video_id
stringlengths 11
11
| text
stringlengths 361
490
| start_second
int64 0
11.3k
| end_second
int64 18
11.3k
| url
stringlengths 48
52
| title
stringlengths 0
100
| thumbnail
stringlengths 0
52
|
|---|---|---|---|---|---|---|
F0CBkT0UVWE | about sender and the recipient types we can look at some options so we have a sender who can be a resource type sender and we can have a recipient who is a resource type recipient thinking about the previous example they are both message participants so maybe we can have a reused resource type that has okay this is a conversation participants and we can reuse them | 1,643 | 1,669 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1643s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | another thing so we can also remove the recipient because we already know that recipients are all the participants in the conversation so again we can simplify our message resource and look at the conversation resource for uh more information uh talking about the resource types so too many resource types can become a problem uh it can be a maintenance problem it can be a design problem | 1,669 | 1,692 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1669s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | because uh you start to maybe get the track lose the track of which resources uh are uh intended for what so for example we can have a user we can have a buyer we can have a seller then we can send a recipient uh customer uh etc so if you look at all of these resources they can all uh be basically come down to a simple user or a customer or some other solution that | 1,692 | 1,720 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1692s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | works for you so again thank you to account it's it's nice to have a really specific uh resources but when you build your apis it can become overwhelming and it can become a maintenance nightmare uh another use case that we can look about is a search so we have a simple request we have one search endpoint for several resource types so we can uh so let's look at the | 1,720 | 1,748 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1720s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | one solution how we can build our search uh functionality so we have a resource type that search result and then it can have an attribute of title and some excerpt so we can show the matching keywords and relationships is where it gets interesting so let's say we have three main resource types that can be in our search results so we can have a classified you can have article and we | 1,748 | 1,775 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1748s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | have category uh as uh the definition of json api says that you should only serve one resource type with endpoint so we need to have a search result and then using a relationship we can say okay this is a classified this is an article and this is a category and then within the included we can have the complete document uh another alternative for this uh can also be we can reuse | 1,775 | 1,799 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1775s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | a relationship so we can have an attribute with uh title and excerpt and then we can have just one relationship which is called the result and depending on the type of the related uh resource we could know that it's a classified it's an article uh or it's a category so this is another example of how we can do this there are many more examples that i could be showing but given that | 1,799 | 1,830 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1799s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | we cannot be here all day so yeah i've selected a few of more interesting ones and i hope you had a great time some takeaways that i would maybe hope to give you today so content and data we are building apis to provide some data some content naming is a big part of our design as well as thinking about how we will name things and design choice design decisions are | 1,830 | 1,861 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1830s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | something that we do every day so it's not a matter of uh the best design choice it's a matter of is this design choice good for us sometimes there are several design choices that can be made that have a similar list of pros and cons it can be difficult but then it's just a matter of okay pick one and uh continue on uh when i talk about the api design and apis in general so this is | 1,861 | 1,889 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1861s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | this is not the all that it's covering this uh this area there are some additional topics uh that could be covered like how can we do documentation on the api design uh how can we do a testing and validation of our apis versioning security so what kind of security are there not just authorization and authentication but also rate limits uh both protection topics like that there | 1,889 | 1,915 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1889s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | is also caching and backend speed optimizations there is some recipes that we can use to improve our apis like that and also a lot of implementation details on how can we implement our apis so that is it for me uh and we can go over some of the questions uh and the q a uh you're muted dragon hi so let's see if uh we have any questions so far yeah yeah uh between | 1,915 | 1,960 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1915s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | okay okay so this is a question where do you draw the line between okay uh requirement engineering and api is the api design so the api design is uh i would say a result of the requirements uh so uh just like you do uh ux design sometimes the wireframes are part of the requirements uh but it's usually the requirements define what we need to achieve and the api | 1,960 | 1,990 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1960s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | design is part of the technical solution how will we achieve our goal so uh if there is some uh let's put like this one thing that can affect the requirements if there is some design choice that can help us save money time and uh bring better results then we can go back to the requirements and ask for change but usually we have requirements and then we get then we go to the design | 1,990 | 2,013 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=1990s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | stage uh does this answer your question great okay another question are we using swagger to build our api docs uh no right now we are using a postman uh to document uh our apis and we also use it to test so right now this is uh the tool we are using uh we have used swagger to build some apis more specifically the open api specification okay another one is there anything that | 2,013 | 2,047 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2013s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | can help better visualizing the api design yes uh so um i'm thinking about how to answer because there are a couple of ways i can answer this uh so i don't know if there is any uh done tool like out of the box tool but one thing that we have been doing is building a map and this is one thing that json api is really good because uh all the resources can be so | 2,047 | 2,075 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2047s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | basically adjacent api enables us to build a relational relationship uh of a relationship diagram of the whole uh api so yeah we have some tools that we developed to help us like build a uml diagram of all of our resources relationships attributes etc yeah okay a lot of questions uh this is a little bit of topic but uh yeah i know from ansible to uh thinking about it right now this uh out | 2,075 | 2,114 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2075s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | of the box yeah we've used ansible for some stuff we used custom solution for some type we used uh symphony provided tools so uh yeah anything anything that works for you this is this is uh really specific to the technology stack to the solution to the infrastructure so there's a lot of factors i would say to answer this question a bit more fairly yeah uh recommendations on how to return | 2,114 | 2,141 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2114s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | errors to the client and what's the importance of using http state to status codes in the in this regard uh uh short answer json api uh so uh this is again a great thing about the standards that we use uh json api standard which already helps us to describe how we should return errors what are the some of the status codes that are already part of the standards but there are some | 2,141 | 2,168 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2141s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | that we use customly so yeah i would say use it semantically if nobody hates me for saying it like that uh yeah so json api is really good for that uh it helps us a lot with this kind of questions yeah uh include testing in the process um i would say testing is part of the development process uh but again depending on which tools you use uh for the design so for example we use uh | 2,168 | 2,197 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2168s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | postman for the api design itself so when we are done with the design we already have a complete postman collections and we can use it for testing we can also automate it so yeah so we can use it uh we can use it in that way oh i i love this one yeah uh i actually have on one slide uh prepared for this yeah this is this was uh a common common question um so uh the graphql is one of those | 2,197 | 2,225 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2197s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | technologies that we have used uh and um how to say it uh we've used with it uh it's they're pretty simple so despite they're being described as uh data and this is a resource uh the json api for us has a clear conventions which helps us speed up the development process uh it has the right amount of implicit versus explicit so graphql is a great tool if you need a bit more | 2,225 | 2,254 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2225s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | let's say freedom but yeah if you like to design your own stuff but uh if you really want to use some standard json api works just fine and if you compare the two they can provide they provide pretty much the same uh functionality so yeah i i wouldn't say that either of them is uh better it's a personal taste maybe okay regarding content of data how would you rate having values in field that | 2,254 | 2,280 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2254s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | reflect earnings instead of easily readable values so a philosopher philosophical answer here would be uh if you're using a strings or integers they're both atoms so we have a key that means something so if you use two or if you use call they are both kind of m's uh but yeah so uh this is this is an example where it's uh really good to separate the two uh pillars so from the back in | 2,280 | 2,310 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2280s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | perspective you're using number constants from the api perspective you want something that is easily uh used by the client so i would probably most of the time i would prefer uh string so call something that when you look at the api as a per as a human or as a uh some kind of uh program you can you can then translate them into uh into your own values so if you use | 2,310 | 2,333 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2310s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | value two value two is very back in specific usually and if you use a uh call it describes the action so it doesn't matter what constant is behind it so yeah this is this kind of the answer this is a really good example where you think about separating the back end from the api api is built for the consumer yeah hello girl uh how much an api script could be complex how do you snap | 2,333 | 2,358 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2333s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | involves in the performance issue um yeah this is a good question uh in short it so the performance of the json api really uh depends on your implementation because you can implement it to be really fast and we have had some good uh cases where our implementation was really fast but if we sometimes we weren't maybe so oriented on performances and yeah you can really quickly | 2,358 | 2,385 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2358s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | degrade the performance but it's again it depends on your technological stack you can improve it but from our example uh a json api can be really performant so there isn't a lot of problems with getting it to be performed yeah uh have you heard of netjs framework uh i don't i cannot say i've heard it uh it sounds like it's a js framework so maybe i should ask if somebody from my front and | 2,385 | 2,419 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2385s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | colleagues is here to answer uh php typescript uh i mean in the company we so recorder is php based for the backend solutions uh we use symphony but we also have uh build building a uh node.js based uh backend solution so we are using symphony and some node.js solutions in the company uh what is the preferred way to implement ignition api uh the json api way again this is uh this | 2,419 | 2,449 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2419s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | is a good question but we followed the json api specification and they just uh defined there's it basically are two main strategies how you can paginate one is offset limit and the other one is a page size so they the json api offers two ways you can pay in it depending on the prefer preferences of the client can talk about the best practices of api version uh | 2,449 | 2,477 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2449s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | in my personal opinion yeah yeah okay i would maybe call it good practices best practices is uh this is kind of where it gets discussion oh something is better than the other um yeah i could talk about it i actually uh have uh in the in the longer version of this talk i actually mentioned a couple of things about the versioning so you could either a version by headers | 2,477 | 2,498 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2477s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | you can either version by url you can version by uh resource yeah uh so i i would say not in the time that i have right now because it's a it's a really really uh a wide topic and i know there's a lot of options you can go with it okay thank you alan a lot of questions today we come to the end uh yeah just maybe to mention if you have any other questions feel free to contact me contact our | 2,498 | 2,528 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2498s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | company or send me an message on you know and the recording of this talk will be in couple you can come back to via developers website and and find the recording uh back to stephanie if he's still with us if not thanks there okay i'm still alive uh there's another question right on implementing so this is kind of what i mentioned with uh constraints in the in the three uh main | 2,528 | 2,572 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2528s | API Design - Getting Started—Alen Pokos | |
F0CBkT0UVWE | principles uh so if you standards you already have some rules in the place another big thing is uh naming so you if you have a ambiguous language some kind of uh dictionary that you agree with your business which is again something that can help uh and yeah that would be probably the two most important uh ones because if you standards if you use uh business terminology you already have | 2,572 | 2,598 | https://www.youtube.com/watch?v=F0CBkT0UVWE&t=2572s | API Design - Getting Started—Alen Pokos | |
3FIo6evmweo | the following is a conversation with jurgen schmidhuber he's the co-director of a CSA a lab and a co-creator of long short term memory networks LS TMS are used in billions of devices today for speech recognition translation and much more over 30 years he has proposed a lot of interesting out-of-the-box ideas a meta learning adversarial networks computer vision and even a formal theory | 0 | 27 | https://www.youtube.com/watch?v=3FIo6evmweo&t=0s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | of quote creativity curiosity and fun this conversation is part of the MIT course and artificial general intelligence and the artificial intelligence podcast if you enjoy it subscribe on youtube itunes or simply connect with me on twitter at Lex Friedman spelled Fri D and now here's my conversation with jurgen schmidhuber early on you dreamed of AI systems that | 27 | 56 | https://www.youtube.com/watch?v=3FIo6evmweo&t=27s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | self-improve recursively when was that dream born when I was a baby no it's not true I mean it was a teenager and what was the catalyst for that birth what was the thing that first inspired you when I was a boy I'm I was thinking about what to do in my life and then I thought the most exciting thing is to solve the riddles of the universe and and that means you have to become a | 56 | 89 | https://www.youtube.com/watch?v=3FIo6evmweo&t=56s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | physicist however then I realized that there's something even grander you can try to build a machine that isn't really a machine any longer that learns to become a much better physicist than I could ever hope to be and that's how I thought maybe I can multiply my tiny little bit of creativity into infinity but ultimately that creativity will be multiplied to understand the universe | 89 | 118 | https://www.youtube.com/watch?v=3FIo6evmweo&t=89s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | around us that's that's the the curiosity for that mystery that that drove you yes so if you can build a machine that learns to solve more and more complex problems and more and more general problems older then you basically have solved all the problems at least all the solvable problems so how do you think what is the mechanism for that kind of general solver look | 118 | 150 | https://www.youtube.com/watch?v=3FIo6evmweo&t=118s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | like obviously we don't quite yet have one or know how to build one who have ideas and you have had throughout your career several ideas about it so how do you think about that mechanism so in the 80s I thought about how to build this machine that learns to solve all these problems I cannot solve myself and I thought it is clear that has to be a machine that not only learns to solve | 150 | 179 | https://www.youtube.com/watch?v=3FIo6evmweo&t=150s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | this problem here and problem here but it also has to learn to improve the learning algorithm itself so it has to have the learning algorithm in a representation that allows it to inspect it and modify it such that it can come up with a better learning algorithm so I call that meta learning learning to learn and recursive self-improvement that is really the pinnacle of that why you then not only | 179 | 211 | https://www.youtube.com/watch?v=3FIo6evmweo&t=179s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | alarm how to improve on that problem and on that but you also improve the way the machine improves and you also improve the way it improves the way it improves itself and that was my 1987 diploma thesis which was all about that hierarchy of metal or knows that I have no computational limits except for the well known limits that Google identified in 1931 and for the limits our physics | 211 | 245 | https://www.youtube.com/watch?v=3FIo6evmweo&t=211s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | in the recent years meta learning has gained popularity in a in a specific kind of form you've talked about how that's not really meta learning with Newall networks that's more basic transfer learning can you talk about the difference between the big general meta learning and a more narrow sense of meta learning the way it's used today the ways talked about today let's take the | 245 | 271 | https://www.youtube.com/watch?v=3FIo6evmweo&t=245s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | example of a deep neural networks that has learnt to classify images and maybe you have trained that network on 100 different databases of images and now a new database comes along and you want to quickly learn the new thing as well so one simple way of doing that as you take the network which already knows 100 types of databases and then you would just take the top layer of that and you | 271 | 306 | https://www.youtube.com/watch?v=3FIo6evmweo&t=271s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | retrain that using the new label data that you have in the new image database and then it turns out that it really really quickly can learn that to one shot basically because from the first 100 data sets it already has learned so much about about computer vision that it can reuse that and that is then almost good enough to solve the new task except you need a | 306 | 333 | https://www.youtube.com/watch?v=3FIo6evmweo&t=306s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | little bit of adjustment on the top so that is transfer learning and it has been done in principle for many decades people have done similar things for decades meta-learning true mental learning is about having the learning algorithm itself open to introspection by the system that is using it and also open to modification such that the learning system has an opportunity to | 333 | 367 | https://www.youtube.com/watch?v=3FIo6evmweo&t=333s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | modify any part of the learning algorithm and then evaluate the consequences of that modification and then learn from that to create a better learning algorithm and so on recursively so that's a very different animal where you are opening the space of possible learning algorithms to the learning system itself right so you've like in this 2004 paper you described get all | 367 | 400 | https://www.youtube.com/watch?v=3FIo6evmweo&t=367s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | machines and programs that we write themselves yeah right philosophically and even in your paper mathematically these are really compelling ideas but practically do you see these self referential programs being successful in the near term to having an impact where sort of a demonstrates to the world that this direction is a is a good one to pursue in the near term yes we had these | 400 | 430 | https://www.youtube.com/watch?v=3FIo6evmweo&t=400s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | two different types of fundamental research how to build a universal problem solver one basically exploiting [Music] proof search and things like that that you need to come up with asymptotic Liam optimal theoretically optimal self-improvement and problems all of us however one has to admit that through this proof search comes in an additive constant an overhead an additive | 430 | 465 | https://www.youtube.com/watch?v=3FIo6evmweo&t=430s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | overhead that vanishes in comparison to what you have to do to solve large problems however for many of the small problems that we want to solve in our everyday life we cannot ignore this constant overhead and that's why we also have been doing other things non universal things such as recurrent neural networks which are trained by gradient descent and local search | 465 | 496 | https://www.youtube.com/watch?v=3FIo6evmweo&t=465s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | techniques which aren't universal at all which aren't provably optimal at all like the other stuff that we did but which are much more practical as long as we only want to solve the small problems that we are typically trying to solve in this environment here yes so the universal problem solvers like the girdle machine but also Markos who does fastest way of | 496 | 522 | https://www.youtube.com/watch?v=3FIo6evmweo&t=496s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | solving all possible problems which he developed around 2012 - in my lab they are associated with these constant overheads for proof search which guarantee is that the thing that you're doing is optimal for example there is this fastest way of solving all problems with a computable solution which is due to Marcus Marcus jota and to explain what's going on there let's take | 522 | 552 | https://www.youtube.com/watch?v=3FIo6evmweo&t=522s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | traveling salesman problems with traveling salesman problems you have a number of cities in cities and you try to find the shortest path through all these cities without visiting any city twice and nobody know is the fastest way of solving Traveling Salesman problems tsps but let's assume there is a method of solving them within n to the 5 operations where n is the number of | 552 | 587 | https://www.youtube.com/watch?v=3FIo6evmweo&t=552s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | cities then the universal method of Marcus is going to solve the same trolley salesman problem also within n to the 5 steps plus o of 1 plus a constant number of steps that you need for the proof searcher which you need to show that this particular class of problems that Traveling Salesman salesman problems can be solved within a certain time bound within order into the | 587 | 622 | https://www.youtube.com/watch?v=3FIo6evmweo&t=587s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | five steps basically and this additive constant doesn't care for in which means as n is getting larger and larger as you have more and more cities the constant overhead pales in comparison and that means that almost all large problems I solved in the best possible way our way today we already have a universal problem solver like sound however it's not practical because the overhead the | 622 | 654 | https://www.youtube.com/watch?v=3FIo6evmweo&t=622s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | constant overhead is so large that for the small kinds of problems that we want to solve in this little biosphere by the way when you say small you're talking about things that fall within the constraints of our computational systems thinking they can seem quite large to us mere humans right that's right yeah so they seem large and even unsolvable in a practical sense today but they are still | 654 | 681 | https://www.youtube.com/watch?v=3FIo6evmweo&t=654s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | small compared to almost all problems because almost all problems are large problems which are much larger than any constant do you find it useful as a person who is dreamed of creating a general learning system has worked on creating one has done a lot of interesting ideas there to think about P versus NP this formalization of how hard problems are how they scale this kind of | 681 | 712 | https://www.youtube.com/watch?v=3FIo6evmweo&t=681s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | worst-case analysis type of thinking do you find that useful or is it only just a mathematical it's a set of mathematical techniques to give you intuition about what's good and bad mm-hmm so P versus NP that's super interesting from a theoretical point of view and in fact as you are thinking about that problem you can also get inspiration for better practical | 712 | 739 | https://www.youtube.com/watch?v=3FIo6evmweo&t=712s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | problems always on the other hand we have to admit that at the moment as he best practical problem solvers for all kinds of problems that we are now solving through what is called AI at the moment they are not of the kind that is inspired by these questions you know there we are using general-purpose computers such as recurrent neural networks but we have a search technique | 739 | 766 | https://www.youtube.com/watch?v=3FIo6evmweo&t=739s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | which is just local search gradient descent to try to find a program that is running on these recurrent networks such that it can or some interesting problems such as speech recognition machine translation and something like that and there is very little theory behind the best solutions that we have at the moment that can do that do you think that needs to change you think | 766 | 792 | https://www.youtube.com/watch?v=3FIo6evmweo&t=766s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | that world change or can we go can we create a general intelligence systems without ever really proving that that system is intelligent in some kind of mathematical way solving machine translation perfectly or something like that within some kind of syntactic definition of a language or can we just be super impressed by the thing working extremely well and that's sufficient there's an | 792 | 815 | https://www.youtube.com/watch?v=3FIo6evmweo&t=792s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | old saying and I don't know who brought it up first which says there's nothing more practical than a good theory and um yeah and a good theory of problem-solving under limited resources like here in this universe or on this little planet has to take into account these limited resources and so probably that is locking a theory in which is related to what we already have sees a sim totally optimal | 815 | 852 | https://www.youtube.com/watch?v=3FIo6evmweo&t=815s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | comes almost which which tells us what we need in addition to that to come up with a practically optimal problem so long so I believe we will have something like that and maybe just a few little tiny twists unnecessary to to change what we already have to come up with that as well as long as we don't have that we mmm admit that we are taking sub optimal ways and | 852 | 882 | https://www.youtube.com/watch?v=3FIo6evmweo&t=852s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | we can y'all not Verizon long shorter memory for equipped with local search techniques and we are happy that it works better than any competing method but that doesn't mean that we we think we are done you've said that an AGI system will ultimately be a simple one a general intelligent system will ultimately be a simple one maybe a pseudocode of a few lines to be able to | 882 | 910 | https://www.youtube.com/watch?v=3FIo6evmweo&t=882s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | describe it can you talk through your intuition behind this idea why you feel that uh at its core intelligence is a simple algorithm experience tells us that this stuff that works best is really simple so see asymptotic team optimal ways of solving problems if you look at them and just a few lines of code it's really true although they are these amazing properties just a few lines of code then | 910 | 946 | https://www.youtube.com/watch?v=3FIo6evmweo&t=910s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | the most promising and most useful practical things maybe don't have this proof of optimality associated with them however they are so just a few lines of code the most successful mmm we can neural networks you can write them down and five lines of pseudocode that's a beautiful almost poetic idea but what you're describing there is this the lines of pseudocode are sitting on top | 946 | 978 | https://www.youtube.com/watch?v=3FIo6evmweo&t=946s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | of layers and layers abstractions in a sense hmm so you're saying at the very top mmm you'll be a beautifully written sort of algorithm but do you think that there's many layers of abstractions we have to first learn to construct yeah of course we are building on all these great abstractions that people have invented over the millennia such as matrix multiplications | 978 | 1,008 | https://www.youtube.com/watch?v=3FIo6evmweo&t=978s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | and real numbers and basic arithmetic and calculus and derivations of error functions and derivatives of error functions and stuff like that so without that language that greatly simplifies our way our thinking about these problems we couldn't do anything so in that sense as always we are standing on the shoulders of the Giants who in the past simplified the problem | 1,008 | 1,043 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1008s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | of problem solving so much that now we have a chance to do the final step the final step will be a simple one oh if we if you take a step back through all of human civilization in just the universe in check how do you think about evolution and what if creating a universe is required to achieve this final step what if going through the very painful and an inefficient process of evolution is | 1,043 | 1,072 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1043s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | needed to come up with this set of abstractions that ultimately to intelligence do you think there's a shortcut or do you think we have to create something like our universe in order to create something like human level intelligence hmm so far the only example we have is this one this universe and you live you better maybe not but we are part of this whole process right so apparently so it might | 1,072 | 1,108 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1072s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | be the key is that the code that runs the universe as really really simple everything points to that possibility because gravity and other basic forces are really simple laws that can be easily described also in just a few lines of code basically and and then there are these other events that the apparently random events in the history of the universe which as | 1,108 | 1,136 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1108s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | far as we know at the moment don't have a compact code but who knows maybe somebody and the near future is going to figure out the pseudo-random generator which is which is computing whether the measurement of that spin up or down thing here is going to be positive or negative underlying quantum mechanics yes so you ultimately think quantum mechanics is a pseudo-random number | 1,136 | 1,165 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1136s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | generator monistic there's no randomness in our universe does God play dice so a couple of years ago a famous physicist quantum physicist Anton Zeilinger he wrote an essay in nature and it started more or less like that one of the fundamental insights our theme of the 20th century was that the universe is fundamentally random on the quantum level and that whenever you measure spin | 1,165 | 1,206 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1165s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | up or down or something like that a new bit of information enters the history of the universe and while I was reading that I was already typing the responds and they had to publish it because I was right that there's no evidence no physical evidence for that so there's an alternative explanation where everything that we consider random is actually pseudo-random such as the decimal | 1,206 | 1,237 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1206s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | expansion of pi supply is interesting because every three-digit sequence every sequence of three digits appears roughly one in a thousand times and every five digit sequence appears roughly one in ten thousand times what do you really would expect if it was run random but there's a very short algorithm short program that computes all of that so it's extremely compressible and who knows | 1,237 | 1,274 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1237s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | maybe tomorrow somebody some grad student at CERN goes back over all these data points better decay and whatever and figures out oh it's the second billion digits of pi or something like that we don't have any fundamental reason at the moment to believe that this is truly random and not just a deterministic video game if it was a deterministic video game it would be | 1,274 | 1,301 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1274s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | much more beautiful because beauty is simplicity and many of the basic laws of the universe like gravity and the other basic forces are very simple so very short programs can explain what these are doing and and it would be awful and ugly the universe would be ugly the history of the universe would be ugly if for the extra things the random the seemingly random data points that we get | 1,301 | 1,331 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1301s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | all the time that we really need a huge number of extra bits to destroy all these um these extra bits of information so as long as we don't have evidence that there is no short program that computes the entire history of the entire universe we are a scientists compelled to look further for that Swiss program your intuition says there exists a shortest a program that can backtrack | 1,331 | 1,368 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1331s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | to the to the creation of the universe so the shortest path to the creation yes including all the entanglement things and all the spin up-and-down measurements that have been taken place since 13.8 billion years ago and so yeah so we don't have a proof that it is random we don't have a proof of that it is compressible to a short program but as long as we don't have | 1,368 | 1,401 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1368s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | that proof we are obliged as scientists to keep looking for that simple explanation absolutely so you said simplicity is beautiful or beauty is simple either one works but you also work on curiosity discovery you know the romantic notion of randomness of serendipity of being surprised by things that are about you kind of in our poetic notion of reality we think as humans require randomness so | 1,401 | 1,436 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1401s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | you don't find randomness beautiful you use you find simple determinism beautiful yeah okay so why why because the explanation becomes shorter a universe that is compressible to a short program is much more elegant and much more beautiful than another one which needs an almost infinite number of bits to be described as far as we know many things that are happening in this | 1,436 | 1,473 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1436s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | universe are really simple in terms are from short programs that compute gravity and the interaction between elementary particles and so on so all of that seems to be very very simple every electron seems to reuse the same sub program all the time as it is interacting with other elementary particles if we now require an extra Oracle injecting new bits of information all the time for these extra | 1,473 | 1,509 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1473s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | things which are currently no understood such as better decay then the whole description length our data that we can observe out of the history of the universe would become much longer and therefore uglier and uglier again the simplicity is elegant and beautiful all the history of science is a history of compression progress yes so you've described sort of as we build up | 1,509 | 1,547 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1509s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | abstractions and you've talked about the idea of compression how do you see this the history of science the history of humanity our civilization and life on earth as some kind of path towards greater and greater compression what do you mean by there how do you think of that indeed the history of science is a history of compression progress what does that mean hundreds of years ago | 1,547 | 1,575 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1547s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | there was an astronomer whose name was Keppler and he looked at the data points that he got by watching planets move and then he had all these data points and suddenly turnouts that he can greatly compress the data by predicting it through an ellipse law so it turns out that all these data points are more or less on ellipses around the Sun and another guy came along whose name was | 1,575 | 1,608 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1575s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | Newton and before him hook and they said the same thing that is making these planets move like that is what makes the apples fall down and it also holds form stones and for all kinds of other objects and suddenly many many of these compression of these observations became much more compressible because as long as you can predict the next thing given what you have seen so far you can | 1,608 | 1,642 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1608s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | compress it you don't have to store that data extra this is called predict coding and then there was still something wrong with that theory of the universe and you had deviations from these predictions of the theory and 300 years later another guy came along whose name was Einstein and he he was able to explain away all these deviations from the predictions of the old theory | 1,642 | 1,670 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1642s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | through a new theory which was called the general theory of relativity which at first glance looks a little bit more complicated and you have to warp space and time but you can't phrase it within one single sentence which is no matter how fast you accelerate and how fast are hard you decelerate and no matter what is the gravity in your local framework Lightspeed always looks the same and | 1,670 | 1,701 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1670s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | from from that you can calculate all the consequences so it's a very simple thing and it allows you to further compress all the observations because suddenly there are hardly any deviations any longer that you can measure from the predictions of this new theory so all of science is a history of compression progress you never arrive immediately at the shortest explanation of the data but | 1,701 | 1,730 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1701s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | you're making progress whenever you are making progress you have an insight you see all first I needed so many bits of information to describe the data to describe my falling apples my video are falling apples I need so many data so many pixels have to be stored but then suddenly I realize no there is a very simple way of predicting the third frame in the video from the first tool and and | 1,730 | 1,757 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1730s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | maybe not every little detail can be predicted but more or less most of these orange blocks blobs that are coming down they accelerate in the same way which means that I can greatly compress the video and the amount of compression progress that is the depth of the insight that you have at that moment that's the fun that you have the Scientific fun that fun in that | 1,757 | 1,781 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1757s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | discovery and we can build artificial systems that do the same thing they measure the depth of their insights as they are looking at the data which is coming in through their own experiments and we give them a reward an intrinsic reward and proportion to this depth of insight and since they are trying to maximize the rewards they get they are suddenly motivated to come up with new | 1,781 | 1,810 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1781s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | action sequences with new experiments that have the property that the data that is coming in as a consequence are these experiments has the property that they can learn something about see a pattern in there which they hadn't seen yet before so there's an idea of power play you've described a training general problem solver in this kind of way of looking for the unsolved problems | 1,810 | 1,837 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1810s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | yeah can you describe that idea a little further it's another very simple idea so normally what you do in computer science you have you have some guy who gives you a problem and then there is a huge search space of potential solution candidates and you somehow try them out and you have more less sophisticated ways of moving around in that search space until you finally found a solution | 1,837 | 1,869 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1837s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | which you consider satisfactory that's what most of computer science is about power play just goes one little step further and says let's not only search for solutions to a given problem but let's search two pairs of problems and their solutions where the system itself has the opportunity to phrase its own problem so we are looking suddenly at pairs of problems and their solutions or | 1,869 | 1,903 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1869s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | modifications are the problems over that is supposed to generate a solution to that new problem and and this additional degree of freedom allows us to build Korea systems that are like scientists in the sense that they not only try to solve and try to find answers to existing questions no they are also free to impose their own questions so if you want to build an artificial scientist we | 1,903 | 1,935 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1903s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | have to give it that freedom and power play is exactly doing that so that's that's a dimension of freedom that's important to have but how do you are hardly you think that how multi-dimensional and difficult the space of them coming up in your questions is yeah so as as it's one of the things that as human beings we consider to be the thing that makes us special the intelligence that makes us | 1,935 | 1,961 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1935s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | special is that brilliant insight yeah that can create something totally new yes so now let's look at the extreme case let's look at the set of all possible problems that you can formally describe which is infinite which should be the next problem that a scientist or power-play is going to solve well it should be the easiest problem that goes beyond what you already know | 1,961 | 1,997 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1961s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | so it should be the simplest problem that the current problems all of that you have which can already sold 100 problems that he cannot solve yet by just generalizing so it has to be new so it has to require a modification of the problem solver such that the new problem solver can solve this new thing but the old problem solver cannot do it and in addition to that we have to make | 1,997 | 2,024 | https://www.youtube.com/watch?v=3FIo6evmweo&t=1997s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | sure that the problem solver doesn't forget any of the previous solutions right and so by definition power play is now trying always to search and this pair of in in the set of pairs of problems and problems over modifications for a combination that minimize the time to achieve these criteria so as always trying to find the problem which is easiest to add to the repertoire so just | 2,024 | 2,055 | https://www.youtube.com/watch?v=3FIo6evmweo&t=2024s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 | |
3FIo6evmweo | like grad students and academics and researchers can spend the whole career in a local minima hmm stuck trying to come up with interesting questions but ultimately doing very little do you think it's easy well in this approach of looking for the simplest unsolvable problem to get stuck in a local minima is not never really discovering new you know really jumping | 2,055 | 2,081 | https://www.youtube.com/watch?v=3FIo6evmweo&t=2055s | Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs | Lex Fridman Podcast #11 |