on and what i'm wondering i guess is um i suppose my real question is more on the for the right for the time being my interest here and a big question mark hanging over in my mind is is is less on the what happens to jobs and employment and you know um uh uh uh all of those arrangements um sort of the general social contract that you exchange your blood and sweat uh and tears for um money that you can then claim against

other value i'm thinking more about you know if if machines start doing all the stuff that involves taking raw materials combining them with energy and some sort of intelligence and rearranging them in ways that then allow people to survive in in in something some semblance of modern comfort then you know we're still doing a lot of the um the the sim not simple not since i don't know the right way to say it we we i'm trying to avoid things like words like production um but maybe i can't um in other words i can imagine a world a scenario in which uh machines do a lot of the toil and um much of the basis of um of human material needs are met fairly straightforwardly um and that creates that transforms the the the condition in which we have found ourselves throughout human history which is primarily one of scarcity into one of abundance at least in a material sense and then of course you know there are you know there's there will be implications of that socially and and for employment for jobs and then what do people do and can we explore entirely new frontiers for the creation of value with our time and so forth but i'm thinking more in terms of sort of the basic um uh uh simple primary sector sort of sense um that what happens if when machines start doing all of that work and what happens if those things as a result of sort of the way that late that machine labor can scale and recursively sort of um uh supply itself with its own needs you know you can in here you can imagine okay if robots need energy then they can build the solar panels they get the energy and if they need to build more robots and robots can do the building of more robots and obviously they can be trained very quickly because they just it's a download it's not a 20-year education process and so on and so you have this sort of you can imagine sort of an explosion of efficiency uh in which very quickly the entire system of material provision basically becomes self-powering um and and and it's that it seems to me that that could lead to a state of abundance in a way that we have have not seen before and i'm wondering what that how quickly that could happen and what that steps along the way look like and anyway that's that's where the most of my my um uh yeah i think that's my question is if that makes sense yeah i think there's a good area to focus on because it it's uh less subject to drifting off into the never never of uh just speculative sci-fi activity yeah while you were talking i was thinking about

if we just step back a bit and think about from the point of view of the plants on earth we already occupy the very easy part of the production scale where they do all the hard work of you know the producing carbohydrates and useful molecules from just sunlight and water and then that's the that's in a sense the true primary production layer underneath everything right yeah yeah you're right we call that in in ecology and environmental science we call that primary production right i mean it's so i guess i mean if if the farming system for example just becomes much more automated or it's precision fermentation and many other things like that i suppose we just stop thinking about it as actually being industrial activity at some level right if it's sufficiently automated it's i mean it's not like farming is really automated it's not it feels like hard work but it's not it's not that the actual productive process taking place inside the plants requires direct supervision or input from us we we manage the environment and the external conditions to make sure that process continues and that still takes effort but it's uh it's at a different level i guess that's probably a good analogy for what the future industrial system will look like in my view right there will be that's a great analogy i love that i mean there is a sense very much in which in which the the primary production certainly of plants you could even extend it to livestock if you wanted but that's all automated already right i mean at you know at a different level of analysis maybe you you know um but yes it's it's that's a really interesting way to look at it um so it's also i mean that's already an accurate way of thinking about say how the training loop works at tesla right so they have this huge fleet they don't drive the cars around they just drive themselves from tesla's point of view maybe the humans are driving sometimes but anyway the devices are out there moving around collecting data it goes back to the data center and the if you watch uh andre kapathy's talks as i'm sure you have and they talk about the way that learning system is designed it's it's just learning all the time and the the engineers are mostly just managing the conditions of the learning they're tagging certain things as being problematic so the system knows to pay attention to them but in some sense that machine is just sitting there processing input without i mean it's like a next level thing beyond deep learning where deep learning

is you just have the input output mapping you want to learn then the machine goes off and loads it and you're not telling it how to do the task right but there's a kind of industrial scale built on top of that of which i think tesla is a very good example but there will be many many more and in protein engineering this is already the case so there's kind of like a race to make something like the tesla of protein engineering not in the sense of having lots of machines necessarily but in terms of having this feedback loop between machines doing experiments and assays feeding back into a learning process and then designing new experiments and having that loop happen to some extent entirely within the machines and that's that's what it will look like in industry after industry i think i think that's the first step along this path you're pointing at where there will be these huge learning systems that are managed by people but increasingly the people are kind of on the outside and just keeping the thing going a bit like the farmer in the field you know spraying pesticides when necessary building fences and shoveling in horse manure from time to time right i think that's a fantastic analogy a few things that are that are usually a few things that are interesting there to me are that um it may be okay so so i'm so take the example of a of a farm and right now you can think that you can think of the system operating autonomously or elements of the system operating autonomously uh already i'm thinking of like plants growing doing their growing doing the business of just you know producing and um that's sort of at the bottom of some some stack of organization or activity perhaps and above that you you know you so you sort of have plants and then at the next level of organization you have farms and farmers and which organize and manage that activity and then at some higher level of organization or perhaps just a larger physical scale you have networks uh uh that that process and um you know they collect and gather and process and refine um and manufacture various different things from these uh the the outputs of farms and then those are in turn distributed and there's you know there's an even larger system of distributed network you know distribution network and infrastructure and all of that stuff and um one can imagine different elements within that within that you know sort of uh that sort of stack um becoming automated or becoming more automated um so that they require less and less human

input and and we sort of again our you know the role would be pushed the role of curating the process and just occasionally intervening you know and stepping in um but the role you know can push human the human role out to the periphery you know out to the edges and at the margins and so forth but what seems to me to be interesting is it is is is that you know you might be tempted to think that that would happen sequentially you know that would that the order in which that automation might occur would be through that scale so okay we've done the plants first the next thing that's going to happen is the farms and then after that it's going to happen you know up at the you know uh secondary processing facilities and one and then after that it would only happen at the distribution networks and and infrastructure level and out to restaurants and out to supermarkets and whatnot and actually probably not at all right probably it just the automation just just surges across all those levels simultaneously um and uh so so that it it's tempting as using space uh and spatial scale as your sort of as a hierarchical or vertical you know some sort of stack structure to think through that may be the wrong structure the wrong way to think about things in terms of how the you know the sequence uh at which automation rolls out and it may be you know maybe a smart thing to do would be to look um for another schema another way another another um stack of organization that actually will track the sequence more appropriately and what would that be what would the framing of that of that structure be like what would determine that could complexity or or what i or or i don't know i'm not anyway sorry just lurching forward towards one weird idea i guess but um one of the first things in my mind yeah it's interesting it reminds me of the discussions around well internet economies of scale and how companies change their structure between you know the microsoft and facebook are very different organizations and the way they their business models are different and partly that's due to what became commoditized and what wasn't i suppose commoditization is a term that goes hand in hand with automation right it's hard to it's hard for something almost by definition right it's a commodity if you can reproduce it uh at large scale and the units are more or less identical and that at least historically it's hard to arrange without some kind of automation um so yeah i don't know if you know um oh what's

his blog i think it's derek is it derek thompson um derek thompson why is that i mean maybe it's eric i'm not sure he has a great blog about tech strategy it's called strategy okay [Music] he's a i think he's an american rider he's living in taiwan [Music] i think it's just strategy.com it's in i don't know where it comes from the name i guess strategy and tech [Music] right [Music] so he's often a very insightful writer he's kind of a six uh i view him as a successor to uh clay christensen um sure yeah he's very much a disciple of christensen and applying some of that but going beyond it i think an analysis of the modern tech industry and he's often writing about aggregators as as a sort of conceptual framework for talking about the internet economy and how how google and facebook destroyed the old advertising based industries by sort of disaggregating and re-aggregating [Music] i wonder whether that's not a very clearly thought through idea but when you were talking it struck me as that might be a useful mental model of thinking about how well once autumn i mean automation applied to this stack probably leaves the stack completely changed in terms of where the value is and how it's organized right and i guess that's uh this is a question for you i mean have you looked at the disruption of agriculture by monsanto i mean this probably doesn't count as a disruption but it's people certainly worried that ownership of ip around very productive variants of crops would change where the value lay in agriculture and just turn farmers into people who were sharecropping because they had to rely on ip from companies like monsanto i never knew how seriously to take those sorts of descriptions doom and gloom about farming due to monsanto taking over everything it doesn't seem to have happened but yeah that's an interesting example i will um i will check with brad about that that particular example of whether or not he he whether or not he has any data or or knowledge of that case and whether or not it's a disruption or qualifies as a disruption i mean my my knowledge is limited of that of monsen about monsanto's particular role but i do know that that one element that certainly would fit within our framework um for understanding disruption is that it really is a new business model for for sure it's a new technology um glyphosate as a you know as a fertilizer that crops could be bred to be resistant to um and uh bread are designed uh in in later cases but but that at any rate

crop species or cultivars of species could be um could be made to be immune to this to this uh defoliating you know this this herbicide and as a result uh you could be hyper productive because you could just douse you could absolutely blanket um a field of these crops and obliterate every other plant all the weeds and all competing you know um uh plants and and stuff um and uh that was a really new way of of of operating uh you know uh even by monocropping standards a new way of operating a farm and um uh so it certainly was disruptive in in that one sense now whether or not that translates into a a market disruption like was there an s-curve of adoption of that particular industrial agriculture monocropping practice in specific uh for specific commodities like corn and um uh and so forth or uh or is it did that business model just enter into the larger agriculture sort of practices ecosystem worldwide and and not really qualify as a full disruption seizing a great deal of market share and so forth i'll check it out but it's a great example because what it points to is a um uh that that one of the key things that new technologies do is they don't simply um uh cause a one-to-one replacement of an existing tool and no other changes you know cetera's parameters nothing else all else is equal all else remains the same that that is typically not the pattern we see what we see instead is that the markets tend to grow when a new when a disruption occurs because the new technology tends to create new opportunities for to create value new ways of doing things and that's a um you know one one aspect of that is uh new business models and the business model that mancento obviously has been demonized for is you know owning the ip behind the plants that can withstand ground up and um you know i i believe that they did i mean correct me if i'm wrong but i'm i'm associating terminator genes um and the the um the the uh basically the manipulation of of crops so that you can't reseed them once you've got them you have to keep purchasing seed from monsanto as well as the um uh you know the roundup fertilizer that they manufacture and because of that they sort of own your entire um they only i pee around your entire operation and uh so they are able to scrape they're able to to to sort of um seize a lot of the value out of that even though you're much more productive that way i mean by some measures i mean not everybody's obviously happy with with with that form of agriculture but in the