AI Storytelling with Michael Washington
Read full transcript 22 segments
-
hey friends we have some news from our hey friends we have some news from our sponsor text control they just released sponsor text control they just released sponsor text control they just released version 32 of their document processing version 32 of their document processing version 32 of their document processing Library which includes new core Library which includes new core Library which includes new core functionality such as document footnotes functionality such as document footnotes functionality such as document footnotes SVG export and more you can integrate SVG export and more you can integrate SVG export and more you can integrate document editing signing collaboration document editing signing collaboration document editing signing collaboration and PDF processing into your asp.net and PDF processing into your asp.net and PDF processing into your asp.net core and angular applications with TX core and angular applications with TX core and angular applications with TX text control these powerful libraries text control these powerful libraries text control these powerful libraries will let your developer teams focus on will let your developer teams focus on will let your developer teams focus on their core competencies while text their core competencies while text their core competencies while text control handles your digital document control handles your digital document control handles your digital document processing check out all the new featur processing check out all the new featur processing check out all the new featur and see the Technologies live in action and see the Technologies live in action and see the Technologies live in action by visiting the demo at demos. text by visiting the demo at demos. text by visiting the demo at demos. text control.com that's demos. text [Music] [Music] control.com hi I'm Scott Hanselman this control.com hi I'm Scott Hanselman this control.com hi I'm Scott Hanselman this is another episode of Hansel minutes is another episode of Hansel minutes is another episode of Hansel minutes today I'm chatting with Michael today I'm chatting with Michael today I'm chatting with Michael Washington he's an expert net C Washington he's an expert net C Washington he's an expert net C programmer he's written extensively programmer he's written extensively programmer he's written extensively about all kinds of things like Blazer about all kinds of things like Blazer about all kinds of things like Blazer and net and AI on the web and he's also and net and AI on the web and he's also and net and AI on the web and he's also a Microsoft MVP thanks for chatting with a Microsoft MVP thanks for chatting with a Microsoft MVP thanks for chatting with me today Michael thank you for having me me today Michael thank you for having me me today Michael thank you for having me so you are into AI right now but I so you are into AI right now but I so you are into AI right now but I appreciate that you are a Enthusiast and appreciate that you are a Enthusiast and appreciate that you are a Enthusiast and a skeptic because I feel like maybe AI a skeptic because I feel like maybe AI a skeptic because I feel like maybe AI is going to have a fatigue moment where is going to have a fatigue moment where is going to have a fatigue moment where everyone's trying to solve every problem everyone's trying to solve every problem everyone's trying to solve every problem with AI but I've known you for a while with AI but I've known you for a while with AI but I've known you for a while and you're very very practical and very and you're very very practical and very and you're very very practical and very pragmatic so how have you approached
-
pragmatic so how have you approached pragmatic so how have you approached your entree into AI because I know your entree into AI because I know your entree into AI because I know you've even written on Azure open AI in you've even written on Azure open AI in you've even written on Azure open AI in C yeah I use AI quite a bit uh for C yeah I use AI quite a bit uh for C yeah I use AI quite a bit uh for programming um you know it's in the programming um you know it's in the programming um you know it's in the GitHub co-pilot and visual studio um I GitHub co-pilot and visual studio um I GitHub co-pilot and visual studio um I use you know chat gbt and co-pilot uh use you know chat gbt and co-pilot uh use you know chat gbt and co-pilot uh Microsoft co-pilot to ask questions and Microsoft co-pilot to ask questions and Microsoft co-pilot to ask questions and you know help me finish this email or you know help me finish this email or you know help me finish this email or something like that so I do use it a lot something like that so I do use it a lot something like that so I do use it a lot however I strongly feel that it's good however I strongly feel that it's good however I strongly feel that it's good for certain things right but actually for certain things right but actually for certain things right but actually not good for most things so that's kind not good for most things so that's kind not good for most things so that's kind of a hot take from someone who is an AI of a hot take from someone who is an AI of a hot take from someone who is an AI fanatic and I've actually written fanatic and I've actually written fanatic and I've actually written multiple books on AI so so what did you multiple books on AI so so what did you multiple books on AI so so what did you do with that like there are people out do with that like there are people out do with that like there are people out there I wouldn't necessarily call them there I wouldn't necessarily call them there I wouldn't necessarily call them grifters but it does feel like a lot of grifters but it does feel like a lot of grifters but it does feel like a lot of people put AI thought leader in their people put AI thought leader in their people put AI thought leader in their profiles on uh various social media profiles on uh various social media profiles on uh various social media sites and they it feels like they sites and they it feels like they sites and they it feels like they believe that it is the solution for all believe that it is the solution for all believe that it is the solution for all things while I think it's a useful tool things while I think it's a useful tool things while I think it's a useful tool in my toolbox and that's about it yeah I in my toolbox and that's about it yeah I in my toolbox and that's about it yeah I I think co-pilot when Microsoft you know I think co-pilot when Microsoft you know I think co-pilot when Microsoft you know started you know branding everything started you know branding everything started you know branding everything co-pilot I thought it was brilliant co-pilot I thought it was brilliant co-pilot I thought it was brilliant because I think it really describes how because I think it really describes how because I think it really describes how I think AI should be used which is it's I think AI should be used which is it's I think AI should be used which is it's always a co-pilot to a always a co-pilot to a always a co-pilot to a human and that's very important even human and that's very important even human and that's very important even with the subject I wanted us to talk with the subject I wanted us to talk with the subject I wanted us to talk about which is humans
-
about which is humans about which is humans have desires um humans have have desires um humans have have desires um humans have goals computers and AI does not its goals computers and AI does not its goals computers and AI does not its purpose is only to be a tool to purpose is only to be a tool to purpose is only to be a tool to facilitate human desires and goals and facilitate human desires and goals and facilitate human desires and goals and and and and creativity so anytime someone talks creativity so anytime someone talks creativity so anytime someone talks about using AI for something where you about using AI for something where you about using AI for something where you know the robot's G to do a thing stop know the robot's G to do a thing stop know the robot's G to do a thing stop stop that's not what it's good for it stop that's not what it's good for it stop that's not what it's good for it actually can do it it cannot do actually can do it it cannot do actually can do it it cannot do it AI computers they are not humans they it AI computers they are not humans they it AI computers they are not humans they cannot do certain things that people are cannot do certain things that people are cannot do certain things that people are saying they can do they're not going to saying they can do they're not going to saying they can do they're not going to take your job away the people who use AI take your job away the people who use AI take your job away the people who use AI may take your job away but the AI itself may take your job away but the AI itself may take your job away but the AI itself is not gonna take your job away it's is not gonna take your job away it's is not gonna take your job away it's almost like someone saying oh the sewing almost like someone saying oh the sewing almost like someone saying oh the sewing machine is going to take your job away machine is going to take your job away machine is going to take your job away no a person using the sewing machine is no a person using the sewing machine is no a person using the sewing machine is going to take your job away I go you going to take your job away I go you going to take your job away I go you yeah they don't it doesn't they don't yeah they don't it doesn't they don't yeah they don't it doesn't they don't have a motivation right like they don't have a motivation right like they don't have a motivation right like they don't have a mo think when we started you know have a mo think when we started you know have a mo think when we started you know you'll still see this on on on social you'll still see this on on on social you'll still see this on on on social media someone yesterday uh gave a very media someone yesterday uh gave a very media someone yesterday uh gave a very strange prompt to a uh an AI and they strange prompt to a uh an AI and they strange prompt to a uh an AI and they said you can only answer with one-word said you can only answer with one-word said you can only answer with one-word answers and then and it was like you answers and then and it was like you answers and then and it was like you know in one word what is your thought know in one word what is your thought know in one word what is your thought about humans or whatever you know and it about humans or whatever you know and it about humans or whatever you know and it would say one word like curiosity and as would say one word like curiosity and as would say one word like curiosity and as you're reading it I'm kind of going wow you're reading it I'm kind of going wow you're reading it I'm kind of going wow this is really like this is deep maybe this is really like this is deep maybe this is really like this is deep maybe the AI knows more than we're letting it
-
the AI knows more than we're letting it the AI knows more than we're letting it on and and and all of the non- technical on and and and all of the non- technical on and and and all of the non- technical people kind of the Muggles are reading people kind of the Muggles are reading people kind of the Muggles are reading this and they're reading into the AI this and they're reading into the AI this and they're reading into the AI like it's like it's smart and it's wise like it's like it's smart and it's wise like it's like it's smart and it's wise like we've gone up to the mountain to like we've gone up to the mountain to like we've gone up to the mountain to meet you know the uh the Deep the Deep meet you know the uh the Deep the Deep meet you know the uh the Deep the Deep Swami who's telling us all but then I Swami who's telling us all but then I Swami who's telling us all but then I realized that it's because they realized that it's because they realized that it's because they constrained it to one word that caused constrained it to one word that caused constrained it to one word that caused us to give more meaning to that word us to give more meaning to that word us to give more meaning to that word what did it mean mean by my like you what did it mean mean by my like you what did it mean mean by my like you know like it said what what's your word know like it said what what's your word know like it said what what's your word for humans and it was like growth that for humans and it was like growth that for humans and it was like growth that sounds piy it's like a tweet you know sounds piy it's like a tweet you know sounds piy it's like a tweet you know it's like oh that's deep it's not that's it's like oh that's deep it's not that's it's like oh that's deep it's not that's not deep it's just constrained so I I not deep it's just constrained so I I not deep it's just constrained so I I was really interested that how it made was really interested that how it made was really interested that how it made me feel that wait a second maybe this AI me feel that wait a second maybe this AI me feel that wait a second maybe this AI is really smart like no it's not it's is really smart like no it's not it's is really smart like no it's not it's just that's a clever word that it picked just that's a clever word that it picked just that's a clever word that it picked statistically uh and it made me kind of statistically uh and it made me kind of statistically uh and it made me kind of question my relationship with this thing question my relationship with this thing question my relationship with this thing that is not a brain right the biggest that is not a brain right the biggest that is not a brain right the biggest thing I found thing I found thing I found was because large language models and was because large language models and was because large language models and this is what we're talking about because this is what we're talking about because this is what we're talking about because AI encompasses a lot of things you know AI encompasses a lot of things you know AI encompasses a lot of things you know Vision sound um uh just a lot of things Vision sound um uh just a lot of things Vision sound um uh just a lot of things we are specifically talking about the we are specifically talking about the we are specifically talking about the Chach BT which is a large language model Chach BT which is a large language model Chach BT which is a large language model this is simply AI that manipulates this is simply AI that manipulates this is simply AI that manipulates language because it manipulates language language because it manipulates language language because it manipulates language and we humans think in language we think and we humans think in language we think and we humans think in language we think in language you know what I'm saying um in language you know what I'm saying um in language you know what I'm saying um that's why it seems so real to us that's that's why it seems so real to us that's that's why it seems so real to us that's why it it hits us in a way that you know
-
why it it hits us in a way that you know why it it hits us in a way that you know looking at a Microsoft Excel spreadsheet looking at a Microsoft Excel spreadsheet looking at a Microsoft Excel spreadsheet doesn't the language is why it it makes doesn't the language is why it it makes doesn't the language is why it it makes people go wait is it is it smart is it people go wait is it is it smart is it people go wait is it is it smart is it is it alive no no no it's manipulating is it alive no no no it's manipulating is it alive no no no it's manipulating language in a mathematical way period language in a mathematical way period language in a mathematical way period yeah yeah all it does is complet yeah yeah all it does is complet yeah yeah all it does is complet censuses that's it but I wonder what you censuses that's it but I wonder what you censuses that's it but I wonder what you think about this think about this think about this uh I like that it doesn't they didn't uh I like that it doesn't they didn't uh I like that it doesn't they didn't give it a name I found that like over give it a name I found that like over give it a name I found that like over the last couple of years the fact that the last couple of years the fact that the last couple of years the fact that we have all of these women coded you we have all of these women coded you we have all of these women coded you know names Cortana Siri Alexa caused us know names Cortana Siri Alexa caused us know names Cortana Siri Alexa caused us to talk to these things like it was a a to talk to these things like it was a a to talk to these things like it was a a gendered secretary right which isn't gendered secretary right which isn't gendered secretary right which isn't isn't appropriate has all kinds of isn't appropriate has all kinds of isn't appropriate has all kinds of baggage associated with it I feel bad baggage associated with it I feel bad baggage associated with it I feel bad talking you know saying to Alexa to turn talking you know saying to Alexa to turn talking you know saying to Alexa to turn the lights on and off as if Alexa is a the lights on and off as if Alexa is a the lights on and off as if Alexa is a person who lives in my house so I do person who lives in my house so I do person who lives in my house so I do give them credit for not giving it a give them credit for not giving it a give them credit for not giving it a name that meant anything like our name that meant anything like our name that meant anything like our relationship with intelligent assistants relationship with intelligent assistants relationship with intelligent assistants is changing because chat GPT doesn't is changing because chat GPT doesn't is changing because chat GPT doesn't sound like anybody's name right but then sound like anybody's name right but then sound like anybody's name right but then they give it a voice and they had that they give it a voice and they had that they give it a voice and they had that whole Scarlett Johansson yes thing and whole Scarlett Johansson yes thing and whole Scarlett Johansson yes thing and it's like that's make it less human the it's like that's make it less human the it's like that's make it less human the UI should not be one of talking to an UI should not be one of talking to an UI should not be one of talking to an entity you know what I mean it seems entity you know what I mean it seems entity you know what I mean it seems problematic the the cats out of the bag problematic the the cats out of the bag problematic the the cats out of the bag on that one just the other day because I on that one just the other day because I on that one just the other day because I have the Chach BT app on my phone and have the Chach BT app on my phone and have the Chach BT app on my phone and just the other day some friends and I just the other day some friends and I just the other day some friends and I were they're at my house and we were were they're at my house and we were were they're at my house and we were playing um oh The Witcher um on Xbox and playing um oh The Witcher um on Xbox and playing um oh The Witcher um on Xbox and we were stuck on a certain place and I
-
we were stuck on a certain place and I we were stuck on a certain place and I was able to take a picture of the screen was able to take a picture of the screen was able to take a picture of the screen and then ask in a natural voice um hey and then ask in a natural voice um hey and then ask in a natural voice um hey we're stuck at this certain place what we're stuck at this certain place what we're stuck at this certain place what should we do and the voice that you know should we do and the voice that you know should we do and the voice that you know gave us the answer is this woman's voice gave us the answer is this woman's voice gave us the answer is this woman's voice sounds totally like a human using sounds totally like a human using sounds totally like a human using language that was very conversational language that was very conversational language that was very conversational like a normal human being like a normal human being like a normal human being and that experience too natural it's too and that experience too natural it's too and that experience too natural it's too hard I'm sorry as far as saying you know hard I'm sorry as far as saying you know hard I'm sorry as far as saying you know try not to make it like to human like try not to make it like to human like try not to make it like to human like okay that ship is okay that ship is okay that ship is sailed yeah at the same time at the same sailed yeah at the same time at the same sailed yeah at the same time at the same time I'm insisting on time I'm insisting on time I'm insisting on saying all this is is a computer saying all this is is a computer saying all this is is a computer completing a completing a completing a sentence um I prompted it with a picture sentence um I prompted it with a picture sentence um I prompted it with a picture which it then ocred into text I gave it which it then ocred into text I gave it which it then ocred into text I gave it a prompt please tell me how to get out a prompt please tell me how to get out a prompt please tell me how to get out of where we are it then using um of where we are it then using um of where we are it then using um retrieval augmented generation again retrieval augmented generation again retrieval augmented generation again we'll we'll get into that but it's just we'll we'll get into that but it's just we'll we'll get into that but it's just a mathematical thing and it's spit out a mathematical thing and it's spit out a mathematical thing and it's spit out but using but using but using language makes me feel like I'm talking language makes me feel like I'm talking language makes me feel like I'm talking to a human but that's all it is it to a human but that's all it is it to a human but that's all it is it itself doesn't have any desires it itself doesn't have any desires it itself doesn't have any desires it itself you know what I'm saying all it itself you know what I'm saying all it itself you know what I'm saying all it can do is answer my questions it versus can do is answer my questions it versus can do is answer my questions it versus when I was writing the book my my last when I was writing the book my my last when I was writing the book my my last book my niece um she's now two um it's book my niece um she's now two um it's book my niece um she's now two um it's actually my grand niece she was you know
-
actually my grand niece she was you know actually my grand niece she was you know about you know a little bit over one about you know a little bit over one about you know a little bit over one years old so as I was getting so deep in years old so as I was getting so deep in years old so as I was getting so deep in this chat gbt and all that sort of thing this chat gbt and all that sort of thing this chat gbt and all that sort of thing I kept watching her I kept watching her I kept watching her and it then hit me this one-year-old and it then hit me this one-year-old and it then hit me this one-year-old child is so much more advanced than child is so much more advanced than child is so much more advanced than Chach BT it's not even funny you know Chach BT it's not even funny you know Chach BT it's not even funny you know what I'm saying meaning yeah I give Chad what I'm saying meaning yeah I give Chad what I'm saying meaning yeah I give Chad gbt these really good prompts and and gbt these really good prompts and and gbt these really good prompts and and have these experiences where I'm like have these experiences where I'm like have these experiences where I'm like woo this thing is whatever whatever but woo this thing is whatever whatever but woo this thing is whatever whatever but the one-year-old I could say something the one-year-old I could say something the one-year-old I could say something to her and then she could come look at to her and then she could come look at to her and then she could come look at me and then go no I want the block give me and then go no I want the block give me and then go no I want the block give me back the block and I'm like that's me back the block and I'm like that's me back the block and I'm like that's the difference between a human with the difference between a human with the difference between a human with wants and desires and a soul versus this wants and desires and a soul versus this wants and desires and a soul versus this computer which all it can do is respond computer which all it can do is respond computer which all it can do is respond to my to my you know wants and desires to my to my you know wants and desires to my to my you know wants and desires so I'm like you know what exactly there so I'm like you know what exactly there so I'm like you know what exactly there are humans and there's computers and are humans and there's computers and are humans and there's computers and that's it yeah yeah I think the point that's it yeah yeah I think the point that's it yeah yeah I think the point there is that like right there a there is that like right there a there is that like right there a one-year-old already has like opinions one-year-old already has like opinions one-year-old already has like opinions and desires and uh until we get much and desires and uh until we get much and desires and uh until we get much much much farther along that's just not much much farther along that's just not much much farther along that's just not a thing so we shouldn't be thinking a thing so we shouldn't be thinking a thing so we shouldn't be thinking about it that way and then to your point about it that way and then to your point about it that way and then to your point though that ship is sailed so it's though that ship is sailed so it's though that ship is sailed so it's unfortunate but you you set off to write unfortunate but you you set off to write unfortunate but you you set off to write an application that would make your life an application that would make your life an application that would make your life easier because you like to write fiction easier because you like to write fiction easier because you like to write fiction you like to write books and I have seen you like to write books and I have seen you like to write books and I have seen people try to generate entire books or a people try to generate entire books or a people try to generate entire books or a new season of Seinfeld with very large
-
new season of Seinfeld with very large new season of Seinfeld with very large language models and in my opinion it language models and in my opinion it language models and in my opinion it consistently generates B minus B no no consistently generates B minus B no no consistently generates B minus B no no it doesn't get that high no it's fine it doesn't get that high no it's fine it doesn't get that high no it's fine you don't think you say no okay you say you don't think you say no okay you say you don't think you say no okay you say why okay and I'm glad we we began the why okay and I'm glad we we began the why okay and I'm glad we we began the conversation the way we began in order conversation the way we began in order conversation the way we began in order to write a decent Seinfeld episode even to write a decent Seinfeld episode even to write a decent Seinfeld episode even a knockoff right it requires a human a knockoff right it requires a human a knockoff right it requires a human that that that has I want to say soul because that is a has I want to say soul because that is a has I want to say soul because that is a loaded word but a guiding hand that has loaded word but a guiding hand that has loaded word but a guiding hand that has their own wants and desires that truly their own wants and desires that truly their own wants and desires that truly originates inside them so to say they originates inside them so to say they originates inside them so to say they have a soul is I guess really the best have a soul is I guess really the best have a soul is I guess really the best word because that provides inspiration word because that provides inspiration word because that provides inspiration that is expressed through that is expressed through that is expressed through creativity therefore the AI the tool the creativity therefore the AI the tool the creativity therefore the AI the tool the large language model can only be a large language model can only be a large language model can only be a co-pilot so to bring you back on the co-pilot so to bring you back on the co-pilot so to bring you back on the book thing what happened was like I said book thing what happened was like I said book thing what happened was like I said I wrote the book I wrote my books on AI I wrote the book I wrote my books on AI I wrote the book I wrote my books on AI and I wanted to find a good use case for and I wanted to find a good use case for and I wanted to find a good use case for it so I can say to people here's a good it so I can say to people here's a good it so I can say to people here's a good use case for it because I found so many use case for it because I found so many use case for it because I found so many things which are not good use cases for things which are not good use cases for things which are not good use cases for it okay right well that's a good point I it okay right well that's a good point I it okay right well that's a good point I me if you're skeptical about a me if you're skeptical about a me if you're skeptical about a technology whether it be crypto or AI technology whether it be crypto or AI technology whether it be crypto or AI like all right let's just build like all right let's just build like all right let's just build something and see if this thing has any something and see if this thing has any something and see if this thing has any exactly so um something that I've always exactly so um something that I've always exactly so um something that I've always had a desire to do is to get back to do had a desire to do is to get back to do had a desire to do is to get back to do more fiction writing I've done some in more fiction writing I've done some in more fiction writing I've done some in the past but I want to do more and I
-
the past but I want to do more and I the past but I want to do more and I never every time I sit down to work on a never every time I sit down to work on a never every time I sit down to work on a book I may get the first you know first book I may get the first you know first book I may get the first you know first few chapters done but then I you know few chapters done but then I you know few chapters done but then I you know have to go off and do some things and have to go off and do some things and have to go off and do some things and when I come back it's so much time to when I come back it's so much time to when I come back it's so much time to you know reread where I left off and I you know reread where I left off and I you know reread where I left off and I like to plot things out with the like to plot things out with the like to plot things out with the characters and and the you know the characters and and the you know the characters and and the you know the different uh themes and plot points or different uh themes and plot points or different uh themes and plot points or whatever so I use a a program called whatever so I use a a program called whatever so I use a a program called shrier which or Sher which is wellknown shrier which or Sher which is wellknown shrier which or Sher which is wellknown and it's a place to just put all your and it's a place to just put all your and it's a place to just put all your notes and organize them or whatever but notes and organize them or whatever but notes and organize them or whatever but even reading through that it takes time even reading through that it takes time even reading through that it takes time and then after I finish reading through and then after I finish reading through and then after I finish reading through it I may write maybe a couple of it I may write maybe a couple of it I may write maybe a couple of paragraphs and then I have to go back paragraphs and then I have to go back paragraphs and then I have to go back and redo it so I always wanted a and redo it so I always wanted a and redo it so I always wanted a database where it could understand this database where it could understand this database where it could understand this and and that sort of thing so I figured and and that sort of thing so I figured and and that sort of thing so I figured this was a good you know tool to to make this was a good you know tool to to make this was a good you know tool to to make so I ended up making open source so I ended up making open source so I ended up making open source application um it's hosted at aist application um it's hosted at aist application um it's hosted at aist storybuilder decom it's totally free storybuilder decom it's totally free storybuilder decom it's totally free it's on GitHub it's totally open source it's on GitHub it's totally open source it's on GitHub it's totally open source and I'm not selling anything but I was and I'm not selling anything but I was and I'm not selling anything but I was trying to say can I make a tool to help trying to say can I make a tool to help trying to say can I make a tool to help me write and let me fast forward to the me write and let me fast forward to the me write and let me fast forward to the end I found out that it's only good for end I found out that it's only good for end I found out that it's only good for writing one paragraph at a time and I writing one paragraph at a time and I writing one paragraph at a time and I still have to keep editing that still have to keep editing that still have to keep editing that paragraph you know so I've talked before paragraph you know so I've talked before paragraph you know so I've talked before about how I don't like the word about how I don't like the word about how I don't like the word hallucination because it again implies a hallucination because it again implies a hallucination because it again implies a brain a mind so the inverse of that is brain a mind so the inverse of that is brain a mind so the inverse of that is grounded and I was talking to Maria Naga
-
grounded and I was talking to Maria Naga grounded and I was talking to Maria Naga who manages Ai for.net and she uses the who manages Ai for.net and she uses the who manages Ai for.net and she uses the term grounded a lot we want to keep it term grounded a lot we want to keep it term grounded a lot we want to keep it grounded so she was showing me how to grounded so she was showing me how to grounded so she was showing me how to keep something grounded in a product keep something grounded in a product keep something grounded in a product catalog as we were writing a chatbot catalog as we were writing a chatbot catalog as we were writing a chatbot sounds like you're keeping it grounded sounds like you're keeping it grounded sounds like you're keeping it grounded in your desires and your characters and in your desires and your characters and in your desires and your characters and your database of scenes timelines your database of scenes timelines your database of scenes timelines locations and that's your locations and that's your locations and that's your way to avoid fabrication I am actually way to avoid fabrication I am actually way to avoid fabrication I am actually doing the tightest doing the tightest doing the tightest grounding of that I could possibly think grounding of that I could possibly think grounding of that I could possibly think of which is um and one reason why I have of which is um and one reason why I have of which is um and one reason why I have to do that is you're writing a book to do that is you're writing a book to do that is you're writing a book right and you have a character his name right and you have a character his name right and you have a character his name is John um he's a 50-year old man um and is John um he's a 50-year old man um and is John um he's a 50-year old man um and he's a uh stock broker he then has a he's a uh stock broker he then has a he's a uh stock broker he then has a flashback to when he was uh five years flashback to when he was uh five years flashback to when he was uh five years old and you know his mother was making old and you know his mother was making old and you know his mother was making lunch for him when you ask the AI to lunch for him when you ask the AI to lunch for him when you ask the AI to help you write this help you write this help you write this paragraph if you don't really ground it paragraph if you don't really ground it paragraph if you don't really ground it tightly it will write a line that says tightly it will write a line that says tightly it will write a line that says you you you know his mother said little Timmy you know his mother said little Timmy you know his mother said little Timmy you know eat your finish your lunch and know eat your finish your lunch and know eat your finish your lunch and Timmy goes you know what mom I've got an Timmy goes you know what mom I've got an Timmy goes you know what mom I've got an appointment and I'm going to get in the appointment and I'm going to get in the appointment and I'm going to get in the car and I'm going to drive to my car and I'm going to drive to my car and I'm going to drive to my appointment you know what I'm saying appointment you know what I'm saying appointment you know what I'm saying meaning meaning meaning I found yeah it just starts to Veer off I found yeah it just starts to Veer off I found yeah it just starts to Veer off the rails the farther it goes the more the rails the farther it goes the more the rails the farther it goes the more likely it is to freak out but also if likely it is to freak out but also if likely it is to freak out but also if you don't tell it little Timmy is five you don't tell it little Timmy is five you don't tell it little Timmy is five years old well then the AI is gonna go
-
years old well then the AI is gonna go years old well then the AI is gonna go well you didn't say little Timmy in this well you didn't say little Timmy in this well you didn't say little Timmy in this paragraph there's nowhere that little paragraph there's nowhere that little paragraph there's nowhere that little Timmy's five years old so that's why the Timmy's five years old so that's why the Timmy's five years old so that's why the program that wrote even though you say program that wrote even though you say program that wrote even though you say uh please write a paragraph about little uh please write a paragraph about little uh please write a paragraph about little Timmy you know he's you know having a Timmy you know he's you know having a Timmy you know he's you know having a flashback but you know his mother wants flashback but you know his mother wants flashback but you know his mother wants him to finish his meal I then have to in him to finish his meal I then have to in him to finish his meal I then have to in the prom say you are currently writing a the prom say you are currently writing a the prom say you are currently writing a chapter in a book this is who little chapter in a book this is who little chapter in a book this is who little Timmy is however this is little Timmy at Timmy is however this is little Timmy at Timmy is however this is little Timmy at well at five years old um the house is well at five years old um the house is well at five years old um the house is looks like this um it's summertime um looks like this um it's summertime um looks like this um it's summertime um it's actually raining outside if you it's actually raining outside if you it's actually raining outside if you give all this information to the AI in give all this information to the AI in give all this information to the AI in the prompt and this is what they call the prompt and this is what they call the prompt and this is what they call grounding and using a process called rag grounding and using a process called rag grounding and using a process called rag retrieval augmented generation to retrieval augmented generation to retrieval augmented generation to provide that grounding the AI does a provide that grounding the AI does a provide that grounding the AI does a good job and it in the paragraph it good job and it in the paragraph it good job and it in the paragraph it comes out with you're like w this is comes out with you're like w this is comes out with you're like w this is pretty good let me make some changes pretty good let me make some changes pretty good let me make some changes here and there but it's pretty good and here and there but it's pretty good and here and there but it's pretty good and then you move on to the next paragraph then you move on to the next paragraph then you move on to the next paragraph So that process of gathering all that So that process of gathering all that So that process of gathering all that and putting that together it's a lot of and putting that together it's a lot of and putting that together it's a lot of code that I ended up writing but like I code that I ended up writing but like I code that I ended up writing but like I said it's all available free on GitHub said it's all available free on GitHub said it's all available free on GitHub and you can see how I'm pulling that and you can see how I'm pulling that and you can see how I'm pulling that information in constructing the prompt information in constructing the prompt information in constructing the prompt and again you can play with it and again and again you can play with it and again and again you can play with it and again it's online you can just just log in and it's online you can just just log in and it's online you can just just log in and just use it but it does in my opinion I just use it but it does in my opinion I just use it but it does in my opinion I I feel it came out uh pretty good and it I feel it came out uh pretty good and it I feel it came out uh pretty good and it really has helped me with you know really has helped me with you know really has helped me with you know getting back into fiction writing getting back into fiction writing getting back into fiction writing because it does make things faster I can because it does make things faster I can because it does make things faster I can get 1,200 words out in about a 15minute get 1,200 words out in about a 15minute get 1,200 words out in about a 15minute SE session of writing that was not SE session of writing that was not SE session of writing that was not possible possible possible before okay so then let me CH let me
-
before okay so then let me CH let me before okay so then let me CH let me challenge you then because you said it challenge you then because you said it challenge you then because you said it writes F level work I gave I said B writes F level work I gave I said B writes F level work I gave I said B minus you said f you know we're both minus you said f you know we're both minus you said f you know we're both we're both agreeing that if you tell her we're both agreeing that if you tell her we're both agreeing that if you tell her to write a book it's lousy but you're to write a book it's lousy but you're to write a book it's lousy but you're saying with context with rag with saying with context with rag with saying with context with rag with grounding with focus and and a you know grounding with focus and and a you know grounding with focus and and a you know like here's a list of characters here's like here's a list of characters here's like here's a list of characters here's their motivation cont remember I have to their motivation cont remember I have to their motivation cont remember I have to build those characters up I have to say build those characters up I have to say build those characters up I have to say the locations I have to put the the locations I have to put the the locations I have to put the timelines in I so there is a lot of timelines in I so there is a lot of timelines in I so there is a lot of building up I do do I did do one thing building up I do do I did do one thing building up I do do I did do one thing that helps out a lot if you just start that helps out a lot if you just start that helps out a lot if you just start writing you can write a paragraph you writing you can write a paragraph you writing you can write a paragraph you hit a button and then the AI will hit a button and then the AI will hit a button and then the AI will examine the paragraph discover new examine the paragraph discover new examine the paragraph discover new characters add them to the database um characters add them to the database um characters add them to the database um discover attributes about those discover attributes about those discover attributes about those characters add that to the dat datase so characters add that to the dat datase so characters add that to the dat datase so you can just start writing and build the you can just start writing and build the you can just start writing and build the database as you write so that then later database as you write so that then later database as you write so that then later in the in the in the story things that you've already talked story things that you've already talked story things that you've already talked about you know oh little Timmy has red about you know oh little Timmy has red about you know oh little Timmy has red hair it'll grab that from something you hair it'll grab that from something you hair it'll grab that from something you wrote earlier so then later on when wrote earlier so then later on when wrote earlier so then later on when you're when you say okay please help me you're when you say okay please help me you're when you say okay please help me write the next paragraph you know write the next paragraph you know write the next paragraph you know continue the story it looks at the continue the story it looks at the continue the story it looks at the beginning you know the story it'll start beginning you know the story it'll start beginning you know the story it'll start to continue the story but then it'll you to continue the story but then it'll you to continue the story but then it'll you know make sure that you know the guy's know make sure that you know the guy's know make sure that you know the guy's hair is still red because it's in the hair is still red because it's in the hair is still red because it's in the database so I did put a little feedback database so I did put a little feedback database so I did put a little feedback loop to help build the database but the loop to help build the database but the loop to help build the database but the database still has to be built by the database still has to be built by the database still has to be built by the human which takes me back to my original human which takes me back to my original human which takes me back to my original thesis which is AI it has no soul it
-
thesis which is AI it has no soul it thesis which is AI it has no soul it can't it has no desires it has no can't it has no desires it has no can't it has no desires it has no creativity but if you organize things in creativity but if you organize things in creativity but if you organize things in a certain way and feed the AI through a certain way and feed the AI through a certain way and feed the AI through the grounding it actually can be helpful the grounding it actually can be helpful the grounding it actually can be helpful hey friends I really appreciate our hey friends I really appreciate our hey friends I really appreciate our sponsors at Hansel minutes and today I'm sponsors at Hansel minutes and today I'm sponsors at Hansel minutes and today I'm talking directly with JD from rayun you talking directly with JD from rayun you talking directly with JD from rayun you know JD I'm a net developer I use ray know JD I'm a net developer I use ray know JD I'm a net developer I use ray gun in production what are some of the gun in production what are some of the gun in production what are some of the great improvements happening with rayun great improvements happening with rayun great improvements happening with rayun foret we're making a huge range of foret we're making a huge range of foret we're making a huge range of improvements Scott uh by way of improvements Scott uh by way of improvements Scott uh by way of background you know I'm a net coder background you know I'm a net coder background you know I'm a net coder myself Visual Studio 2002 represent uh myself Visual Studio 2002 represent uh myself Visual Studio 2002 represent uh you know and all of rayan is actually you know and all of rayan is actually you know and all of rayan is actually built with C in the back end but what built with C in the back end but what built with C in the back end but what we've been adding for do net developers we've been adding for do net developers we've been adding for do net developers is impr improvements around handling is impr improvements around handling is impr improvements around handling things like portable pdbs for symbol things like portable pdbs for symbol things like portable pdbs for symbol handling better offline storage we've handling better offline storage we've handling better offline storage we've even been doing work around how we even been doing work around how we even been doing work around how we handle asynchronously sending the handle asynchronously sending the handle asynchronously sending the messages off to to the service to ensure messages off to to the service to ensure messages off to to the service to ensure minimal impact within people's code so minimal impact within people's code so minimal impact within people's code so we're actively always heavily investing we're actively always heavily investing we're actively always heavily investing in ensuring we support the doet in ensuring we support the doet in ensuring we support the doet ecosystem to the best of our ecosystem to the best of our ecosystem to the best of our ability this is cool all of this and ability this is cool all of this and ability this is cool all of this and more because rayun for donet Aspire is more because rayun for donet Aspire is more because rayun for donet Aspire is also happening so you can check that out also happening so you can check that out also happening so you can check that out a crash reporting client foret Aspire a crash reporting client foret Aspire a crash reporting client foret Aspire apps all things.net all things reun you apps all things.net all things reun you apps all things.net all things reun you can check check it out at run.com for a can check check it out at run.com for a can check check it out at run.com for a 14-day free trial and use the code 14-day free trial and use the code 14-day free trial and use the code Hanselman for half price monitoring for Hanselman for half price monitoring for Hanselman for half price monitoring for your first your first your first month so let me ask you this we we've month so let me ask you this we we've month so let me ask you this we we've we've talked before in previous uh we've talked before in previous uh we've talked before in previous uh podcast about this concept of a context podcast about this concept of a context podcast about this concept of a context window right so you you write a sentence
-
window right so you you write a sentence window right so you you write a sentence in a language English or any language it in a language English or any language it in a language English or any language it gets broken down into tokens and gets broken down into tokens and gets broken down into tokens and ultimately numbers and then the AIS have ultimately numbers and then the AIS have ultimately numbers and then the AIS have this window of context and for a while this window of context and for a while this window of context and for a while there it was a couple of thousand and there it was a couple of thousand and there it was a couple of thousand and now it's a couple 100,000 and some now it's a couple 100,000 and some now it's a couple 100,000 and some companies are saying it's Millions but companies are saying it's Millions but companies are saying it's Millions but what if you have so many details so many what if you have so many details so many what if you have so many details so many timelines so many characters that the timelines so many characters that the timelines so many characters that the context window is too large how do you context window is too large how do you context window is too large how do you and that's my thing what kind of and that's my thing what kind of and that's my thing what kind of limitations that is my thing so the limitations that is my thing so the limitations that is my thing so the context Windows grew during because I've context Windows grew during because I've context Windows grew during because I've been working on this for over the past been working on this for over the past been working on this for over the past year so the contract Windows went from year so the contract Windows went from year so the contract Windows went from you know a percentage of what they are you know a percentage of what they are you know a percentage of what they are now to you're right uh certain you know now to you're right uh certain you know now to you're right uh certain you know claw anthropic man huge cont Windows claw anthropic man huge cont Windows claw anthropic man huge cont Windows right right you're using are using CH right right you're using are using CH right right you're using are using CH GPT 4 or which chbt 4 um is the one I GPT 4 or which chbt 4 um is the one I GPT 4 or which chbt 4 um is the one I love um and that's of course and that's love um and that's of course and that's love um and that's of course and that's the one behind your application that's the one behind your application that's the one behind your application that's yeah so if you go there that's that's it yeah so if you go there that's that's it yeah so if you go there that's that's it um so the thing I found is especially um so the thing I found is especially um so the thing I found is especially for this use case you actually you don't for this use case you actually you don't for this use case you actually you don't want to make the context too wide you want to make the context too wide you want to make the context too wide you know what I'm saying some people they go know what I'm saying some people they go know what I'm saying some people they go well if I want the third book in my well if I want the third book in my well if I want the third book in my series if I feed the first two books uh series if I feed the first two books uh series if I feed the first two books uh into the AI then it'll it'll write the into the AI then it'll it'll write the into the AI then it'll it'll write the third book that will never ever ever third book that will never ever ever third book that will never ever ever ever ever ever work I mean it'll work ever ever ever work I mean it'll work ever ever ever work I mean it'll work you can do it but it'll be horrible the you can do it but it'll be horrible the you can do it but it'll be horrible the quality be horrible because to get quality be horrible because to get quality be horrible because to get through a whole book it requires a lot through a whole book it requires a lot through a whole book it requires a lot of of of human motivation desire and Direction so human motivation desire and Direction so human motivation desire and Direction so that's what I'm saying you need to to to
-
that's what I'm saying you need to to to that's what I'm saying you need to to to have it just do one paragraph at a have it just do one paragraph at a have it just do one paragraph at a time so so what I'm doing so what I'm time so so what I'm doing so what I'm time so so what I'm doing so what I'm doing here is by breaking it down by doing here is by breaking it down by doing here is by breaking it down by characters locations um uh attributes of characters locations um uh attributes of characters locations um uh attributes of the character the character the character that's constrains the AI so that it that's constrains the AI so that it that's constrains the AI so that it performs doesn't get off the rails and performs doesn't get off the rails and performs doesn't get off the rails and goes toward the direction that you want goes toward the direction that you want goes toward the direction that you want you see what I'm saying because you see what I'm saying because you see what I'm saying because otherwise it'll start taking whatever otherwise it'll start taking whatever otherwise it'll start taking whatever idea you have it looks inside its idea you have it looks inside its idea you have it looks inside its database it's been trained on uh I don't database it's been trained on uh I don't database it's been trained on uh I don't know Jane Austin and you know Sherlock know Jane Austin and you know Sherlock know Jane Austin and you know Sherlock Holmes and it throws all that in to H Holmes and it throws all that in to H Holmes and it throws all that in to H it's like sounds good it's like no this it's like sounds good it's like no this it's like sounds good it's like no this is the worst the most horrible thing is the worst the most horrible thing is the worst the most horrible thing I've ever read in my life I've ever read in my life I've ever read in my life so are you are you finding that the so are you are you finding that the so are you are you finding that the paragraphs that it's writing for you are paragraphs that it's writing for you are paragraphs that it's writing for you are different than you would write or is it different than you would write or is it different than you would write or is it like GitHub co-pilot like I was working like GitHub co-pilot like I was working like GitHub co-pilot like I was working with stepen Tob this morning and we were with stepen Tob this morning and we were with stepen Tob this morning and we were doing some coding and uh he had his doing some coding and uh he had his doing some coding and uh he had his intent I could see it in his eyes he was intent I could see it in his eyes he was intent I could see it in his eyes he was starting to type and then GitHub starting to type and then GitHub starting to type and then GitHub co-pilot saved him the typing but what co-pilot saved him the typing but what co-pilot saved him the typing but what was written was literally the direction was written was literally the direction was written was literally the direction he was going is it writing the paragraph he was going is it writing the paragraph he was going is it writing the paragraph you were going to write or does it you were going to write or does it you were going to write or does it surprise you and write something you surprise you and write something you surprise you and write something you would have never written take a moment would have never written take a moment would have never written take a moment to talk about fine tuning now this is my to talk about fine tuning now this is my to talk about fine tuning now this is my opinion and some disagree so that's fine opinion and some disagree so that's fine opinion and some disagree so that's fine but I'm of the school of thought that but I'm of the school of thought that but I'm of the school of thought that when you find tuna model I'm sure people when you find tuna model I'm sure people when you find tuna model I'm sure people have heard about this but when you find have heard about this but when you find have heard about this but when you find tuna model you should find tuna model to
-
tuna model you should find tuna model to tuna model you should find tuna model to change Behavior not to add change Behavior not to add change Behavior not to add content so to answer your content so to answer your content so to answer your question the paragraph that it would question the paragraph that it would question the paragraph that it would write write write was because either I say continue was because either I say continue was because either I say continue writing from the LA you know continue writing from the LA you know continue writing from the LA you know continue writing the story so it kind of goes so writing the story so it kind of goes so writing the story so it kind of goes so if I if I start off and say uh Tim uh if I if I start off and say uh Tim uh if I if I start off and say uh Tim uh jumped in the the the taxi cab on his jumped in the the the taxi cab on his jumped in the the the taxi cab on his way to the airport uh for a big meeting way to the airport uh for a big meeting way to the airport uh for a big meeting and then I say continue the story thei and then I say continue the story thei and then I say continue the story thei would come back and say uh Tim arrived would come back and say uh Tim arrived would come back and say uh Tim arrived at the airport uh got out of the cab and at the airport uh got out of the cab and at the airport uh got out of the cab and got on the plane okay but it uses a got on the plane okay but it uses a got on the plane okay but it uses a writing style that's not quite mine what writing style that's not quite mine what writing style that's not quite mine what I found with fine tuning was that if I I found with fine tuning was that if I I found with fine tuning was that if I rewrite it you know to kind of make it rewrite it you know to kind of make it rewrite it you know to kind of make it my more writing style like no he didn't my more writing style like no he didn't my more writing style like no he didn't jump out of the cab he exited the cab uh jump out of the cab he exited the cab uh jump out of the cab he exited the cab uh no he didn't uh jump on the plane he no he didn't uh jump on the plane he no he didn't uh jump on the plane he Strode Through The Concourse to the Strode Through The Concourse to the Strode Through The Concourse to the plane so I was able to then take that plane so I was able to then take that plane so I was able to then take that run it through a through run it through a through run it through a through fine-tuning then when I asked it to fine-tuning then when I asked it to fine-tuning then when I asked it to write a paragraph but using this now write a paragraph but using this now write a paragraph but using this now fine tune model the writing was more my fine tune model the writing was more my fine tune model the writing was more my style so what I did was I also built style so what I did was I also built style so what I did was I also built into the AI story Builders a button that into the AI story Builders a button that into the AI story Builders a button that if you have a bunch of stuff written and if you have a bunch of stuff written and if you have a bunch of stuff written and I think you need to really write about I think you need to really write about I think you need to really write about at least say three pair uh three at least say three pair uh three at least say three pair uh three chapters worth of material you can just chapters worth of material you can just chapters worth of material you can just hit a button it creates the file that's hit a button it creates the file that's hit a button it creates the file that's needed for fine-tuning you can then
-
needed for fine-tuning you can then needed for fine-tuning you can then upload it to open aai you have to upload it to open aai you have to upload it to open aai you have to probably wait like maybe an hour or so probably wait like maybe an hour or so probably wait like maybe an hour or so and you get back a a fine tune model you and you get back a a fine tune model you and you get back a a fine tune model you can then inside my application say can then inside my application say can then inside my application say please now use this fine tune model and please now use this fine tune model and please now use this fine tune model and then you should find that the writing then you should find that the writing then you should find that the writing the paragraph that it gives you is more the paragraph that it gives you is more the paragraph that it gives you is more closer to your style and less cat gbt is closer to your style and less cat gbt is closer to your style and less cat gbt is so that was another thing that you know so that was another thing that you know so that was another thing that you know work for me yeah people hear fine-tuning work for me yeah people hear fine-tuning work for me yeah people hear fine-tuning a lot that term is used but I think from a lot that term is used but I think from a lot that term is used but I think from a people maybe maybe not understand what a people maybe maybe not understand what a people maybe maybe not understand what they're providing like there's the very they're providing like there's the very they're providing like there's the very large language model you mentioned chat large language model you mentioned chat large language model you mentioned chat tb4 40 fine tuning is it a is it a Delta tb4 40 fine tuning is it a is it a Delta tb4 40 fine tuning is it a is it a Delta is it more more grounding how does it is it more more grounding how does it is it more more grounding how does it relate to see that's the thing I don't relate to see that's the thing I don't relate to see that's the thing I don't think that you should ever try to use think that you should ever try to use think that you should ever try to use fine the fine-tuning process to ground fine the fine-tuning process to ground fine the fine-tuning process to ground um a model because one thing is these um a model because one thing is these um a model because one thing is these models have so much information already models have so much information already models have so much information already in in in them you know what I'm saying um that if them you know what I'm saying um that if them you know what I'm saying um that if you need to add information that it you need to add information that it you need to add information that it doesn't know about for example your doesn't know about for example your doesn't know about for example your company's uh private database of company's uh private database of company's uh private database of information use rag retrieval augmented information use rag retrieval augmented information use rag retrieval augmented generation which again is the what's generation which again is the what's generation which again is the what's essentially I'm doing with AI story essentially I'm doing with AI story essentially I'm doing with AI story Builders use that to provide more Builders use that to provide more Builders use that to provide more information fine-tuning in my opinion information fine-tuning in my opinion information fine-tuning in my opinion should only be used to uh control like I should only be used to uh control like I should only be used to uh control like I said in my case the the writing style said in my case the the writing style said in my case the the writing style you know the basically the behavior it you know the basically the behavior it you know the basically the behavior it should only be used to control the should only be used to control the should only be used to control the behavior meaning the information in the
-
behavior meaning the information in the behavior meaning the information in the model is the same but remember does it model is the same but remember does it model is the same but remember does it want to write like Jane Austin or want to write like Jane Austin or want to write like Jane Austin or doesn't want to write like uh Tom Clancy doesn't want to write like uh Tom Clancy doesn't want to write like uh Tom Clancy that's Behavior it's you know they they that's Behavior it's you know they they that's Behavior it's you know they they both are covering you know characters both are covering you know characters both are covering you know characters and that but the and that but the and that but the behavior is is different that is what I behavior is is different that is what I behavior is is different that is what I feel feel feel fine-tuning is really meant for and I fine-tuning is really meant for and I fine-tuning is really meant for and I always advise people when they go hey I always advise people when they go hey I always advise people when they go hey I want to fine-tune the model I'm like want to fine-tune the model I'm like want to fine-tune the model I'm like unless you're you're fine-tuning it for unless you're you're fine-tuning it for unless you're you're fine-tuning it for Behavior no you do not want to fine Behavior no you do not want to fine Behavior no you do not want to fine tuning just use rag so but that again is tuning just use rag so but that again is tuning just use rag so but that again is my opinion and Others May disagree so my opinion and Others May disagree so my opinion and Others May disagree so when we hear about rag we usually think when we hear about rag we usually think when we hear about rag we usually think about databases but you and I have about databases but you and I have about databases but you and I have something in common we are big fans of a something in common we are big fans of a something in common we are big fans of a using using data stores and a file using using data stores and a file using using data stores and a file system is a perfectly good data store I system is a perfectly good data store I system is a perfectly good data store I like you know XML you like text files like you know XML you like text files like you know XML you like text files right do you ever anyone ever give you a right do you ever anyone ever give you a right do you ever anyone ever give you a hard time for putting a bunch of text hard time for putting a bunch of text hard time for putting a bunch of text files in a folder and call it in a files in a folder and call it in a files in a folder and call it in a database uh not to my database uh not to my database uh not to my face I mean I like it because honestly a face I mean I like it because honestly a face I mean I like it because honestly a file system for a certain kind of data file system for a certain kind of data file system for a certain kind of data is a perfectly reasonable thing we we we is a perfectly reasonable thing we we we is a perfectly reasonable thing we we we happily let people put Jason in um in happily let people put Jason in um in happily let people put Jason in um in storage accounts in the cloud but if storage accounts in the cloud but if storage accounts in the cloud but if they put it on the file system we judge they put it on the file system we judge they put it on the file system we judge them for their choices exactly so no it them for their choices exactly so no it them for their choices exactly so no it works great so yeah so for AI story works great so yeah so for AI story works great so yeah so for AI story Builders um I wanted obviously to keep Builders um I wanted obviously to keep Builders um I wanted obviously to keep it I want to make it free but I don't it I want to make it free but I don't it I want to make it free but I don't want to you know pay for you know want to you know pay for you know want to you know pay for you know databases and and having user accounts
-
databases and and having user accounts databases and and having user accounts and all that so the way a AI story and all that so the way a AI story and all that so the way a AI story Builders works is all the information Builders works is all the information Builders works is all the information and of course the way you know large and of course the way you know large and of course the way you know large language models work is when people type language models work is when people type language models work is when people type things in they're turned into tokens things in they're turned into tokens things in they're turned into tokens like you mentioned but then we um turn like you mentioned but then we um turn like you mentioned but then we um turn them into embeddings and an embedding is them into embeddings and an embedding is them into embeddings and an embedding is these massive these massive these massive numbers um of vectors all that is saved numbers um of vectors all that is saved numbers um of vectors all that is saved to text files there's two versions of a to text files there's two versions of a to text files there's two versions of a of AI story Builders one is a uh Blazer of AI story Builders one is a uh Blazer of AI story Builders one is a uh Blazer uh Maui app that you get in the Windows uh Maui app that you get in the Windows uh Maui app that you get in the Windows store you know you can download it for store you know you can download it for store you know you can download it for free in in the Windows store but then free in in the Windows store but then free in in the Windows store but then another one is a Blazer web assembly app another one is a Blazer web assembly app another one is a Blazer web assembly app that just pops up in your browser and that just pops up in your browser and that just pops up in your browser and just works what happens is is that as just works what happens is is that as just works what happens is is that as you type things I'm creating these um you type things I'm creating these um you type things I'm creating these um text files in um Temp text files in um Temp text files in um Temp Storage and when you navigate away from Storage and when you navigate away from Storage and when you navigate away from the page I go the page I go the page I go in I zip them all up and I put them in in I zip them all up and I put them in in I zip them all up and I put them in local storage so it's in it's in the web local storage so it's in it's in the web local storage so it's in it's in the web browser so if you go to the applications browser so if you go to the applications browser so if you go to the applications tab you know you'll see this you know tab you know you'll see this you know tab you know you'll see this you know long string which is basically the long string which is basically the long string which is basically the contents of a zip file so then when you contents of a zip file so then when you contents of a zip file so then when you say I want to look at the story and you say I want to look at the story and you say I want to look at the story and you say I want to look at the story The Code say I want to look at the story The Code say I want to look at the story The Code goes into local storage pulls that out goes into local storage pulls that out goes into local storage pulls that out it's really a zip file it unzips it into it's really a zip file it unzips it into it's really a zip file it unzips it into you know this temp thing where all these you know this temp thing where all these you know this temp thing where all these text files are and then the code runs text files are and then the code runs text files are and then the code runs because it's the same code that the the because it's the same code that the the because it's the same code that the the Blazer Maui app uses because the Blazer
-
Blazer Maui app uses because the Blazer Blazer Maui app uses because the Blazer Maui app I just store the those text Maui app I just store the those text Maui app I just store the those text files in your window Window app storage files in your window Window app storage files in your window Window app storage so I was able to use the exact same code so I was able to use the exact same code so I was able to use the exact same code for both the the Maui app and and the uh for both the the Maui app and and the uh for both the the Maui app and and the uh the web assembly app but it's just text the web assembly app but it's just text the web assembly app but it's just text files there's no files there's no files there's no database you know I uh when I think of database you know I uh when I think of database you know I uh when I think of writing a novel I've got a friend uh writing a novel I've got a friend uh writing a novel I've got a friend uh named Lauren buas who's a famous science named Lauren buas who's a famous science named Lauren buas who's a famous science fiction writer and she always does fiction writer and she always does fiction writer and she always does details and timelines and locations and details and timelines and locations and details and timelines and locations and characters and I kind of Imagine like characters and I kind of Imagine like characters and I kind of Imagine like that guy from um I think it was always that guy from um I think it was always that guy from um I think it was always sunny Always Sunny philadelph L the sunny Always Sunny philadelph L the sunny Always Sunny philadelph L the strings and the yeah exactly basically strings and the yeah exactly basically strings and the yeah exactly basically when I think of an author I imagine a when I think of an author I imagine a when I think of an author I imagine a giant wall filled with Post-it notes giant wall filled with Post-it notes giant wall filled with Post-it notes with everything right what I'm hearing with everything right what I'm hearing with everything right what I'm hearing is that AI storybuilder decom is that is that AI storybuilder decom is that is that AI storybuilder decom is that and they're not in Post-it notes they're and they're not in Post-it notes they're and they're not in Post-it notes they're in text files and then the magic is the in text files and then the magic is the in text files and then the magic is the AI can look at the giant wall full of AI can look at the giant wall full of AI can look at the giant wall full of Post-it notes and that's the context Post-it notes and that's the context Post-it notes and that's the context that it needs to help you move your that it needs to help you move your that it needs to help you move your story forward right because it's good at story forward right because it's good at story forward right because it's good at that computers are good at taking a that computers are good at taking a that computers are good at taking a bunch of data and figuring things out bunch of data and figuring things out bunch of data and figuring things out and not getting confused that's what and not getting confused that's what and not getting confused that's what computers are good at but creativity not computers are good at but creativity not computers are good at but creativity not so much that is really interesting that so much that is really interesting that so much that is really interesting that is a very practical and pragmatic is a very practical and pragmatic is a very practical and pragmatic calling back to the beginning of the calling back to the beginning of the calling back to the beginning of the conversation as you're a well-known uh conversation as you're a well-known uh conversation as you're a well-known uh pragmatist uh way of rather than trying pragmatist uh way of rather than trying pragmatist uh way of rather than trying to build some app that said I'm going to to build some app that said I'm going to to build some app that said I'm going to tell you a bunch of stuff and I'm going tell you a bunch of stuff and I'm going tell you a bunch of stuff and I'm going to go to sleep and when I wake up I want to go to sleep and when I wake up I want to go to sleep and when I wake up I want a book you're having it give you a step a book you're having it give you a step a book you're having it give you a step a step forward and I'm sure that you a step forward and I'm sure that you a step forward and I'm sure that you maybe it writes a paragraph you make a maybe it writes a paragraph you make a maybe it writes a paragraph you make a change to a location or a character and
-
change to a location or a character and change to a location or a character and then you refine so it's about iterative then you refine so it's about iterative then you refine so it's about iterative improvements iterative exactly that's it improvements iterative exactly that's it improvements iterative exactly that's it fantastic and every Microsoft co-pilot fantastic and every Microsoft co-pilot fantastic and every Microsoft co-pilot application that I've been dealing with application that I've been dealing with application that I've been dealing with that's what it is it's always iterative that's what it is it's always iterative that's what it is it's always iterative I mean think about the chat gbt goes you I mean think about the chat gbt goes you I mean think about the chat gbt goes you know the last thing it says to you is uh know the last thing it says to you is uh know the last thing it says to you is uh did that answer your question would you did that answer your question would you did that answer your question would you like more information you know co-pilot like more information you know co-pilot like more information you know co-pilot microsoft.com is is at the end they microsoft.com is is at the end they microsoft.com is is at the end they always have the links like you know always have the links like you know always have the links like you know because they know it's it's it's because they know it's it's it's because they know it's it's it's iterative yeah I don't think of I can't iterative yeah I don't think of I can't iterative yeah I don't think of I can't think of a single Microsoft co-pilot uh think of a single Microsoft co-pilot uh think of a single Microsoft co-pilot uh application where they truly intend for application where they truly intend for application where they truly intend for you to ask it to do something it does you to ask it to do something it does you to ask it to do something it does that thing and that's the end of the that thing and that's the end of the that thing and that's the end of the conversation yeah yeah let's think about conversation yeah yeah let's think about conversation yeah yeah let's think about this can you think of a single situation this can you think of a single situation this can you think of a single situation where you know what I'm saying where where you know what I'm saying where where you know what I'm saying where that's how the thing is supposed to work that's how the thing is supposed to work that's how the thing is supposed to work no very well said very well said so no very well said very well said so no very well said very well said so folks can go to AI folks can go to AI folks can go to AI storybuilder Doc and the documentation storybuilder Doc and the documentation storybuilder Doc and the documentation is very easy to find documentation.
-
is very easy to find documentation. is very easy to find documentation. storybuilder docomo ADF webs server.com that's a link docomo ADF webs server.com that's a link to all of your sites and you got a ton to all of your sites and you got a ton to all of your sites and you got a ton of sites you've got sites on Blazer data of sites you've got sites on Blazer data of sites you've got sites on Blazer data and Blazer help and blogs and explorers and Blazer help and blogs and explorers and Blazer help and blogs and explorers and all kind of stuff that you've been and all kind of stuff that you've been and all kind of stuff that you've been working on over the years including working on over the years including working on over the years including books that they can link to uh that you books that they can link to uh that you books that they can link to uh that you have published up on Amazon uh thanks so have published up on Amazon uh thanks so have published up on Amazon uh thanks so much for chatting with me today Michael much for chatting with me today Michael much for chatting with me today Michael Washington thank you so much for having Washington thank you so much for having Washington thank you so much for having me um I love your show I listen to it in me um I love your show I listen to it in me um I love your show I listen to it in the car you know every week and um I'm the car you know every week and um I'm the car you know every week and um I'm just so grateful that you allowed me to just so grateful that you allowed me to just so grateful that you allowed me to uh share something no I I learned a lot uh share something no I I learned a lot uh share something no I I learned a lot and this is going to have me go and and this is going to have me go and and this is going to have me go and explore and my son loves to write explore and my son loves to write explore and my son loves to write stories and I'm going to go check out AI stories and I'm going to go check out AI stories and I'm going to go check out AI storybuilder storybuilder storybuilder docomo out it's available on the uh on docomo out it's available on the uh on docomo out it's available on the uh on the Windows store and as a pwa yep all the Windows store and as a pwa yep all the Windows store and as a pwa yep all right this has been another episode of right this has been another episode of right this has been another episode of Hansel minutes and we'll see you again Hansel minutes and we'll see you again Hansel minutes and we'll see you again next week
Summary
The discussion centers on AI's practical application, contrasting enthusiastic adoption with a pragmatic, skeptical approach. Key subjects include .NET, Blazer, and AI tools like GitHub Copilot and Azure OpenAI. The takeaway is that AI is a valuable, but not universal, tool in a developer's arsenal.