저자: Dr.Know-It-All채널진행자 · 외국인
요약 보기
영상 결론: 진행자는 AI 시장 규모 확대에는 동의하지만 AI 황금기는 끝나가고 있다고 봄. 진행자 기존 주장: 에너지 비용 급증 대비 보상이 줄고, AI 연구의 대중화가 끝나 소수 기업 과점으로 이어져 미래 AI 발전에 문제가 된다고 주장. Cathie Wood(ARK Invest) 주장: AI 학습 비용이 연 68%씩 하락, AI 관련 기업가치 기회를 10년 내 30조 달러에서 80조 달러로 상향. 유전체·로봇·블록체인 등 디플레이션 기술이 중간 유통을 없애 기존 기업에 타격. 바이오 분야: ARK는 R&D 투자 수익률이 20%에서 7%로 떨어졌으나 AI로 임상 실패를 25% 이상 줄이고 출시 기간을 25% 단축할 수 있다고 보며, 화이자·리제네론·버텍스 등을 주목. 진행자 근거: 이미지 분류 비용이 수천 달러에서 3센트로 떨어졌지만 이면의 막대한 연구·인프라 투자가 가려져 있다고 지적. 테슬라 완전자율주행은 소비자가 차량 대금과 주행 데이터로 비용을 분담하는 구조이며, 구글·메타·바이두·텐센트 등이 막대한 자금을 투입 중이라고 설명. 리스크/조건: 연구가 소수 기업(과점)에 집중되면 발견 가능성이 줄고, 소비자 가격 경쟁이 약해져 프라이버시·권리 양보와 빈부 격차가 커진다고 경고. 출연자 전망: 진행자는 자신을 주식 분석가가 아니라고 밝히며 투자 권유를 하지 않음. 최종적으로 Wood와 자신 모두 맞다고 보며 '최고의 시대이자 최악의 시대'가 올 것이라고 결론.
다듬은 글
just a couple of weeks ago i did an episode about how i thought that the current golden age of ai might be on the wane interestingly enough yesterday i saw a video where kathy wood of arc invest said just the opposite so i wanted to do an episode where i examined who might be right about that i guess it's dr know it all versus kathy wood for those of you interested in investing check out weibull an amazing platform for buying and selling stocks and now cryptocurrencies like bitcoin dogecoin and others open an account and get a free stock valued at up to 200 and fund your account and get another free stock valued at up to sixteen hundred dollars check out the link in the description and help the channel at the same time hey y'all it's dr know it all so first of all you should probably start with like watching my episode if you haven't watched it yet i'll do the tldw here really really fast but i go into a lot more detail about the reasons why in that episode so you might want to pause this video watch that one and then come back and watch this one but basically my two contentions were number one that energy costs were getting very very high for the rewards that we were getting and number two and more importantly was that the sort of democratization of ai research was ending and that was a big potential problem for the future of ai so cool that's my stance on the issue what did kathy wood have to say well let's watch a couple of quick video clips artificial intelligence training costs are dropping 68 per year think about that we've never seen anything like that and artificial intelligence is starting really a technology war the companies with the biggest pools of data the highest quality data the best ai expertise with the best domain expertise in the genomic space you know curing disease and here's the incentive you know so the human being has to have has to set the goal and the reward system but the machine's now just training itself and those costs are dropping 68 per year ai is going to be a part of every industry and and that's not the only one dna sequencing is deflationary robotics is deflationary blockchain deflationary it's taking all the middlemen out of the equation massive deflation which is going to hurt companies who haven't spent enough on innovation we were assuming that in the next 10 years artificial intelligence would would deliver in the enterprise software space would deliver a market cap opportunity of 30 trillion dollars okay from from tens of billions today our new number is 80 trillion and from tens of billions today so we think that is the big new frontier but it's going to be in transportation autonomous taxi technology autonomous trucks drones and so forth it is in genomics and so it's going to make these uh platforms come alive and evolve much faster than even we thought was possible right this 30 trillion was a two year ago number now we're up to that's double now we're almost triple we think one of the most interesting opportunities out there that is in the value space is biotech and maybe some pharma those companies that we see uh embracing the convergence between and among dna sequencing artificial intelligence and crispr gene editing gene therapies generally were very interested in because the returns on their r d investment in the last 20 years have dropped this is on average for the industry from 20 to 7 percent we believe with especially artificial intelligence trial failures will uh reduce there will be a reduction by at least 25 could be a lot more the time to market could be 25 less and so we could go back to the golden age now not every not every company is embracing these but those who are and pfizer with some of its move into moves into base editing uh where has is intriguing us here as is regeneron and vertex so as you just heard kathy wood who represents arkhan vest and if you don't know arkhanvest they are they pretty much are they've made the news a lot they talk about disruptive innovation in a lot of different sectors and their one index fund arc q is all about ai as a disruptive influence in the commercial sector so anyway just a few months ago arkhanvest thought that the upside or the potential market cap for the ai industry in in general was about 30 trillion dollars and now they've increased it to 80 trillion dollars which is an amazing amount of money right so if you're thinking about that essentially what kathy wood is saying is that artificial intelligence is going to integrate itself into our lives in so many ways that it's going to become an essential industry and pretty much like you know everything so as people have often said artificial intelligence is going to be the oil of the 21st century so as the oil was to the 20th century right everything kind of grew up around oil and other energy sources left to a lesser degree but certainly oil i guess coal and natural gas as well but that was the main focus of the 20th century it also caused several wars to happen and what they're saying now is that ai artificial intelligence is going to be that same sort of thing in the 21st century so oil and other things are becoming less important of course with the advent of evs that's obvious because you can power evs not just with traditional types of energy but with green energy and so forth so it becomes less and less important to have oil and fossil fuels available more important to start thinking about evs as a green energy source and of course we can also power our grid in that way as well so oil and fossil fuels are on the decline what's taking its place right now is computing power and very specifically machine learning or artificial intelligence so how can the two of us disagree so radically on all of this well my contentions were that we see ai kind of go in cycles since the 1950s with the first you know main golden age of artificial intelligence that people thought at that time sort of like fusion energy they thought in 10 or 20 years that we were going to have robots around that could do the work we could talk to them they would be very very intelligent in the sense that a human being is that did not come to pass then we had another wave you know in the 80s that sort of happened and then it waned as well and you know again they were like predicting that in 10 to 15 years we were going to see these assistants that were essentially human being level intelligence that didn't come to pass and now since the 20 early 2010s or so so about the last decade we've seen a gigantic rise in artificial intelligence and predictions again that we are going to have human level general artificial intelligence within the next you know 10 years or something so the problem with that contention is that we're seeing sort of a law of diminishing returns right now which is that the amount of computing power is getting really really big for less and less gains so at the beginning of this you know thing in like 2011 2012 you see a small increase in computing power and a large increase in the amount of actual work that we can get done and as the decade has gone on it's gone less and less so at least that was my contention and i was saying that we were seeing you know the law of diminishing returns meant that we were going to have a kind of another retrenchment period in artificial intelligence what kathy wood and arkhan vest say instead is that what we're seeing is an exponential decline in the amount of money it costs to do a specific task so for example classifying a billion images has gone down to like three cents as opposed to thousands of dollars a decade ago so right that that kind of decrease is incredible but i think what that covers up is how much compute power is necessary to get to that stage so i guess analogously this would be like saying look i can go out and i can spend a thousand dollars on a new iphone 13 and it only costs a thousand dollars to have this you know incredible cutting edge bionic computing chip inside of it and all of this memory and other things like that and of course you can get other phones for even cheaper than that so you look at that and you say like wow it's incredible how cheap this technology is what you're ignoring is trillions upon trillions upon trillions of dollars that have gone in over the decades into manufacturing into research you know into all of this stuff that's been going on since the 1960s at a major major clip in order to make this happen so there's a little bit of a hidden thing when you say it only costs three cents to classify this many images what you're actually saying here is that it currently costs that little to do that you're not taking account of what's gone before that and you're not taking account of the infrastructure that's had to be built to be able to do this and of course if we look at full self driving which i've done several videos on you can check those out if you're interested but if you look at that as an industry you can see that it is getting really really close to happening but at the same time when people say it's like super close to happening they make it sound like it's sort of come out of thin air there's not a lot of really taking account of the amount of cost involved so what's happening is of course that tesla has geniusly shifted the cost on to us consumers so we're paying 50 or 60 000 for a car in order to give them information that's part of what's going on part of that deal is that we are driving billions of miles collectively and we're giving that to tesla so tesla as opposed to having to pay say it costs them 40 000 to build a car on average right instead of having to put out a fleet of a million cars at forty thousand dollars per car they're actually making a profit to have the exact same thing happen and while that's genius on a commercial level and it's probably going to solve full self-driving i'm convinced that they are the front runner and only real runner in terms of getting full self-driving working at a level four level five type of autonomy it does sort of mask the fact that we're looking at trillions of dollars to make this happen right it's just being diffused into the economy because we consumers are effectively paying for it but we're not paying for all of it a lot of the r d stuff that's going on i mean building the dojo chip building their current super computer cluster all of that kind of stuff cost you know hundreds of millions to billions of dollars i don't know exact figures because of course they don't really reveal that kind of stuff but it's not cheap so all of this stuff costs a ton of money the same sort of thing when you get to like google when they're doing their search and facebook or meta now as they're doing their work and of course baidu and tencent and other companies in china as well so you've got these companies who are investing a fortune like more than kind of almost imaginable amounts of money into this technology to make it look easy and it's not easy and then of course we've got big pharma and as kathy wood talks about she's very excited about pharma and genomics and all of that and so am i quite frankly i don't know enough to be a good investor in it but i'm very excited in the possibility partially just because i'm you know a 56 year old person and that would be nice if i could add a couple of decades to my life due to having some genetic breakthroughs before i get too old for that so that would be really cool so in a selfish way i'm very very into genomics and into that sort of research but again you've got companies that are willing to throw billions of dollars at artificial intelligence because the payoff is so high and absolutely that market that they get so let's say that a company throws two billion dollars at ai for an anti-aging sort of thing and they come up with that and you can take a pill and it takes 10 years off your life how many people are going to be willing anybody basically who is able to afford that pill whatever they charge for it anybody who can afford that is going to buy it and of course that gets into cultural and societal issues of the habs and the have nots right so if you're not wealthy enough to afford to have this pill to make you live longer you live a shorter amount of time and that's a tragic sort of situation but that's you know that's the subject for a whole other video but basically they will make as much money as they want to on this pill if they can make a pill like that and if they can use ai to do that and to make it happen faster all to their benefit right it's an arms race in the genomics and medical industry whoever gets to that point first is going to take the lion's share of the profit just like whoever gets to full self driving first takes the the lion's share of the profit so we're looking at this sort of thing where there's an arms race going on and absolutely i agree that with kathy wood that if these companies are able to do this if they're able to use ai to make full self-driving complete reality if they're able to use ai to get rid of certain types of cancer or to make us live longer etc yes absolutely we could be looking at an 80 trillion dollar industry but i think my points are still quite valid because what we're looking at here is a potential diminishing return but the more important aspect is it's focusing the research into very very narrow silos now right so there's only so many companies like just one or two that can do what tesla's doing in terms of the amount of money that they're willing to spend on full self-driving ai and there's only a few companies that are willing to do this kind of thing in biotech or even have the funds to do that in biotech and universities don't have that kind of funding anymore which is a to a large extent where a lot of this type of research comes from and ideally it would be even cheaper than that where you or i on the street could just buy a computer and we could run some ai stuff and we could be very clever about it and we could come up with solutions that nobody has thought about before so again the more democratized the more people who are playing in this sandbox the better chance you have of making massive breakthroughs and that's the problem and so while i completely agree with kathy wood that we're going to see trillions you know tens of trillions of dollars of market cap get materialized from all of this at the exact same time we're very very much focusing or siloing or pyramiding or whatever it is it's getting to be fewer and fewer players in this market and that leads to an oligopoly and not a monopoly but just a few players right so an oligopoly of players in the market and that reduces the potential for discoveries in the future and so the problem is while we see the commercialization and the market cap go way up and we see the advantages of ai in all of our lives what we're giving up at the same time is the large group of people who are able to play in that field and we're also of course giving up some of our privacy and other rights to these companies at the same time because the fewer of these companies that are able to do this the more money and the more privacy and the more things we're going to have to give up in order to get the benefits of those companies so if you had a hundred companies that saw full self-driving at the same time there would be kind of a race to the bottom in terms of the price for it and so we consumers would win out because somebody would sell it to us for 100 bucks or something for full self-driving if it's just a monopoly if tesla is the only one and who's able to do it they will then license out their hardware and software to other car companies and they will make you know as much money as they want to right now it's ten thousand dollars to purchase full self-driving and it could go up from there so right that's great for the company it's great for us shareholders but it's bad for society so in conclusion i believe that kathy wood and i am both right she's absolutely right that the market cap for artificial intelligence and machine learning is going to skyrocket and it's going to become very very high i also believe that i'm absolutely right that we are seeing the end of a golden age of ai and the end of it again comes from the law of diminishing returns which is you have to throw more and more money to get smaller and smaller results out of it and at the same time that there are fewer and fewer players because of that massive amount of money who are able to play in this sandbox and so while the market cap could take off it's going to be concentrated into very few companies around the world who are going to have all of that money within them it's going to become very concentrated within just a few players around the world and so all of that you know 80 trillion dollars is going to be compressed into a small number of companies not just industries but companies who are able to afford to you know put up the money to be able to make these kinds of things happen and of course we've seen this with pharmaceutical companies overall right as it becomes more and more expensive to make drugs and to find new drugs and to come up with cures for things there are fewer and fewer players and they merge or they're acquired or they go out of business or whatever and the fewer companies there are the more money that those companies can charge when they finally do come up with something so it's a thing where the consumer ends up losing out in the end because there are fewer players out there and there's less competition so again while these big players like google and tesla and apple and meta not facebook etc are out there and are going to be creating money and making more and more money for themselves it's going to be a have and a have not sort of situation so i'm not a stock analyst at all i'm terrible at it believe me as i say on twitter all the time i'm like the worst stock purchaser in the world but again i'm too risk-averse to do this myself sadly enough but what i see if i was more you know if i was willing to take the risk more would be to invest in these companies because what's going to happen is they're going to squeeze out the money from the rest of the world and it's going to get put into these companies so we're going to have these bloated gigantic corporations that are incredibly wealthy because they've sunk the money into this sort of ai universe and we're going to have a lot of other people out there who are losing out because they didn't get involved in that or just simply couldn't afford to do that and so that's a kind of a sad situation it can be very good for the people who are shareholders in these companies but it's going to be a sad situation i think for the world at large so again i think kathy would and i am correct and i think it's going to be the best of times and the worst of times going forward so why not end this episode quoting charles dickens all right i hope you found this episode fun and interesting and thought-provoking if you did definitely like it so other people can find it and of course consider subscribing for more of this kind of stuff as always a huge shout out to my patrons on patreon thank you all so much been wonderful conversations on discord recently if you want to join the club and you want to join the discord channel and everything check out the link in the description and if you're interested in some awesome merch check out the link in the description we've got the teslabot t-shirt which is super popular these days don't mess with tesla lots of other t-shirts tumblers mugs etc so check out the link in the description and help the channel out and for those of you interested in investing check out weibull an amazing platform for buying and selling stocks and now cryptocurrencies like bitcoin dogecoin and others open an account and get a free stock valued at up to 200 and fund your account and get another free stock valued at up to 1600 check out the link in the description and help the channel at the same time thank you and finally don't forget we are both tesla and amazon affiliates if you look in the description you can see how going shopping for a solar roof a wall or anything on amazon helps out the channel and as always feel free to ask me questions in the comments or at my email address which is doctor know it all knows gmail.com until next time bye-bye you
Comments 0