← The ChatGPT Experiment - Simplifying AI For Curious Beginners
The ChatGPT Experiment - Simplifying AI For Curious Beginners
From the Vault: Documents and ChatGPT
00:31:23
Loading summary
Transcript1 lines
- [00:00]
A
This episode is brought to you by Progressive Insurance. Do you ever think about switching insurance companies to see if you could save some cash? Progressive makes it easy to see if you could save when you bundle your home and auto policies. Try it@progressive.com Progressive Casualty Insurance Company and affiliates. Potential savings will vary. Not available in all states. Guys, thanks for helping me carry my Christmas tree. Zoe, this thing weighs a ton. Drew Ski, live with your legs, man. Santa. Santa, did you get my letter? He's talking to you, Bridges. I'm not. Of course he did. Right, Santa, you know my elf Drew Ski here. He handles the nice list and elf. I'm six' three. What everyone wants is iPhone 17 and at T Mobile. You can get it on them. That center stage front camera is amazing for group selfies, right, Mrs. Claus? I'm Mrs. Claus much younger sister and AT T Mobile there's no trade in needed when you switch. So you can keep your old phone or give it as a gift. And the best part, you can make the switch to T Mobile from your phone in just 15 minutes. Nice. My side of the tree is slipping. Kimber. The holidays are better. AT T Mobile switch in just 15 minutes and get iPhone 17 on us with no trade in needed. And now T Mobile is available in US cellular stores with 24 monthly bill credits for well qualified customers plus tax and 35 device connection charge. Credit engine balance due if you pay off earlier. Cancel financing agreements. 256 gates 830 eligible for in a new line $100 plus a month plan with auto pay plus taxes fees required. Check out 15 minutes or less per line. Visit t mobile.com hey gang, it's Carrie. Hey. This week we're talking documents. I'm bringing an episode out from the vault from back in March of 24. Going to talk about how to use ChatGPT to create, compare, analyze documents. One of the topics that comes up time and again. And so during this holiday break, I'm finding some episodes that have continued to serve value and provide answers to questions and situations that I get all the time. So I'm going to bring back documents, creating, analyzing, replicating, all that fun stuff. If that sounds interesting to you, I'll see you on the other side of the music. Hey there. Welcome to the ChatGPT experiment. This is the podcast designed to help you better understand what ChatGPT is and find some nuggets personally and professionally that you can put in place right away to help you be a little bit more efficient and help you be a little bit more effective using ChatGPT My name is Kerry Weston. I'm your host. I'm on a holiday vacation here. So I'm bringing some episodes out from the Vault and this week I've got Creating, Editing and Analyzing Documents. First came out in March of 24 and it continues to be a topic that I get a lot of questions about. And certainly since this episode has come out, there's been some updates into what kind of documents you can give ChatGPT and how you do it. But the principles you're going to find here are the same. And there's some remarkable power and efficiency in feeding documents and having it analyze documents. I've used it for legal, I've used it for school, I've used it for business. So there's a number of things here that you're going to get out of it and I hope you can find a nugget, like I said, to apply to your personal or professional needs. ChatGPT experiment.com is the website, is an episode archive and articles and ways of getting a hold of me. But in the meantime, I'll be back in the new year. Hope you enjoy the Vault episode here on creating and editing and analyzing documents. See you in the new year. And as always, until we talk again, do stay curious. Okay, bye. Bye. So, documents, yeah, that's the topic for today. Want to share some things that I've been playing with, some experiments, some things I've heard, right? Really two things, creating and analyzing, right. And I think the couple points I want to make before we dig in is, you know, the name of the show is the Chat GPT Experiment, and that's really what this whole thing is. It's just trying stuff out, right? It's experimenting. And so I'm going to share a couple of examples and. But they're not absolute context. It's not. This isn't the only thing, or if you don't do this one thing, this tip or these lessons or whatever isn't valuable. So I want to give you some homework, at least set you up for the proper frame. Mind, right? Is as I'm sharing, I want you to picture something that might be relevant in your world. I want you to think of a document or a task or a duty or some sort of repetitive action, right? Where the things I'm going to share might be relevant. So even if it's not literally in context for what I'm saying and what I'm doing in the document or the situation, the principle, right. The framework is certainly relevant and can be relevant to you. In your world. So that's really the mindset, that's the assignment, that's the homework. That's kind of the perception and the perspective that I'd like you to kind of listen to this through because I want you to be able to find things that are applicable in your world. And I'm just going to share an analogy of something that I've done in hopes that it can trigger, as I talked about before, that idea treadmill. Get some things spinning in your mind. Yeah. And so a couple things I've noticed in using GPT is, is we want to combine it with some of the other tips. Yeah, that's here in the show. And one of them specifically is background and goals. You know, who am I, what do I need, what do I. Goals, what do I. You know, that kind of thing. Right. So give it background, give chat background, give it details. And then when you feed it a document. Right. Oh yeah, let's back up because I never want to really assume that you know what we're talking about. I want to make sure we're all on the same page here. Right. So I played with this this morning and it looks like the free version now has, if you're using 4, I think the free version got updated to 4 ChatGPT 4 as well as a 4.0, which is the 4 Omni. That's the newest update that came out. It looks like both versions now have the paperclip kind of icon at the where you type in and put your instructions and feedback. Clicking on that paperclip allows you to upload a document from your computer. And something that I've seen recently, this is the update I was mentioning in the beginning. I'm not sure how new this is. I just saw it recently, which is pretty cool, is you can link your ChatGPT account to a Google Drive account and you can also link it to Microsoft OneDrive. Those are the two I see for now. I'm sure there's more coming later, but that's really cool because if you're like me, I've got a bunch of stuff in Google Drive. I have my clients folders, I have writings, I have stuff that I've done from content point of view. I've got proposals. Right. Memos and whatnot. And the ability to not have those be on my computer means I have more access to things that I've done in the past to reference libraries to documents and whatnot. Whether I'm creating or analyzing or just using for a basic reference. I can pull those in Right. So I'm thinking in the future there's going to be more capabilities of doing that. We're going to get into larger data sets and larger access and I can see the path. We're going to be able to use this as a personal library and background of knowledge. Right. So you can beg and borrow and steal from the work that you're doing. So cool little tip there is the documents can now come from somewhere other than your computer. Right. So why would you find benefit in that? So let's get into an analysis or an analogy. That's the word I want an analysis yet of how might this come in handy? Again, here's that scenario that I'm setting up. But if it doesn't apply to you, can you find ways in which this does? Let's say you are in sales and let's say that you get an rfp. Right. There are documents out there, requests for proposals, requests for qualifications. And you really want to kind of analyze this and do your best work in a proposal. Yeah. So you could upload that document to ChatGPT and then have a conversation. You can try to find some insights, trying to find some nuggets. Right. You can try to see how you know language that they're using or words that they're using or phrases that you can ask it for. What are they repeating? What are some of the emphasis. Right. What are some of the things you want to pull out and really pay attention to now, again, combining with previous topics or lessons or recommendations. Let's say you were able to, in that RFP process, have a call with the person that issued the rfp. You were able to ask questions and you were proactive enough to record that. Let's say it's a video call or whatnot and you recorded this. Now you have a transcript. Right. So now you can compare or combine actually the RFP and the transcript, up those both, upload those both to ChatGPT and then ask it to analyze the transcript compared to the RFP and let's pull out some insights. Is there anything that got emphasized in a different way? Is there anything that was mentioned in the phone call that wasn't mentioned in the document? Are there things that you could pull out and draw attention to that you know are important? Right. So you can have it be that analysis side. You want to get in the head of the person. In this case, you don't know if it's going to be an individual or a committee. You don't know the perspective by which they're going to Be looking at your proposal. So the more information you can gather, the more insights you can combine, right? The more data sets you can combine, the better. Let's say you had that conversation but you didn't record it, right? You're, you're not out of luck because another thing that you can do is do a massive brain dump into chatgpt and just tell it, right? Organize your thoughts, get it all out of your head. This is the conversation I had with this person. These are the topics that came up. Just get it out, combine that with the RFP and then say, hey, listen, let's compare, right? Let's organize my thoughts, compare it against the rfp. You know, what are some of the goals, what's being emphasized, maybe more than others? What are your key takeaways, what's being repeated, right? That kind of thing. So analysis of getting in there, because some of these documents are pretty long and if we stick with the RFP analysis for a second or analogy for a second, you know, some of those are templated, right? And so they're just borrowed from someone else's, right? Downloaded off the Internet or they're developed by a committee. So you end up with a platypus, right? You end up with a little bit of everything. And there's not really an emphasis anywhere. So anytime you can compare that or combine that with, you know, another data set, transcript or meeting notes, you can get perspective to tell you what's more important. Because having signs of what's important, right? Having signals and triggers of what's important means you can pull the meaningful stuff out. In this case, you can reverse engineer that and make a meaningful proposal that hopefully will stand out against your competitors. So the ability of pulling it out and tell a quick story here in terms of, you know, triggers and knowing what to emphasize, you know, if you're like me, you've spent time watching crime movies or crime shows, right? And I remember a While back now, 25 years ago or so, I was picked to be on a murder trial. I was on a five day jury here in town. And you know, as you watch those TV shows and movies, right, there's always music, there's always triggers, right? There's always some sort of sign, maybe a heightened pace or voice change or some sort of musical undertones, right? That tells you something's really important and in court there really isn't, right? There's no big payoff moment. When you're in court, everything comes at you in this same cadence, right? This, everything is the same. So when you're reading a document, it's much like that. Everything is weighted the same unless it's pointed out and bolded or italicized. You really need to figure out for yourself really what they're saying. And so from an emphasis from reading into it, some of this stuff could be pretty dry and straightforward. So you can figure out what's important to them when you can figure out what's important to success, right? What's what should you rank, what you prioritized reading? The intent, right. The intent is always difficult. So that's a great way to figure out what they really mean, what they really want. Especially like I said, if you get the chance to talk to them, that's goal too, right? Trying to figure out what success looks like. And so you get a little background. You can ask ChatGPT to help you analyze that. Hey, even feed it. By the way, here's another thing that picked up off listening to other podcasts around the topic is why not combine that document? We'll talk about analysis. And now that ChatGPT, if you're using 4o the Omni, you can have it search in real time for the Internet. Why not do a little work and say, hey listen, this is the situation. I have this RFP from this company, right? This organization, and here's maybe even give it to it, right? Do the analysis of what, of what they want, right? Help me analyze it. And then let's go to the web and let's figure out some recent news, some background about this particular organization. What can you tell me that's relevant to some of the business goals or needs that they have here? Is there anything on their website? Are there any news releases, any articles, any announcements, anything happening that would be relevant to help me paint a bigger picture. So now the document that they've sent, this RFP becomes a 3D kind of look at their company. So you can combine multiple sources, right? Multiple triggers of information. Maybe you've got your conversation, the rfp, you do some research, can bring it, bring it all together. Yeah. So the ability of having it be a problem solving analysis partner from the document to help you get a clear picture of their perspective, what they're looking for, what success looks like, and then combine it with other elements is really tight. That's a great way to get a bigger, more effective kind of action steps in this case, maybe, you know, a more effective proposal. So the more that you can dig in and understand them, convey that you understand them, the more likely of success going forward. But it doesn't have to be that complex, that's kind of a multi stepped kind of thing. Again, maybe you can find some ways to apply variations of that to your world. But what if you just wanted to just get a quick summary, right? Maybe you've got an article or a research report or something lengthy. Right. And what are they? Tldr. Right, the TLDR kind of thing. I know what TL means. Too long. Actually, I don't even know what the doctor. I don't know what the doctor means. Maybe I gotta get, maybe I get some more coffee and I'll Google that later or I'll chat GPT it later. Right, but it's one of those acronyms. You mean it's too long? The thing's too long. Just give me the highlights. Right? Just give me the highlights. So let's say you have an article or a paper, New York Times or something, and, or reading the Wall Street Journal, some industry trade magazine, and you think, hey listen, this would be great if either a, I don't want to read it all, you know, give me the summary, give me an executive summary, or you know, give me a bulleted executive analysis or you know, help me understand it in context of highlights or a certain audience or a certain industry. Maybe you're going to share it with a client, maybe you're going to share it with a colleague, whatever it might be. And here's, here's something else. I remember once where I was working with a university and you know, they said, you know, based on your question, I was asking questions about process or whatnot. And they had a, it was a research center and they gave me a 39 page PDF and y', all, they didn't waste any space. In this PDF there was, there was words front to back, left to right, top to bottom, you know, research data, heavy, heavy content. And I enjoy to read, you know, I enjoy reading this as much as the next person. But you know, it can get a little much, right? You can get a little much. So I fed the PDF in and I said, here's the context, here's my position, here's my reason for getting this. These are the things I want to pull out of it. Can you give me a summary? Divide this up into sections, give me some bullets, help me understandability. And it did that. So analyzing documents from multiple points of views, multiple reasons is an excellent way of using chat. I found some terrific value from that. And documents doesn't have to be words, documents could be spreadsheets. So another use here that I've played with is uploading a spreadsheet and asking for help in reading patterns and reading analysis. In fact, I just saw this isn't relevant to this topic because I don't know much about it. Maybe some of you have, but I saw that there's now a plugin for. I don't know if it was Excel or Google Sheets. It's a browser extension, so it must be Google Sheets, I would imagine where you can bring an AI plugin to Chrome. And as you're using Google Sheets, I'm guessing you can actually write formulas. You know, you tell this plugin what you needed to do and it's going to show you the formula right off the bat. But more than just formulas, right? More than just the stuff that we don't know how to do from a formulaic point of view. This one had like tell me more about this topic. So they had columns of or rows of countries and by highlighting, using France for instance, as a variable, you can say tell me something about France, right? And it puts a one sentence summary about France and then you can copy and paste that command to the other countries and it gives a one sentence overview. You can talk about what's the capital. I mean you can do counts. So I'm only talking from something that I saw a passing video on. I don't have any experience in this, so I can't share how to's and firsthand knowledge. But just to say that the ability of interacting with data, regardless of whether it's numbers or words, is a very powerful way and a very time efficient way of using right chatgpt and we spend so much time thinking about how to use ChatGPT to create something that, that I don't hear enough about people using ChatGPT to analyze and to break down. Right. And to give insights and so, you know, think about this for a second. You take that document, whether it be a spreadsheet or words, and you layer it on top of perspectives that we've talked about before. You know, take this information and rephrase it like I'm an 8th grader or I'm a brand new newbie or I'm a rocket scientist and I want to know all the details, right? So getting real analytical, real deep or just getting summarized, right, Just getting high level executive summaries. Yeah. So it's really, it's an interesting, it's just interesting to see where we can take this and I hope that this kind of generates a few. A few for you. Yeah, it's kind of like a second brain, right? And getting into the analysis and the insights. And it doesn't always have to be linear, right? So meaning it doesn't always have to be. Here's a document. Do this, right? You could layer multiple things on top of it and having do some really robust and powerful ways of pulling and analyzing and looking at things in different way, getting meaningful insights, especially when it comes to trying to understand for another purpose or another perspective. So layering, layering, layering. Give it background, give it goals. What did success look like? Yeah, it's really, it's really powerful. So I hope you see how this might be powerful in the world that you're using. And as always, if you've got a case study here, you know, share it with me. You know, if you've got something cool, if you found a specific use, the best use, I've heard this term thrown at a couple of times, AI literacy, right. Which is basically just learning how to use the basics, the principles, right. That we've talked about, the evergreen kind of guiding principles. So if you've got something that has applied to your daily world, to your daily life that's been useful and practical, share that. Because that's the whole point. It's not about learning this software. It's not about mastering a set of computer prompts. It's about can this thing, can this problem solving partner do something meaningful, productive and practical that means something to me. Right? And if the answer is yes, then that's really cool. Hopefully you get value out of that. Right? So how about, here's another use. How about for training? What if you have to share your experience at scale? Let's get into creating here a little bit. We'll move beyond analyzing. What if you have to replicate something that you do and you want others to do it as well, or, you know, you want to create like a learning library or a resource library for clients or co workers or for onboarding new employees, right? Maybe you're creating a knowledge set to help people get trained at your company. You know, the ability for you to create meaningful documents and meaningful instructions. You know, ChatGPT is super powerful and super helpful for that. Let me give you a specific. For instance here, just we can kind of paint the picture where you can go, let's say there's a client report or some sort of structured document that you need to replicate over and over again. For me, I probably mentioned it here before. A document that I have is called a recap memo, and it's a memo that I put together after an initial sales call and of course, the information in the memo changes. It always varies depending on the business and the situation. Right. But the structure is the same. It's something that can be repetitive from a section to section and format point of view. It's very structured, very repetitive. And if you remember, I think it was episode 31 and I'm layering multiple things here again, but we talked about custom GPTs and some of the initial uses that I've been experimenting with and finding uses for custom GPTs, and I think one of them was the recap memo, now that I see this out loud. So this is kind of a repeat performance of the recap memo here on the podcast. But if you wanted to create a pattern based on things you've done before, you could simply share it with ChatGPT, upload it like we've talked about, and just say, hey, listen, I've got this document. I want to do this over and over again. I'm going to provide you samples with it. And it's always great if you have more than one, right? So if you have more than one sample of whatever it is you're going to feed. And I say that because anytime you have more than one, you start to have a pattern, you can start to look for consistencies and the reasoning, right? And so as you upload it, the more the better. And when I went through this exercise, I think I gave it three or four, and I think they were in Google Drive or they're in Google Drive now. So I could have linked it, but I uploaded them before, right? And I just asked it, hey, listen, take these documents and could you break these down for patterns, look to see what I do over and over again, you know, what do I replicate and then what other unique variables, you know, and break down the sections and break down the logic behind it and, you know, give me your best analysis. And it did. It gave me a skeleton, if you will, right? It kind of. It said, you know, here's what I see, here's the outline, right? And I said, great, that's great. And I've seen this before because when I went in to do the custom GPT, I did the same kind of thing, thing, but here's what I haven't done before, which is really cool in this particular exercise, getting ready for this episode, and from the training and sharing and guiding side of things, this is going to be really powerful. So I took that analysis and I basically, it said, here's what I see, right? So it said, here's the heading and here's the Bullet points, right? And here's the logic and rationale behind those bullet points. And it was good. You know, it was a model. Then I said, great. And I always want to compliment it, right? I always want to compliment it because the robot's going to take over the world at some point, and I want to be on the good side. But then I said, hey, here's my goal. I want to create a guide to teach others how to do this, okay? So this is where it shifts from my custom GPT. My goal with custom GPT was to understand what I was creating and why it went in there so I could give the custom GPT instructions, do it itself, right? So with minimal information or an interview, it could create us on its own. Now, what I want to do is I want to teach other people, okay? I want to put this into a training document. I want to teach and guide others how to do this, right? And because I'm going to pause here for a second because I don't want to create dependency on chat GPT for everything that we do, I think that's an important aside here, is that, yes, it's wonderful to create custom GPTs and configurations and instructions, but if you're only passing, and this is comes into play, if you're teaching and sharing with others, if you're only passing that code, that GPT, the configuration, the output of what you did to somebody else so that they can use it, the logic and reasoning and insight and value of why and how they can do it themselves goes away, right? It's like passing somebody a calculator and telling them exactly what buttons to push and then wondering why they don't understand the concept of the mathematical equation if they don't have the calculator, Right? And so I've been probably more aware of this recently. We have a very encouraging environment around curiosity here at the office and getting people to use tools and sharing and learning. But I also don't want them to become dependent. I need them to be critical and independent thinkers, Right? So what I'm doing here is I am saying this is the document that I create, and these are the outline and insights and things that you've seen. I want you to be the mentor. I'm talking to Chad now. You. I want you to be the mentor, and I want you to break out each of these areas that you see in the document and give some guidance, give some context, give some suggestions and recommendations in a teaching manner to help people understand not only what these sections are, but the rationale behind it. And then again, take a look at the, from a consistency point of view of your analysis and see if you could add any additional tips or insights. Again, I'm talking to Chad as I'm doing this. That would be helpful to have folks better understand how to create this on their own. Let me repeat that. So not only did I say I want you to reverse engineer what you see here and put some analysis and insights, but I want you to look even further, be a mentor, take on the position that perspective as the guide and add any additional tips or insights that would be helpful to have folks better understand how they can create this on their own. Right? And it did it and it was really good. And it gave me like four different parts, you know, and it broke down the sections and it gave me rationale on what I would put there and why. And then it gave me key considerations and context and went back through each one and it walked me, it taught me and it said, consider this and consider this, and these are importance. And here's a key insight, here's a key takeaway. It gave me a variety of information and that allowed me to see how powerful this could be if I was creating a training doc for my company. And I don't think I've mentioned this before, but in the essence of documents, While you're using ChatGPT, if you get an output that you like, you can simply tell it, hey, can you put this into a document format that I can download? So let me just pause there for a second. I've gone through a memo, a document that I use over and over again. I've fed a few of them to ChatGPT. I've asked it to analyze for structure, right. And rationale as to what type of information is in there. And then I've asked it to take that and create a training doc so that outside of CHAT GPT, I could teach others how to do this, what to think about to guide them. And then this last piece is you can tell chatgpt when you're done, hey, I want you to create this information into a document that I can download. And it will, it will actually create a file for you that you can download to your computer. Right? So I played with this for a little bit and I went through it and gave considerations and insight. And I told it that you need to consider two things. One, that the person using this document is familiar with the reasoning behind the document, but they're not necessarily used to or experienced in actually doing it this way, meaning the information that they're Gathering. Right. The insights that they're putting together is not new. It's just a format and purpose behind the document that's new. So I want them to see how to organize their thoughts and how to organize that stuff so they could actually use it and put it into this format. Right. And when we were done, I said, could you please take this and create it into a file that I can download and that helps me. Right. Put them into a learning document or some sort of shared folder that becomes shared intelligence, Scaled intelligence. Right. And the last piece I didn't share with you, I should have thrown this in here, is as part of that document, I always like to include a section called what does success look Like? Because I want to wrap it all up. Because it's not enough to just do these things. Back to the TPS report, right. From office space. Doing it over and over again isn't the goal. Right? Getting it done isn't the goal. There must be some reason, there must be some measurement of success. There must be something that we can say, this is worth it. And sometimes that's conveying information, sometimes it's getting buy in, sometimes it's just organizing thoughts. But there has to be some reason that we're doing it. Otherwise we're just hamsters on a treadmill. Right. And so that was cool. The ability of teaching in my absence. Right. The ability of taking the expertise and experiences and routines that I have that end up being a document of some sort and then feeding it back in to ChatGPT and then creating recipe. Right. Insights, guidance. Right. In a meaningful and teaching way so I can have someone else see how to do this for themselves. I think that's, I think that's really cool use of chat GPT and I hope a few ideas came to you there. So that's it. Right? So your curiosity is key. I always talk about the best way of moving forward, the evergreen guiding principles of being patient and looking at perspective and, you know, and asking the right questions. It just comes down to being curious and that's going to serve you well in all aspects, regardless of what tool you're using, whether it's ChatGPT or anything else that's out there or what features come around. Right. Or variation we're at. Right. There was some big updates made a couple weeks ago, but the principles are still true, right? The principles. There are some new features and whatnot, but if you apply the principles, you're always going to be well served. And so that's how we'll wrap this up. Is that curiosity is key. I appreciate you being with us and until we talk again, do stay curious. Okay, talk soon. Foreign. Deserves to be connected. That's why T Mobile and US Cellular are joining forces. Switch to T Mobile and save up to 20% versus Verizon by getting built in benefits they leave out. Check the math@t mobile.com switch and now T mobile is in US cellular storage savings versus Comparable Verizon plans plus the cost of optional benefits, plan features and taxes and fees vary. Savings with three plus lines include third line free via monthly bill credits. Credit stop if you cancel any lines. Qualifying credit required. It's the most wonderful time of the year, and Valpak is here to make it even better. This month as you sip through holiday mail, don't miss the blue Valpak envelope. From dining to holiday shopping, there's a sleigh full of savings in your mailbox, plus a chance to instantly win $100. That's right, you could find $100 Christmas cash inside. Want to save even more money on what you love? Go to valpak.com for local coupons and offers. It pays to open Valpak. No purchase necessary for instant win. Voip were prohibited. Prices are randomly inserted. See specially marked Valpak envelopes for details.