AI芯視界

Chat YouTube:
OpenAI 的 CEO 討論了 AI 技術的現狀，並展示了一款用於聊天 GPT 的新應用程序，該應用程序使用 AI 生成圖像和文本。 該應用程序能夠保存信息供以後使用，並且可以與其他應用程序集成。 該技術旨在方便用戶使用，無需明確說明。


Q:用於聊天 GPT 的新應用程序的用途是什麼？
Chat YouTube:
新的聊天應用程序 GPT 的目的是使用 AI 生成圖像和文本，使用戶無需明確指示即可更輕鬆地創建內容。 它還可以保存信息供以後使用，並與其他應用程序集成。


Q:該應用程序如何生成圖像和文本？
Chat YouTube:
該應用程序使用 AI 生成圖像和文本，特別是使用一種稱為 GPT（生成式預訓練轉換器）的 AI 語言模型。 人工智能能夠理解用戶的意圖並以自然語言生成相關響應，並根據提供的提示生成圖像。 該應用程序旨在不斷學習和提高其生成質量更好的圖像和文本的能力。


Q:該技術如何使用戶界面更加人性化？
Chat YouTube:
該技術通過提供統一的語言界面使用戶界面更加人性化，用戶無需了解他們正在使用的各種應用程序的每個細節或菜單選項。 人工智能語言模型理解用戶的自然語言輸入，每次都無需明確指示即可進行選擇。 這節省了用戶瀏覽不同應用程序和工具的時間和精力。 此外，人工智能可以保存信息供以後使用，並與其他應用程序集成，使用戶體驗更加無縫和高效。


0:00 foreign we started open AI seven years ago 0:06 because we felt like something really interesting was happening in Ai and we wanted to help steer it in a positive 0:12 direction it's honestly just really amazing to see how far this whole field has come since 0:19 then and it's really gratifying to hear from people like Raymond who are using 0:24 the technology we are building in others for so many wonderful things we hear from people who are excited we 0:31 hear from people who are concerned we hear from people who feel both those emotions at once and honestly that's how we feel 0:40 above all it feels like we're entering an historic period right now where we as a world are going to define a technology 0:48 that will be so important for our society going forward and I believe that we can manage this 0:55 for good so today I want to show you the current state of that technology and some of the 1:01 underlying design principles that we hold dear 1:09 so the first thing I'm going to show you is what it's like to build a tool for an AI rather than building it for a human 1:17 so we have a new Dolly model which generates images and we are exposing it as an app for chat GPT to use on your 1:24 behalf and you can do things like ask you know suggest a nice 1:30 post Ted meal and draw a picture of it 1:38 now you get all of the sort of ideation and creative back and forth and taking 1:43 care of the details for you that you get out of chat GPT and here we go it's not just the idea for the meal but uh very 1:51 very uh detailed spread um so let's see what we're gonna get but 1:56 chat gbt doesn't just generate images in this case sorry it doesn't need to generate text it also generates an image 2:02 and that is something that really expands the power of what it can do on your behalf in terms of carrying out 2:07 your intent and I'll point out this is all that I've demo this is all generated by the AI as we speak so I actually 2:13 don't even know what we're going to see um this looks wonderful 2:20 thank you now I'm getting hungry just looking at it now we've extended Chachi petite with 2:26 other tools too for example memory you can say save this for later 2:32 um and the interesting thing about these tools is they're very inspectable so you get this little pop-up here that says 2:37 use the dolly app and by the way this is coming to all chat EPT users over upcoming months and you can look under 2:43 the hood and see that what it actually did was write a prompt just like a human could and so you you sort of have this ability 2:50 to inspect how the machine is using these tools which allows us to provide feedback to them now it's saved for 2:56 later and let me show you what it's like to use that information and to integrate with other applications too you can say 3:03 I now make a shopping list for the tasty 3:09 thing I was suggesting earlier and make it a little tricky for the AI 3:16 and tweet it out for all the Ted viewers out there 3:23 so if you do make this wonderful wonderful meal I definitely want to know how it tastes 3:28 but you can see that chat EBT is selecting all these different tools without me having to tell it explicitly 3:34 which ones to use in any situation and this I think shows a new way of 3:39 thinking about the user interface like we are so used to thinking of well we have these apps we cook between them we 3:45 copy paste between them and usually it's a great experience within an app as long as you kind of know the menus and know all the options 3:51 um yes I would like you to yes please always good to be polite 4:00 and by having this unified language interface on top of tools the AI is able 4:07 to sort of take away all those details from you so you don't have to be the one who spells out every single sort of 4:13 little piece of what's supposed to happen and as I said this is a live demo so uh sometimes the unexpected will happen to 4:20 us but let's take a look at the instacart shopping list while we're at it you can see we sent a list of 4:27 ingredients to instacart here's here's everything you need and the thing that's really interesting is 4:33 that the traditional UI is still very valuable right if you look at this uh you still can click through it and sort 4:39 of modify the uh the actual quantities and that's something that I think shows that that they're not going away 4:46 traditional uis it's just we have a new augmented way to build them and now we have a tweet that's been drafted for our 4:52 review which is also a very important thing we can click run and there we are we're the manager we're able to inspect 4:58 we're able to change the the work of the AI if we want to and so I after 5:05 this talk I you will be able to access this yourself 5:17 and there we go cool so thank you everyone 5:23 [Applause] 5:29 so we'll cut back to the slides now the important thing about how we 5:35 build this it's not just about building these tools it's about teaching the AI how to use them like what do we even 5:41 want it to do when we ask these very high level questions and to do this we use an old idea 5:48 if you go back to Alan turing's 1950 paper on the Turing test he says look you'll never program an answer to this 5:53 instead you can learn it you could build a machine like a human child and then teach it through feedback have a human 6:00 teacher who provides rewards and punishments as it tries things out and does things that are either good or bad 6:06 and this is exactly how we train chat GPT it's a two-step process first we produce what touring would have called a child machine through an unsupervised 6:12 learning process we just show it the whole world the whole internet and say predict what comes next in text you've 6:18 never seen before and this process imbues it with all sorts of wonderful skills for example if you're shown a 6:24 math problem the only way to actually complete that math problem say what comes next that green line up there is 6:30 to actually solve the math problem but we actually have to do a second step 6:36 too which is to teach the AI what to do with those skills and for this we provide feedback we have the AI try out 6:41 multiple things give us multiple suggestions and then human rates them says this one's better than that one and this reinforces not just the specific 6:48 thing that the AI said but very importantly the whole process that the AI used to produce that answer and this 6:54 allows it to generalize it allows it to teach to sort of infer your intent and apply it in scenarios that it hasn't 7:00 seen before that it hasn't received feedback now sometimes the things we have to teach the AI are not what you'd expect 7:06 for example when we first showed gpd4 to Khan Academy they said wow this is so great we're going to be able to teach 7:12 students wonderful things only one problem it doesn't double check students math if there's some bad math in there 7:19 it will happily pretend that one plus one equals three and run with it so we had to collect some feedback data 7:25 Sal Khan himself was very kind and offered 20 hours of his own time to provide feedback to the machine alongside our team and over the course 7:33 of a couple months we were able to teach the AI that hey you really should push back on humans in this specific kind of 7:39 scenario and we've we've actually made lots and lots of improvements to uh to to the 7:45 models this way uh and when you push that thumbs down in chat GPT that actually is kind of like sending up a bat signal to our team to say here's an 7:52 area of weakness where you should gather feedback um and so when you do that that's one way that we really listen to our users 7:58 and make sure we're building something that's more useful for everyone now providing high quality feedback is a 8:06 hard thing if you think about asking a kid to clean their room if all you're doing is inspecting the floor you don't know if 8:12 you're just teaching them to stuff all the toys in the closet this is a nice Dolly generated image by 8:17 the way and the same sort of uh reasoning applies to AI as we 8:25 move to harder tasks we will have to scale our ability to provide high quality feedback 8:30 but for this the AI itself is is happy to help it's happy to help us provide 8:36 even better feedback and to scale our ability to supervise the machine as time goes on and let me show you what I mean 8:42 for example you can ask for you know gpd4 question like this of how much time 8:48 passed between the these two foundational logs on uh unsupervised learning and learning from Human 8:53 feedback and the model says two months passed but is it true like these models 8:58 are not 100 reliable um other they're getting better every every time we we provide some feedback 9:04 um but we can actually use the AI to fact check it's and it can actually check its own work you can say fact 9:10 check this for me now in this case I've actually given 9:15 the AI new tool this one is a browsing tool where the model can issue search queries and click into web pages and 9:22 actually writes out its whole Chain of Thought as it does it it says I'm just going to search for this and it actually does the search it then it finds the 9:29 publication date in the search results um it then is issuing another search query it's going to click into the blog 9:34 post and all of this you could do but it's a very tedious task it's not a thing that humans really want 9:40 to do it's much more fun to be in the driver's seat to be in this manager's position where you can if you want triple check the work and outcome 9:47 citations so you can actually go and very easily verify any piece of this whole chain of 9:53 reasoning and it actually turns out two months was wrong two months in one week that was correct 10:01 [Applause] 10:07 and we'll come back to the slide and so thing that's so interesting to me about this whole process is that it's as many 10:14 step collaboration between a human and an AI because a human using this fact checking tool is doing it in order to 10:20 produce data for another AI to become more useful to a human and I think this really shows the shape 10:27 of something that we should expect to be much more common in the future where we have humans and machines kind of very 10:33 carefully and delicately designed in how they fit into a problem and how we want 10:38 to solve that problem we make sure that the humans are providing the management that oversight the feedback and the machines are operating in a way that's 10:45 inspectable and trustworthy and together we're able to actually even create even more stress worthy machines and I think 10:50 that over time if we get this process right we will be able to solve impossible problems 10:55 and to give you a sense of just how impossible I'm talking um I think we're going to be able to 11:01 rethink almost every aspect of how we interact with computers for example think about spreadsheets they've been 11:08 around in some form since you know we'll say 40 years ago with visicalc I don't think they've really changed that much 11:14 in that time and here is a specific spreadsheet of 11:19 all the AI papers on the archive for the past 30 years there's about 167 000 of 11:24 them and you can see their the data right here but let me show you the chat CPT take on how to analyze a data set 11:31 like this 11:37 so we can give chat GPT access to yet another tool this one a python 11:42 interpreter so it's able to run code just like a data scientist would and so you can just literally upload a file and 11:49 ask questions about it and very helpfully you know it knows the name of the file and it's like oh this is CSV comma 11:55 separate value file I'll parse it for you the only information here is the name of the file the column names like 12:02 you saw and then the actual data and from that it's able to infer what these 12:07 columns actually mean like that semantic information wasn't in there it has to sort of put together its World Knowledge 12:13 of knowing that oh yeah archive is the site that people submit papers and therefore that's what these things are 12:18 and these are integer values and so therefore it's a number of authors on the paper like all of that that's work for a human to do and ai's happy to help 12:25 with it now I don't even know what I want to ask so fortunately you can ask the machine 12:31 um can you make some exploratory graphs 12:37 and once again this is a super high level instruction with lots of intent behind it but I don't even know what I 12:42 want and the AI kind of has to infer what I might be interested in and so it comes up with some good ideas I think so 12:47 a histogram the number of authors per paper time series of papers per year word cloud of the paper titles all of 12:53 that I think will be pretty interesting to see and the great thing is it can actually do it here we go nice bell curve you see that 13:00 three is kind of the most common um it's going to then write a it's going to make this nice plot of the papers per 13:07 year something crazy is happening in 2023 though looks like we're on an exponential and it dropped off a cliff what could be going on there and by the 13:14 way all this is python code you can inspect and then we'll see the word cloud and so you can see all these wonderful things 13:20 that appear in these titles but I'm pretty unhappy about this 2023 thing it looks makes this year look 13:26 really bad of course the problem is that the year is not over so I'm going to push back on the machine 13:44 so April 13th was the cutoff date I believe can you use that 13:51 I think a fair projection so we'll see this is a kind of ambitious 13:56 one so you know again I feel like there was 14:03 more I wanted out of the machine here I really wanted it to like notice this thing it's maybe it's a little bit a 14:08 little bit of of an overreach for a two of sort of inferred magically that this is what I wanted but I inject my intent 14:16 I provide this additional piece of of you know sort of guidance and under the 14:21 hood the AI is just writing code again so if you want to inspect what it's doing it's very possible 14:26 and now it does the correct projection 14:35 if you notice it even updates the title I didn't ask for that but it knows what I want 14:41 now we'll cut back to the slide again this slide shows a parable of how I 14:49 think we you know a vision of how we may end up using this technology in the future 14:54 a person brought his very sick dog to the vet Who and the veterinary made a bad call to say let's just wait and see 15:01 and the dog would not be here today had he listened in the meanwhile he provided the blood 15:07 test like the full medical records to gpt4 which said I am not a vet you need to talk to professional here are some 15:14 hypotheses he brought that information to a second vet who used it to save the dog's life 15:21 now these systems they're not perfect you cannot overly rely on them but 15:26 this story I think shows that the human with a medical professional and with 15:33 chat EBT as a brainstorming partner was able to achieve an outcome that would not have happened otherwise I think this 15:38 is something we should all reflect on think about as we consider how to integrate these systems into our world 15:44 and one thing I believe really deeply is that getting AI right is going to require participation from everyone and 15:51 that's for deciding how we want it to slot in that's for setting the rules of the road for what an AI well and won't 15:56 do and if there's one thing to take away from this talk it's that this technology just looks 16:01 different just different from anything people had anticipated and so we all have to become literate and that's 16:06 honestly one of the reasons we release chatgpt together I believe that we can achieve 16:11 the open AI mission of ensuring that artificial general intelligence benefits all of humanity thank you 16:34 Greg wow I mean 16:39 I suspect that within every mind out here uh apart from there's a feeling of 16:45 reeling like I suspect that a very large number of people viewing this you look at that and you think oh my goodness 16:51 pretty much every single thing about the way I work I need to rethink like there's just new possibilities there am 16:57 I right who thinks that they're just they're having to rethink the way that we do things yeah I mean it's it's amazing but it's 17:04 also it's also really scary so let's let's talk Greg let's talk we have absolutely I mean I guess my first 17:09 question actually is just how the hell have you done this you know like open openai has a few hundred employees 17:16 Google has thousands of employees working on artificial intelligence 17:22 why is it you who's come up with this technology that shocked the world yeah well I mean the truth is we're all building on shoulders of giants right 17:29 there's no question if you look at the compute progress the algorithmic progress the data progress all of those are really really industry-wide but I 17:35 think within openai we made a lot of very deliberate choices from the early days and the first one was just to 17:41 confront reality as it lays and you know that we just sort of like thought really hard about like what is it going to take 17:47 to make progress here we tried a lot of things that didn't work so you only see the things that did and I think that the the most important thing has been to get 17:53 teams of people who are very different from each other to work together harmoniously 17:59 can we have the Water by the way I just brought here I think we're going to need it so try it hope you try a dry amount 18:04 off topic um but there's isn't there something also just about the fact that that you 18:11 saw something in these language models that meant that if you continue to 18:16 invest in them and grow them that something at some point might emerge 18:21 yes and I think that I mean honestly I think the story there is is pretty illustrative right I think 18:28 that at a high level deep learning like we always knew that was what we wanted to be was a deep learning lab and exactly how to do it like I think that 18:34 in the early days we didn't know we tried a lot of things and one person was working on training a model to predict 18:41 the next character in in Amazon reviews and he got a result where 18:46 this is a syntactic process you expect you know the model will predict where the commas go where the nouns and verbs 18:52 are but he actually got a state-of-the-art sentiment analysis classifier out of it that this model 18:58 could tell you if a review was positive or negative I mean today we were just like ah come on like anyone can do that 19:03 but this was the first time that you saw this emergence this sort of semantics that emerged from this 19:11 underlying syntactic process and there we knew you got to scale this thing you got to see where it goes so I think this helps explain the the Riddle That 19:18 baffles everyone looking at this because these things are described as prediction machines and yet what we're seeing out 19:24 of them feels it just feels impossible that that could come from a you know prediction machine 19:29 just the stuff you showed us just now and the key idea of emergence is that 19:34 when you get more of a thing suddenly different things emerge it happens all the time that ant colonies single ants run around when you bring enough of them 19:40 together you know you get these ant colonies that have show completely emergent and different Behavior or a 19:45 city where a few houses together it's just houses together but as you grow the number of houses things emerge like 19:52 suburbs and cultural centers and traffic jams um 19:57 give me one moment for you when you saw just something pop that just blew your mind that you just did not see coming 20:03 yeah well uh so if you you can try this in chat apt if you add 40 digit numbers 40 digits 40 digit numbers the model 20:10 will do it which means it's really learned a internal circuit for how to do it and the funny the really interesting 20:16 thing is actually if you have an ad like a 40 digit number plus a 35 digit number it'll often get it wrong 20:22 and so you can see that it's really learning the process but it hasn't fully generalized right it's like you can't 20:28 memorize the 40 Edition table that's more add-ins that aren't the universe so it had to have learned something general but that it hasn't really fully yet 20:35 learned that oh I can like sort of generalize this to adding arbitrary numbers of arbitrary lengths 20:40 so what's happened here is is that you've you've you've allowed it to scale up and look at an incredible number of pieces of text and it is learning things 20:47 that you didn't know that it was going to be capable of learning well yeah and it's it's more nuanced too because so 20:54 one science that we're starting to really get good at is predicting some of these emerging capabilities and to do 20:59 that actually one of the one of the things I think is very undersung in this field is sort of engineering quality like we had to rebuild our entire stack 21:05 and get you know like when you think about building a rocket like you know every tolerance has to be like incredibly tiny same is true in machine 21:12 learning you have to get every single piece of the stack engineered properly and then you can start doing these predictions there are all these 21:18 incredibly smooth scale on curves but I think tell you something deeply fundamental about intelligence if you look at our gpt4 blog post you can you 21:24 can see all these Curves in there and now we're starting to be able to predict so we were able to predict for example the performance on coding problems from 21:32 you know we basically look at some models that are ten thousand times or a thousand times smaller and so there's something about this that is actually 21:39 smooth scaling even though it's still early days so here is one of the big fears then 21:45 that arises from this if it's fundamental to what's happening here that as you scale up things emerge that 21:51 that you you can't you can maybe predict in some level of confidence but they still 21:57 it's capable of surprising you why isn't there just a huge risk of 22:02 something truly terrible emerging well I think all these are questions of degree and scale and timing and I think one 22:09 thing people miss too is sort of the integration with the world is also this like incredibly emergent like sort of 22:14 very powerful thing too and so that's one of the reasons that we think it's so important to deploy incrementally and so 22:19 I think that what we kind of see right now if you look at this talk a lot of what I focus on is providing really high 22:24 quality feedback today the task that we do you can inspect them right that it's very easy to look at that math problem 22:30 and be like no no machine like seven was the correct answer but even summarizing a book 22:36 like that's a hard thing to supervise like how do you know if this book summary is any good you have to read the whole book no one wants to do that 22:44 and so I think that the the important thing will be that we take this step by step and that we say okay as we move on 22:51 to book summaries we have to supervise this task properly we have to build up a track record with these machines that 22:56 they're able to actually carry it up carry out our intent and I think we're going to have to produce even better more efficient sort of more reliable 23:02 ways of scaling this sort of like making the machine be aligned with you so we're going to hear later in this 23:08 session there are critics who say that you know this there's there's no real 23:13 understanding inside the system is it going to always we're never going to know that it's not generating errors 23:20 that it doesn't have common sense and so forth is is it your belief Greg that that that is true at any one moment but 23:27 that the expansion of the scale and the human feedback you know that you talked about is basically going to 23:34 take it on that journey of actually getting to things like truth and wisdom and so forth with a high degree of 23:40 confidence how can you be sure of that yeah well I think that the opening eye I mean the three answers yes I believe 23:45 that is that is where we're headed um and I think that the open AI approach here has always been just like let reality hit you in the face right it's 23:52 like this field is the field of broken promises of all these experts saying X is going to happen why is how it works people have been saying neural Nets 23:59 aren't going to work for 70 years they haven't been right yet they might be right you know maybe 70 years plus one 24:04 or something like that is what you need but I think that our approach has always been you've got to push to the limits of this technology to really see it in 24:10 action because that tells you then oh here's how we can move on to a new paradigm and we just haven't exhausted 24:16 the fruit here I mean it's quite a controversial stance you've taken that the right way to do 24:21 this is to put it out there in public and then harness all this you know instead of just your team giving 24:27 feedback the world is now giving feedback but 24:33 if you know bad things are going to emerge it is out there so so you know 24:38 the original story that I heard on open AI when you were founded as a non-profit well you were there as the great sort of 24:45 check on the big companies doing their unknown possibly evil thing with AI and 24:51 you are going to you were going to build models that sort of um uh you know somehow held them 24:56 accountable and could was capable of slowing the field down if need be or at 25:02 least that's that's kind of what I had and yet what's happened arguably is the opposite that you that your release of 25:09 GPT the special chapter GPT put shock waves through the tech world that now 25:14 Google and meta and so forth are all scrambling to catch up and some of their criticisms have been you are forcing us 25:21 to put this out here without proper guardrails or we die you know how how do 25:26 you like make the case that what you have done is responsible here and not Reckless yeah we think we think about 25:32 these questions all the time like like seriously all the time and I think that that I don't think we're always going to 25:38 get it right um but one thing I think has been incredibly important like from the very beginning when we're thinking about how to build artificial general intelligence 25:44 actually have it benefit all of humanity like how are you supposed to do that right and that the default plan of being like well you built in secret you kind 25:51 of like you know get the super powerful thing and then you like figure out the safety of it and then you push go and you hope you got it right 25:57 like I don't know how to execute that plan okay maybe someone else does but for me that was always terrifying it 26:02 didn't feel right and so I think that that this alternative approach is the only sort of other path that I see which 26:09 is that you do let reality hit you in the face and I think you do give people time to give input you do have well before these machines are perfect before 26:16 they are super powerful that you actually have the ability to see them in action and we've seen it from gpt3 right 26:21 gpd3 we really were afraid that the number one thing people are going to do with it was generate misinformation try 26:27 to tip elections instead the number one thing was generating Viagra spam 26:34 hmm so Viagra spam is bad but there are things that are much worse if Here's a thought experiment for you 26:40 suppose you're you're sitting in a room there's a box on the table you believe that in that box is something that 26:47 there's a very strong chance it's something absolutely glorious it's going to give beautiful you know gifts to your family and and to everyone but there's 26:54 actually also a one percent thing in the small print there that says Pandora and uh there's a chance that 27:02 this actually could unleash unimaginable evils on the world do you open that box well so absolutely 27:09 not I I think I think you don't do it that way um and actually honestly like I'll tell 27:14 you a story uh that I haven't actually told before which is that uh shortly after we started open AI I remember I 27:19 was at I was in Puerto Rico for an AI conference I was sitting in the hotel room just like looking out over this wonderful water all these people having 27:25 a good time and you think about it for a moment like if you could choose for a like you know sort of potentially like 27:31 basically that Pandora's Box to be you know five years away or 500 years away 27:36 which would you pick right and like on the one hand you're like well like you know maybe for you personally it's better to like have it be five years 27:42 away but if it gets to be 500 years away and like people get more time to get it right like which do you pick and like 27:48 you know I just like really felt it in that moment I was like of course you do the 500 years like for real like there's many people like my brother is in the 27:54 military at the time and you're like he puts his life on the line in like a much more real way than like any of us you know typing things in in computers and 28:01 developing this technology um at the time and so yeah like I'm I'm really sold on the you've got to 28:06 approach this right but I don't think that's quite playing the field as it truly lies like if you look at the whole 28:13 history of computing like that I I really mean it when I say that this is a 28:18 industry-wide or even like sort of just almost like a human development of technology-wide shift and the more that 28:24 you sort of don't put together the pieces that are there right we're still 28:29 making to faster computers we're still improving the algorithms like all these things they are happening and if you don't put them together you get an 28:35 overhang which means that if someone does or you know that the moment that someone does manage to connect the 28:41 circuit then you suddenly have this very powerful thing no one's had any time to adjust like who knows what kind of 28:46 safety precautions you get and so I think that that one thing I take away is like even you think about the 28:52 development of other sort of Technologies think about nuclear weapons people talk about being like a zero to one sort of like you know sort of change 28:58 in what humans could do but I actually think that if you look at at capability it's been quite smooth over time and so 29:04 the history I think of every technology we've developed has been you got to do it incrementally and you've got to 29:10 figure out how to manage it for each moment that you're sort of increasing using it so what I'm hearing is that you that the 29:17 model you want us to have is that we have birthed this extraordinary child that may have superpowers that take 29:24 Humanity to a whole new place it is our Collective responsibility to provide the 29:31 guard rails for this this child to collectively teach it to be wise and not to pterosol or down is that basically 29:38 the model I I think it's true and I think it's also important to say this may shift right like we gotta take each 29:44 step as we encounter it and I think it's incredibly important today that we all 29:49 do get literate in this technology figure out how to provide the feedback decide what we want from it and I think 29:55 that my hope is that that will be continued to be the best path but it's so good we're honestly having this debate because we wouldn't otherwise if 30:01 it weren't out there Greg Brockman thank you so much for coming to Tad and blowing our minds thank you appreciate it