ChatGPT & Emerging AI Use Cases

hello and thank you for watching this video about Advanced chat GPT strategies the inspiration for this video and video series comes from my experience with chat GPT as I was beginning to learn to use it and saw the tool as I saw a lot of guides about basic information basic usage advice on how to use chat GPT and I thought there has to be more detail out there and so I've started to explore and provide information for you so what follows in this video is an interview with my brother who is also a business owner he's a graphic designer and we've been discussing a lot lately chat GPT mid Journey which is an AI art generator and different ways that you can use these tools at work and in your daily life and so it's a very interesting interview I hope you enjoy it I want to apologize for the video quality I recorded that with a different computer and realized shortly thereafter half I'm going to be doing this YouTube thing I need I need a more powerful computer so we're working on that um it just to keep a good high level of quality for all of you who are viewing but I still hope you like the content of this video there are some screen images that are shared as well so you can kind of see prompts that are being used and different strategies in this video later on in the series we'll have specific videos that address specific legal topics currently the only videos that are going up today are this video this interview with my brother and a set of videos for Dungeons and Dragons that will show you how to use Advanced input to get Advanced structured output and and so it's just really fascinating to see how different people are using artificial intelligence Tools in particular chat GPT which this video guide is about so again I hope you enjoy the interview I hope the choppiness of the video here in the interview isn't too distracting and that you get great benefit from watching [Music] thank you so thanks for coming on bro um oh yeah I I I I know that you have been very interested in artificial intelligence and I have been playing with artificial intelligence a lot and so I just thought it would be fun to talk to you a little bit about your thoughts and feelings about uh both mid-journey but I'm going to focus more on chat GPT because what I'm doing is I'm recording a series of videos about how to use chat GPT as a lawyer I've seen some videos where lawyers log on to chap GPT and say well it can make a simple contract but it kind of sucks and so it's not a usable tool there you go right that's the end of the story I just don't think that's a realistic view I don't think lawyers and I haven't found a video like I've found lots of videos about about chat GPT online and they're all like here's the beginner's guide or The Beginner's use case or whatever nobody's making the advanced guide to how to use this for their specific industry and so what I wanted to do was make a more of an advanced guide for lawyers to use chat GPT and I wanted to get your thoughts while I do it well your your screen is right here your thoughts while I do it yeah your thoughts you don't need to look at that one uh no I and and I think you're totally right a there are a lot of videos on how to use Chachi PT the beginner kind of setting and it it it it it's exactly what you said you're the lawyers that you've seen they get in there and they try a couple prompts and then they're like oh this kind of sucks and they don't really dig into you know learning the tools because right now everybody I mean this is all new technology to just about everybody except for those people that have built it right and so everybody's trying to figure out ways to uh engage with it and how to use it well uh in fact a friend of mine talking about mid-journey she recently changed her name on LinkedIn to her name and then A.I conjurer that's what she put as her title and yeah uh which is fantastic but she also produces so many amazing things uh in the journey and in other things but uh it's like I've told many people it is the writers people who write well are the ones that are doing the best with these tools in my opinion uh and so chat GPT you you have to learn how to talk to it and get it to engage uh I mean like the example the other night when we were like write some insults and it's like well I can't be mean and then we explain the situation about why we wanted the insults and for our our Dungeons and Dragons game as a Bard for vicious mockery a spell that it's involved with insults and so it created some really fun kind of campy insults that were just good fun kind of things and that were game appropriate yeah we're game appropriate and so you know it you have to direct these Technologies to what the focus you want them to be uh so I I think your your lawyers gave up too quickly in my opinion they well so they've looked at it and said oh yeah it does okay but it's not making you know the multi-million dollar contract to sell some business to you know right right and it will at some point it will do that because those contracts will be available online well so I had a couple of things in mind that I wanted to talk to you about and the first of which is specificity and then also delegation so and I think that this is a helpful discussion to have when it comes to AI because also as a business owner you are learning and we have started to have those conversations about what it takes to effectively delegate something and then and then I also wanted to talk a little bit about kind of some of the limitations around GPT so to effectively delegate something I have learned as a business owner that that you have to give very specific instructions and I used to avoid this because I didn't want to micromanage I wanted to let people you know be themselves do their own thing but I have learned over time that the more specific I am with the instruction I give the better results I get and that's with um inputting information to a human and expecting a result back right and so what are your thoughts about that when it comes to chat GPT Della essentially it will be an assistant that we delegate tasks to and so how do you feel about that well like like you just said I think it's I think you're totally right we have to give and I love that word specificity and we use it a lot in CSS coding in for websites there's a certain level of specificity that you have to dedicate if you want things to function properly and it goes to the same thing with humans and with uh AI in this example and with humans it's obviously our learned experience Shades or gives us a typical type of interpretation with what the directions that we've been given and the same thing with AI it has learned quote learned something but it's only learned in a certain way and what it's been fed uh in the example of chat GPT it doesn't know anything about uh 2022 and things that came out in 2022. and so like I mentioned a couple days ago I had an argument with it about mid journey and it being an AR art platform and it was like no it's a behavioral health it's a behavioral assessment platform and I'm like what are you talking about right right and did those constraints and we've had number of chats I love limits and constraints too because so I want to talk about the constraints as well we'll get there yeah and and so it all comes into that specificity these are the constraints you know this is what you don't want it to act as something that you don't want the technology to act as a a circus clown you want it to act as a lawyer or you wanted to act as a social media expert so let's get into that for a second so one one of the things and I do believe that it's helpful that if you want to use chat GPT well that you should also get on and use some of the uh generative art programs as well like mid-journey or maybe stable diffusion or something because I think that there's skills that are overlap and lessons that overlap between the two AI tools so um to get a specific kind of image you have to give it a specific kind of input and as I've poked around with mid-journey and and played with that tool one of the things that I noticed when I was looking through the Community Showcase was I'd see these extremely they look like pictures and I said well how do you get that like how do you tell it to give you that level of specificity and with mid-journing it can happen by accident like you can ask for you know show me a picture of a field and then it might show you one of the options looks like it's an actual photo or a picture but I saw a couple of examples that to me were fascinating where The Prompt included you know picture of whatever it was picture of a woman in her 20s on a sunny afternoon taken from a Canon EOS R6 aperture size this uh focal length this and so I'm all like whoa the specificity in that instruction just by and in fact what's interesting is that you don't have to be that specific if you say taken from this kind of camera you know all of these pictures have metadata and so I assume I don't know but I assume that a program like mid-journey when it looks at the images is also looking at the metadata and so it must be able to see well every single picture that was taken with that camera with that focal length that I have in my database I can make it look exactly like that and so that to me is is mind-blowing that level of specificity allows you to get a very detailed and a very certain kind of picture and so when we look at people drafting prompts for chat jpt the prompts that seem to work best are the ones that give a similar level of specificity the the language side as well and so that's how you start to get better and better results from chat GPT exactly when you tell it right in the style of it will it will it will take that and I did an exercise last week with that where uh I I had it look at my website and it reviewed the copy I I put in it and it said well here I'll I'll pull it up and it's going to show your traditional writing style I can I don't know can I share my screen let's see it might let you I don't know actually well here I'll do a Chrome tab it looks like it will let me so here's my here's the tab can you see that um you know I'm gonna make it the only screen so your Chrome tab is now showing in the video go ahead okay uh now that was uh yes here it is okay so I it put my website copy in here and I told it you know acting as an expert copywriter analyzed the following copy for writing style voice of tone because I I don't know what my tone voice in tone is and so it it came in and said oh well um you know here's the style you are well organized your subheadings you know thorough The Voice is authoritative and confident position yourself as an experts in the field you're enthusiastic with a passion for the work and then the professional tone is just like but not overly formal or stiff uh with a sense of approachability upbeat and a sense of collaboration partnership so now I can take that and when I want to write a piece of copy I will say something like acting as a professional copywriter write a blog or a bit of copy that is business-like but not overly formal and stiff with a sense of collaboration and partnership and then it will write a somewhat nice now again being said there's nothing that I have just taken from chancypt and loaded to whatever I'm using I'm using it more for ideation in the capacity that I use it in or I'll edit it heavily there's only been one piece that I edited two words and said this is pretty fantastic and moved on so now you've been using traffic not only for web content but like primarily your social media posts right to kind of help you get through those quicker is that accurate correct I use it to generate ideas and uh wording uh to help me March through the process at a quicker rate um I'm an okay writer at present I think I will improve with using this technology because it gives me ideas about how to go about it and I in the past I've told my rider Buddies the ones that I work with closely over the last 10 years or so you guys are the ones that taught me English and how to write your marketing writer bodies uh right right because they they and it goes again back to the specificity you know they wrote with the AP style guide and we followed these particular guidelines and this is the and I like those particular guidelines so it helped me identify how to write better now I want to jump in and and add something there when it comes to Legal writing so I have obviously been playing with that and and in my video series what I'm going to do is is show people how you can write a legal brief with chat GPT how you can write a contract an extensive contract not just a short contract with chap GPT and and that's my plan for those videos but it's interesting that you mentioned AP style format and those kinds of things so one of the catches one of the difficulties in using the tool for legal writing is with legal writing you tend to need to cite your work and that's also been one of the biggest problems that people have with chat GPT is they're like we can't show the homework that's happening in the back end so what's fascinating is is I have learned and figured out how to get it to what site some work yes so I'll ask it to write something for me and then I'll say now cite your sources and the first time I did it it wrote a similar copy and then provided websites and sources that it got from and then I said no no no I need you to do this again but I need you to cite legal sources including case law and then after every conclusory sentence I need you to provide the citation in Blue Book format which is the lawyer's standard format for citing you know case law and whatnot and then what it spat out was exactly what I needed I couldn't have read it wrote and written it better myself uh with and it wrote it for you in oh yeah five seconds right and it wrote it wrote These rule statements for me uh with proper proper legal citation and so and there was one case that seemed made up I checked it it's a legitimate case but it's not about what chat GPT thought it was and so that's the piece where you know the law you're working together with the AI but the fact that I got it to spit out like 95 correct information with legal citations I'm the only person I know who's been able to do that and that's why I've decided I was going to make this video series and say look guys this is a real legit usable tool can save you hours and hours of time and effort and work um and so yeah I think it's it's something that needs to be talked about well yeah I completely agree with what you just said about the the you know you have to review it you can't you can't just go yeah yeah this is good I'm gonna load it up because you never know what that citation means because it doesn't it just like a person it's going to interpret one thing one way and see it one way and you have to it brings to mind trust but verify you know so yeah yeah yeah I like that trust but verify but chat GPT yeah so uh yeah that that's fantastic when you started into that talking about the the the citations I was like oh my gosh I didn't think about that with you know they I'm sure you've seen the articles about New York City School Board saying no chat GPT we won't allow this and uh you know but we're kids writing essays but were they writing them without citations and then you know how can you get it to do that now so let me let me let me actually talk about that and I do plan to create like a separate video just propose I mean maybe there's a little high-minded on my part but I was thinking you know we should talk about the regulatory framework that that should surround GPT but its uses and all of that kind of thing oh yeah I heard you say that like you mentioned chat GPT has a very high likelihood of making you a better writer right and as I've talked to one of the other lawyers in the office about it as well he was saying oh yeah because you get opportunities to see multiple ways of writing it and then you can choose the word choices that you like and make good choices about the word choices the grammar the sentence treasure and and frankly as a tool to help teach writing better chat GPT could be very very useful and so I think it's perhaps wrong-headed or backwards looking to say no we can't use this tool right like I remember a moment in middle school when someone told me I it was it was my middle school English teacher said all of the papers that you write in college you will write in cursive by hand and you will never have a calculator with you at all times right like come on you know give me a break it's just not true none of that is true there was a a I can't remember where I heard it but it was so funny it was so true so all through school we're told not to collaborate not to do work together uh because it's called cheating And yet when you get into business World we're also to collaborate and work together to accomplish the goal uh and then these type of Technologies I think you know it like you said it's wrong-headed they're doing a disservice to their students because today's technology will be tomorrow's Norm right and this is disruptive right now but tomorrow you know tomorrow 10 years from now when those kids are out of school it it this is going to be the tool they refer to I mean as you showed we are already dependent upon these I mean research has uh 90 of adults in the United States have a smart device in their pocket that they use on a regular basis to look up information and find these things out so how is this new technology going to benefit us tomorrow where we're just kind of muddling around in the mud right now with it so yeah yeah it's an interesting one yeah I I think that it's fascinating to kind of consider what direction this could all go but one of the things that I do hope is that schools kind of quickly grasp onto this I think part of the resistance is this idea that people won't learn to write but hopefully again we'll start to see it for what it is as a tool to help you write better right like go ahead well there's where you talk about not learning how to write so uh there's a story and I need to look up this Mythos to to verify it completely and and really understand the structure of this but they're in there's a story that I read ages ago and it talks about writing and how writing came to be in Egypt hieroglyphs and whatnot where prior to that really uh our knowledge the bank of human knowledge was transferred by Oral histories and a oral historian oral historian when they started scratching stuff into stone was like this is going to ruin the world we'll won't we won't be able to remember anything because we're writing and and where yes that is true but look look at the breadth of human knowledge now where we've written all these things down and I mean you look at it I mean you could say the same with disruptive technology like video now there is a breadth of human knowledge larger than the Library of Alexandria online at a a just a quick search to figure out anything from plunging your bathroom toilet to fixing a car to learning how to run AI to improve your life so it all these Technologies yes we're we're losing out on something but is the benefit does the benefit outweigh what we're losing an add-on and are we truly losing it because we still communicate orally with each other and we still remember things but we relegate our memory to the things that are truly important to us rather than just the random things that we well you know or the things that are more meaningful or fun for us right like everybody knows how to tell a story everybody knows how to tell a joke uh these are the things that we want to remember um Tanya just started using um her digital calendar like more and it was kind of funny because now that she's using her calendar More there are things that she's just not telling me because she doesn't remember them off the top of her head and so it's actually become a little bit disruptive to our lives uh you know and it's not like the the digital calendar hasn't been around for forever right it's just one of those new habits that she started on and so it's kind of funny how that that has affected us but I do definitely think that uh oh you know in the long run it's gonna be better to use that digital camera excuse me not camera calendar and we'll be able to share it and be more on top of things ultimately but in the short term there are going to be hiccups along the way I think the same is true for artificial intelligence and I don't think chat GPT is the end-all and be-all I mean I'm excited to see what Google bard comes out with frankly very excited chat gpd's amazing tool I hope to keep using it but um I'm also excited to see what else develops in our in our capitalist Society filled with competition and other options so that's fair and and the only the technology is only going to improve and I'm excited to see what happens when they finally do insert 2022 into chat GPT or any other softwares that are limited by a time frame uh so you know it and that's that's one of the wonderful scary interesting thoughts is you know they they've put information into this from 2021 but they have they put the breadth of human knowledge into it uh you know have they well which currently is it possible you know the amazing things though about chat GPT is this is and so I do want to talk about this for a second is its ability to be an expert in a variety of topics at the drop of a hat so back to the specificity argument right so you and I have watched some videos together and shared some links and so I know that one of the examples that's out there is you could ask chat GPT to give you a workout program and it'll give you some exercises but if you ask it to act as a personal trainer and then you give it your own stats and say develop a personalized workout program for me then what you're going to get is going to be significantly better and you know I also am aware that what is it none of us no human being alive could ever watch all that will be on YouTube right like more than a bajillion hours of video or uploaded every second right and so but the thing is about anything 82 82 years of content is loaded to YouTube the last stat that I heard daily 82 years crazy that's crazy no so anyway here is though is that an artificial intelligent computer could watch all of YouTube and could tell us what's most important or could could guide us to recommendations for TV shows or videos that we might like better than than we could figure out ourselves right and so that's that to me is what's amazing about chat GPT is that it will be a better writer it's the same argument that they make for self-driving cars which which I and and I and I want to talk about regulations here in a bit why I think they're important and all of that kind of thing but there's this argument there that a self-driving car is going to be an infinitely better driver than any human very soon because you know while I can only have you know so many years of Driving Experience multiply that times 100 or times a thousand or times a million for any self-driving car they're going to have 20 years of experience times 100 million people it will be a better driver than I will ever be period right and and so there's this there's there's this this I guess mental leap that we have to take uh to accept artificial intelligence as well um on the other side of that coin though is people pushing too quickly or not recognizing certain things and and risking human life or human sanity as a result of pushing technology forward particularly when it comes to like car safety like I read one report about artificial intelligence in cars uh they didn't train it to look out for jaywalkers and so a car that was self-driving had to hit somebody because because it had only trained been trained to look out for crosswalks and people crossing at crosswalks not people crossing at unexpected places right and so that's obviously a problem that a human wouldn't make right and so there's that caveat to yes it will be a better driver but also it's going to make dumb mistakes that a human would not make and we'll have to work through some of those problems and and that goes into the constraints that you give it you know and and us we're creating we're creating this technology I mean and humans we're all flawed and we can't think of everything so someone didn't think hey there might we might need to have this thing look out for jaywalkers and yeah worry about that so you know they didn't think about it and the same with a number of things the just any think of any scenario and and I like to I I'm not I try not to be pessimistic by nature but anytime there's a decision that is risky I like to treat try to think of five of the worst outcomes that could possibly happen and then backtrack that backtrack the decision and go okay is that the worst thing that can happen oh great then that's the decision I'm going to make and you know so we have to look at this technology yeah it's going to be great and like you just said they're going to have years upon years and years of experience it's like uh it made me think just now of any of those websites that you see we we have combined 120 years of experience while each of those individual peoples in that business only has nobody actually has 15 years right yes they're going to bring their minds together and they're going to have that collaboration and that's what I think you know we should be doing with AI is having a collaboration not just a this is how we're going to do it not a full-on replacement so and part of the reason that it can't or shouldn't replace in particular I would say law of course is one of those areas where a lot of writing is required a lot of analysis of words happens and so I could see AI moving very heavily into the space and yet at the same time because of unintentional biases and how you know AI can be programmed I think it's very important to have human guard rails on the use of this technology for the foreseeable future um so but I wanted to talk a little bit about constraints so you have a book that you're working on I know Thinking Inside the Box and how creativity loves constraints um in my experience with chat CPT there are a couple of constraints one both of which I find a little disappointing I also kind of Wonder out loud like if they are unintentional security features as well because as we learn to use about it so chat GPT even within a conversation I've had some conversations go quite long will forget after about three or four thousand words it doesn't and I look this up because it couldn't refer to something from earlier we discussed we I'm talking about the chat GPT as a wee something I discussed with GPT we're going to humanize it because that's just who we are and we right we're going to humanize it so tell me more yeah no I already I already humanized it so so um and what's interesting is when I realized that it wasn't remembering I did feel like just a short twinge of loss I was like oh no it doesn't remember and so that's when I Googled and found go ahead the AI has memory issues and we're feeling sad about it that's true it's true well so then I Googled and realized that that it only could remember so much right so that was that was part of a I forget how open AI categorizes it but it only has a memory of so many tokens which turns out to be about 3000 words that's what it's posted somewhere yeah well and and one thing that's interesting that I've found is that yes those three thousand words but if you're good about setting it up say let's say I want 10 ideas about something okay I want you to write a post about each of those 10 ideas in this format it doesn't do a great job and it's hit that 3000 limit but you have numbers one two three four and you can say referring back to number one and it will look back through that thread and then come back with that content and ultimately I got and I got a thread so long it wouldn't do that oh really even with numbers yeah even even with that it got so the the thread was so long that it just couldn't even refer back to the conversation now maybe that was an error and if so I'd love for somebody to reach out to me and explain that to me from from openai but yeah that's interesting yeah because I thought it because it couldn't refer to its own words previously now yesterday when I did that it it locked up at number seven I think it wrote it wrote too much content and I was like okay so that's probably what it was it probably got to that limit where it could no longer remember um right so so then let's talk about the other limit that I've found or run into and I've done the word count multiple times to double check myself it won't spit out more than 595 words at a time on on a couple of occasions I got it to write out 596 words but it wouldn't go beyond 595 596 words and that six words was it dragging itself across the finish line exactly what it was and period no I've had it cut off a couple times in the middle of a sentence and so and so that to me probably understanding that constraint is my biggest argument against all these people that are like Well it can't write a contract and I'm like no you can't just ask it to write a 20-page contract about a specific issue but what you can do is you can have it help you right you can break it up you can have it write each section for you and so then your job as the human is the the person to guide the contract to guide the specifics of the contract to guide the organization of the contract to structure it and so one of my videos is going to show exactly how you I picked a non-disclosure agreement so we're going to draft like a multi-page non-disclosure agreement just something you know generic but common and and and see how it does um and how quickly it can do it because I have a feeling it's going to move pretty darn quick uh and I've already done some some basic testing on it but yeah I mean it'll help you generate the sections it'll help you and then there's a big difference between asking it to just write a section versus asking it to write as and I did a couple of different tests because everybody has their law review as somebody working on the Harvard Law review or the Columbia Law review or as a Virginia lawyer you get different kinds of responses from it as well uh and so that's been fascinating to kind of see how it analyzes prompts with that extra specificity right right and and what's neat about the that type of application is that there is so much content online that it can refer to or that has been fed to it to refer from that it can generate those things uh pretty easily I imagine and in that format let's talk about a couple of other um constraints so I know that you can give it a URL I've I've started to play with the summarize command and and I've given it URLs it's not very good at summarizing case law yet so for as a as a lawyer what you'd love to be able to do is to say check out this case here's a link give me the best Holdings and then write a paragraph with proper Blue Book citation format uh about the primary Holdings on this issue right now I think we're about one hop stepping a jump away from that like I I know that they're training gpt4 right that's being whispered about and so it's supposed to be released later this year and and chat GPT is technically just GPT 3.5 or trained on top of that and so I could see that being possible soon but I I've had difficulty getting it to look at specific websites and getting truly usable information from it unless unless the content is less than like a thousand words or so but like you said this technology is moving rapidly and mid-journey when I first got onto it uh the art was interesting but very amorphous and it was okay and people that you know Managed IT they were able to pull off some really interesting things but the quality wasn't really there like it is now and you talk about early in our conversation you talked about the specificity of the prompts in there and just the other day I wanted you know some people in a business suit for the cover of a post that I was doing and so I did uh two individuals one male one female on a Runway catwalk at a Ralph Lauren fashion show editorial photorealistic uh blah blah blah and it pro it made these two you know fashion model people in business suits even the hands were pretty good which these programs are terrible at hands um just because hands are hard to draw in the first place even for humans uh but it came up with this really cool image that I used for for that thing so definitely the specificity and the quality is going to improve as time goes on like you're saying chat gpt4 you know what's in store with that uh and I'm excited to see where they go because mid journey is now on version you can do version four I think four is the cap and just the amazing quality of the work that it can be put out now is phenomenal as long as you're prompting it with specificity and again it really comes down to that specificity because if you put a one word prompt it's going to do some interesting stuff and maybe you'll get lucky so yeah but again it goes into that interpretation and you know it's it's us learning prompt engineering is the word that I'm hearing a lot now for prompt engineering yeah working in this sector and learning how to make the best prompts again I think the writers are have hit their Heyday if they can really tap into this technology they can make it through Leaps and Bounds Beyond well what everybody else what's fascinating is you could go into chat PT I saw one person play around with this example going back and forth where they'd go into chat GPT and describe in a lot of detail and say give me the prompt for mid-journey to create this yeah and and then and then copy and paste The Prompt into mid journey and yeah get a better more specific image than they would have gotten had they just kind of tried to come up with it on their own it's fascinating so yeah it's it's really interesting the cyclical nature of this you know it's like okay I'm gonna have an AI write a prompt for me so that I can make another AI create a piece of art for me yeah I mean eventually so so the let's let's talk a little bit about um interoperability which I'm sure there are smarter people than I looking at the technical side of things but I also wanted to bring up this and the regular regulation stuff and then then we'll have been talking for about 40 minutes maybe we should jump off but um no so the first thing that occurs to me is when so I'm I've been a long time WordPress user and I I wouldn't call myself a web developer but I'm I know how to do a thing or two with some code and some CSS and HTML and connect different tools and I know that a lot of tools landed on this rest API which allows a lot of these Technologies to talk to each other and so one of the things that has already occurred to me is what are are the people in the AI space doing to make tools interoperable interconnected and in a way that's truly fantastic and I I just haven't seen anything people are probably working on this but I just don't know what they're doing have you heard or read anything about that uh I know there's some connectivity in that chat gbt is being run through they've got an API of their own yeah yeah and and you you ask if you need to write you an API for itself and it will yeah no I I have done that as well yeah and so there are a number of technology I think that are pulling in I can't remember which ones are I remember some people talking about that and they mentioned several platforms that are using it so like a lot of these AI platforms for naming things and for other things are utilizing the API to from GPT and kind of piggy banking on what they've learned so I think it's going to cut through a number of sectors not just code or HTML I mean I was I was raised on uh making a website by hand coding it so yeah I've had a lot of experience tables and all of that stuff yeah no I know what you're talking about tables and then you know the the epiphany of CSS which just made it so much awesome right yeah and but and now you can just go on to chat CPT and say right X you know web page for me at HTML and it'll probably spit out something pretty good now I haven't tried it but I imagine it'll it'll it's pretty awesome because I've seen more specific your prompt the better it is right writing some code to do that but I mean all of this is code anyway all of our programs that we're using all of our uh the Adobe software that's all code it's just they've specified it to do a certain thing right so so yeah I'll jump away from that oh go ahead no I think they're just gonna this technology is gonna cut through a lot of things and just be really interesting I mean I can only imagine what they're already thinking of to help automate uh manufacturing plants you know so that the robots that are you know doing XYZ work already do it and you talk about uh AI technology and cars and driving cars what are their applications in farming or mining or other Industries are available more productive I think the the connectivity between these AI things and and robots and devices that are actually in the real world are that and so and so that brings me to the the regulation framework so I was watching uh the John Oliver skit this week is about artificial intelligence that's a good one you should check it out but he mentions regulations about and but but no real like policy positions or Solutions but but we should be regulating this and we should be you know and I agree I mean I think uh companies don't have the greatest history of of thinking in the best interest of the world but they're thinking of shareholder value and that's that's what a company is supposed to do so I don't take issue with that but um obviously as a society we need to be thinking about okay well how do we protect ourselves Elon Musk in a lot of videos is out there talking about how he's I mean he sounds like he's scared of AI when he talks about it so like they're coming for us already and and they haven't even been made yet right but um so we know that there's concern and there's reason for concern one of the things that that John Oliver mentioned in his skit was that um the computer doesn't show its work and so and I do plan to interview Todd too I'm going to reach out to him but uh and ask him this because I don't know the answer but I would hope that these programmers put in place some kind of metadata behind each generation so that each generation of texts so that we could see so that we could see where it you know so we can see its homework so we can see what it did or how it arrived at the conclusion to create those word choices you know what I mean because one of the issues is it's so Advanced already that we don't know how it drew the conclusions of true but I think that that's an important like legal and technological guard rail that we should put up we should be able to see how it arrived at its conclusions every single time and and that's a fair request I don't know the viability of that request and here's why neither do I neither do I but well well here's why so my kids have always complained and I imagine yours will too when they get into the higher grades and they get to the point where they're doing math they don't want to show them all my kids are very good at doing it in their heads now I'm not I I have to write it down but they've come up with the solution But the teacher is asking them to show their work and and you as a as a person who grew up on web code HTML code and not giving been giving the best practices from an organization that did that being in that wild in space of just writing code it wasn't until late later on where they're like you need to comment in your code right commenting if you asked very important well let's let's ask you know and this is all me supposing this is also a position but if you asked 100 coders do you comment your work I'll bet you 90 of them do not I think I think coders have gone but I mean I bet professional coders like I bet if you ask Todd I bet he comments his work like good ones I would love I would love to hear that answer I would love to hear that answer so especially for people who work in large organizations it's impossible not to code like like I'm thinking that if I were employing coders and one of them didn't comment his work he would be fired pretty soon because you just can't do that in a large organization when you have a large code set but here's also here's also the thing what tools are they using that's just generating code for them so with WordPress we're not coding anymore the code is there it's just we've manipulated it with a wysiwyg on our side to place it yeah a lot of times there's no comments right and and it's the same for me like I try very hard when I'm building a an illustrator graphic to label my things so that it's a it's more a courtesy really of just giving yeah when I hand that off to another designer so they know where those things are but a lot of people don't go into those business best practices because they've been just Freelancers and they've always just done the work themselves and so they know yeah who works in an organization or who you're used to working with in a group will comment but those who are not may not right and so I I would be interested in I mean Todd's experience is definitely working in mostly groups so I would be interested in his answer about you know do they comment and how often they do that and what does that look like uh because obviously the Chachi PT code is not commented it just spits out the exec exactly I better asked it to provide comments on what the code is doing I bet it would but but what I'm saying though is like when you take a digital picture it that digital picture isn't just the image it's the metadata behind it and my thought my thought is that you know mid-journey chat zpt all of these programs should should generate some kind of metadata with each generation of content um yes and there should be a tag on it right right to say this is how I this is so the computer can say this is how I arrived at this conclusion and and I can't I can't imagine a world where that wouldn't be helpful to the engineers as well in training and retraining it um there's another whole piece to this conversation which I'm going to have with some lawyers which is the copyright issue um James and I have been debating how the copyright office actually just rejected copyright for some mid-journey created images and said no you can't get copyright on this which I think is stupid so incredible wrong here here's my thought on that that Frontier and it is a frontier that is interesting that they said no copyright but early on in the mid-journey thing there was a a person who posted that they won an art competition with a piece of art I know what you're talking about it it is it is a pretty piece and it it you know I liked it for various reasons uh but it also brings the question what is art you know and what is that inspiration and you know where does that where does that intersection reside with you know because and I and then this I hate some Modern Art just because it's dumb and like there was a piece that we saw at the high and it was just this square and its name was red and it was red like crazy right anybody and they're anybody in their blind goat could make that you know right it's like uh really is this a piece of art no it's a square that I could insert in my child's build a block thing and and call it good um so where where does art where is that art lie in both writing because writing is Art in my opinion it is I mean we read for sure art all the time for novels and those kind of things so where is the does the art um versus this well technology enhanced art lie from for me and this is where I'm thinking too is chat GPT generated words are those copyrightable right and and and so for me there's this Doctrine in copyright law that it's not sweat of the brow it's that you know the the amount of effort and work doesn't make it copyrightable it's the the you know I think there's even a case law about like how the lightning strikes or even just hitting the button on a camera like the person holding the camera owns the copyright to that or not necessarily right like if somebody else is taking a picture for you and you're telling them how to position everything and then they take the picture you're the one in control and so you're the one creating the art and I just I the way that I think about it especially after having played with mid-journey and creating a prompt and seeing how particular some people are in their prompts and they get certain results like a lot more effort can go into mid-journey style art Than People realize and and and so I definitely definitely think it merits copyright protection but that is an argument that I'm sure I will pick with other lawyers later and also include in this video series it's worthy to be picked because because it is a question I mean the iter the possibility for iterations too is is so fantastic and do you copyright all of those iterations and it brings to to mind the rage and nfts non-fungible tokens with these graphics and art and how they were making all billions of these monkeys and you could create your own monkey with a particular hat and sunglasses and that would be your token but you know what what is the functionality of it and for art the functionality is enjoyment for the most part but there there was a phrase in Creative circles if it doesn't sell it's not creative right so you know what does it does it provide some use Beyond just the feeling and the emotion of it uh and I think a lot of a lot of good design is coming out of it good art is coming out of it on both chat GPD and the the art side of things but it does bring up a lot of questions you know where where those regulations where the copyright law uh plagiarism concerns all of these things where where does that reside and lie I mean I think it's interesting the current lawsuit with stable diffusion and Getty Images um where they're you know stable diffusion fed their library of images to their Ai and and now it's creating these things but but you know as a human I could have looked at each one of those images too and drawn some inspiration from it but not plagiarize them and I go through the web daily looking for inspiration for like yesterday I was looking up inspiration for a poster I want to create with just text and so I was like a typography poster and so I was looking through posters and I captured a couple of those to put in my inspiration folder to look at later when I'm really going to dive into that process so where does that and this is where you guys are going to come in where does that balance lie with these AIS taking inspiration from content that's available I'll invite you to my conversation with the copyright lawyers that I want to invite and uh because because you know taking the screenshots is probably a copyright violation like saving those because somebody else owns that copyright and use now you looking at it and then coming up with your own thing probably not a copyright violation like bookmarking the web page looking at it later but actually taking a screenshot saving in a folder might very well be a copyright violation um and then when it comes to like stable diffusion looking at now when a computer looks at an image it has to make a copy and when you look at an image you're not making a copy right it's not fixed in anything or anything so there's a difference but here's the thing with Getty Images you can download for free a comp of any of those images right right and I can store them on my computer I can store every last one of them and use them for reference so you know where does that balance lie because now like you said ai ai may make a copy or they but their memory is larger they have servers that are going to maintain more memory of right does it undermine the person my memory from having posted the image or provided that policy to download a comp right so my memory is a little short more short term and so I have to go back and refer to that folder if I want that same inspiration or different inspiration where all they have to do is access a folder on server 4452 right you know XYZ database where I have to go uh the file pass where was that folder so I mean I have a library of of just interesting content that I found over the years that I've really loved the design and refer back to it frequently to gather inspiration yeah yeah okay cool well that was a good chat man I appreciate you taking the time uh it's it's it's such a brave New Field Brave new adventure I'm excited to kind of see what happens but I wanted to be part of the conversation especially since I've seen lawyers kind of look at the tool and then give up I think that's not the right approach I think this is the time to dig in and so I wanted to show um and I'll have some other videos that show how a lawyer can dig in um and actually generate usable contracts usable documents um and then I mean everybody already has seen that this is usable for marketing content but I think it's also usable today today with chat GPT not the next iteration for a variety of writing tasks that you might use in the legal profession so I I'm going to be making those videos I'll post them and maybe we talk again about it later yeah let's do it I would welcome any conversation about this just want to learn more and and just like you said be part of the conversation uh because it is fascinating and if you're not part of the conversation how can you guide guide the efforts or the the ship you know so I'd rather have one hand on the wheel with the other four million people hand on the wheel as well to give it a nudge when I can well and to potentially influence you know maybe somebody out there will see this and say oh that's how you could use the tools so um anyway thanks again I would love to help anybody figure out any of these things because it is fascinating and you know that's my personal Mantra is how can I help you choose success today so thanks again for watching that interview as you can tell there are a number of fascinating issues surrounding artificial intelligence uh surrounding legal issues like copyright and those kinds of things but also just surrounding its usage generally and the potential applications for artificial intelligence in a wide variety of Industries including in the legal industry and how you can use that there like I said I and as I mentioned in the interview I've been exploring a lot of these uses and have actually begun to use chat GPT frankly in my daily practice of Law and my daily skills that I Implement throughout the day uh here here at the firm so if you are interested in this topic please like And subscribe because additional videos will be coming out about this topic very soon and I will be sharing with you specific prompts that I use to help me generate structured output for legal purposes for now uh just kind of honor the coming release of the Dungeons and Dragons movie we've gone ahead and released our Dungeons and Dragons prompts which has been kind of a fun project but frankly it has taught me so much about how to get structured data out of chat GPT so I could run that game on the fly but I also can now already see uses that I could use structured data out of chat GPT in a very rapid fashion and a legal setting and so we'll go more into that like And subscribe this video And subscribe to our YouTube channel so that you can see more as we publish more about really interesting ways that you can use chat GPT in your business in your law office and we'll be seeing you here around on our Channel at tingen law um

Channel: Tingen Law

Views: 204

Keywords:

Id: vZ-EOt5uCa8

Length: 57min 28sec (3448 seconds)

Published: Thu Mar 23 2023