 on the machine, but I'm gonna assume that we are live and everything is going okay. So I just wanna say welcome to designing with AI, ask me anything with me and Tibi. Thanks for joining wherever you are and whenever you are. Let's start with quick introductions if it's your first time joining one of these lives and I'll go first. I'm Chris, a product designer that used to be ahead of design but I quit my job in 2020 and started a UX education company. It's called UX Playbook. So I help designers of all levels whether it's building your portfolio or thinking about managing your first design team. The idea is to take you from zero to unicorn designer wherever you are in your journey. And then we have my special guest today that we're gonna delve into AI, it's Tibi. So let's welcome him to the stage. Hey man, how you doing? Oh, what up everyone? Okay, so I'm Tibi and I started like back in 2010 as a designer doing logos on Fiverr for $5. And after that, I stumbled upon trying to do some dropshipping for a while. It didn't work out anyways. Long story short, I went back to design, become a UX designer, worked for Huawei HP and right now I transitioned to a full-time content creator and talking about AI and how to leverage as a designer AI tools such as JGPT, cloud, perplexity, mid-journey, you name it. So yeah, that's pretty much it in a nutshell. Hey man, that's cool. So Tibi, let's just, before we dive in, just wanna talk about what we're talking about today, right? So the topic is designing with AI. Like the big problem, at least in my opinion is the AI buzzword is getting thrown around a lot, right? But I guess here what we wanna discuss is, what the hell is it? How do we use it and everything there? So let me tell the audience quickly about what to expect from this live, okay? So today's topic is designing with AI. So hopefully we can delve into some actionable things to get you started. How and what and where to use these tools, okay? We'll be touching into things like what is it? Why UX, some workflows, some process and tools and if we have time, future of predictions. And of course, we encourage you guys to drop anything in the comments and we'll hopefully get to it if we have time. But yeah, just ask any question you want as you're listening to this, just comment. And also to start us off and make this interactive, why don't you drop where you're dialing in from, okay? And that's all I have to say, but I guess I'm gonna go to the actual ex, not the expert as we'll get into this, but someone who's better at AI than me. So you can think of this as a conversation of two designers trying to figure out what this AI world is. Tibi has been using this tool and diving way deeper than I have. So you can think of this as a casual conversation between two designers who are fascinated with technology. So let's just bring Tibi back in. I guess the first thing is where do we start? Where's the best place to start in terms of talking about AI? Yeah, so that's a really good question, Chris. So from my point of view, I was watching the other day, like a clip on TikTok or whatever, and it was super funny. It was something like a person on the street asking other people, what is JNAI? And nobody knew what to ask. And the reporter, the guy got like super funny answers back. And I was thinking, bro, this is like super smart, you know, in order to start using this kind of tools, I mean, yeah, you can jump straight in and fool around and see what happens. This is how I started as well. Like one year and a half ago, I was like super curious about new tools and stuff like that. I stumbled upon Dali after that charge EPT. And this is how like I actually started to get into this world, let's put it this way. But in order to, because things are evolving and are evolving super fast, I think right now is the best way to jump into this. Like I don't want to say the right way because there isn't like a right or wrong way. And as you said it, we're not like experts. And I don't want to call myself like an expert. I'm just like a super curious designer about these new tools and this new technology that yeah, it's revolutionary and it's gonna change the world. And like to get back to your question, I would start with the basic terminology. Like it's like when you're building a house, right? So when you're building a house, you need to have a strong foundation. And this is when you're learning a new field. You need to have like a strong foundation and you need to know the basics, the ABCs. And from my point of view, I would like to, since I'm not good at remembering stuff like a small parenthesis, I'm gonna, I'm having like, I'm building a second brain on notion. And I have a bunch of notes that I've popped out right now and I'm gonna read. And after that, we're gonna expand on those notes if that's okay for you as well, Chris. Yeah, let's do that, let's do that. Yeah, cool, perfect. So let's start with what is artificial intelligence and what is the difference between AI and machine learning in the first place. And after that, we're gonna move on to what is gen AI, what is deep learning, what are large language models. Maybe we can talk about, because the other day I was invited at Jeremy's podcast to talk about these kinds of things. And he asked me like a super nice question and I did my research after that because at the moment I didn't knew what to respond. He asked me, bro, okay, so we're talking about large language models, but there are any small language models, for example. And yeah, I did my research and I can say right now there are small language models, are domain specific language models, but we're gonna dive deeper into those as we proceed. So first things first, what is AI and what is the difference between AI and machine learning? And I'm gonna read this one and after that we're gonna expand on this. So one way to think about AI is like a discipline. For example, let's say it's like physics or like math or English or name it whatever. And AI is a branch of computer science that deals with creating intelligent systems that can reason, learn and act autonomously. And so AI is the theory and development of computer systems to be able to perform tasks normally requiring human intelligence. And that being said, we can think about AI as a discipline. That's the take. So AI is the discipline. Now what is the machine learning? Machine learning is like a subfield, is a subfield of AI. And it is a program or a system that trains a model from input data. And this train model can make like good guesses about new data or data that it hasn't seen before, but this new data is similar to the data used in the train model. So unlike traditional AI, which might rely on hard coded rules and logic. Like for example, if this, then that, if four legs, then it's a horse. You know what I'm saying? Machine learning is centered about the idea that system can learn from data, identify patterns and make decision with minimal human intervention. So AI is the big umbrella that covers all the ways of making a computer smart while machine learning is a specific method under this umbrella where computers learn from data to get smarter in a nutshell. So that's the difference between AI and machine learning. Will that make sense for you Chris? That makes a lot of sense. Okay, so what about gen AI? Is that also a big umbrella? And also I hear this phrase called deep learning as well. Can you tell us the difference between like those things? Sure thing, Matt, sure thing. So gen AI refers to a type of artificial intelligence that can create new content. And it's called generative because it doesn't just analyze and understands information. Like recognizing images or understanding spoken words. It actually generates, that's the beauty of it. It generates new data that didn't exist before. So that's the thing with gen AI. And since you're asking me about deep learning, like deep learning fits as a subset of machine learning methods. And this one, I won't go in depth for the sake of time, but deep learning uses artificial neural networks to process complex patterns than traditional machine learning. So yeah, okay, okay. So, okay, so AI is the umbrella. Then you have machine learning and then you have deep learning under that. And then gen AI is like one type of thing that generates a bunch of other things. Yeah, so gen AI just, it doesn't just analyze and understand the information, but it actually generates new data that didn't exist before. Right, and the gen AI that we're using today, let's just say chat GPT, the kind of most famous one. Is that deep learning? Is that using deep learning to generate stuff? Yeah, that's a good question. I mean, let me think about it. Let me think about it. Sure, sure. Yeah, I mean, that's the thing as well. I mean, maybe it doesn't matter really, but because these concepts are pretty, also hard to grasp, right? What exactly is deep learning? What exactly is machine learning? So, okay, well, let's talk about what might matter, which is like chat GPT is this language model. Yeah, okay, so what is that? Maybe that's a better way to look at it, yeah. Yeah, sure, sure, sure. That'll make more sense. So like large language models, so a large language model is a type of AI over artificial intelligence that can generate human quality text. And we can think of this like in easy terms, let's think of it like as the big brains of language models. And they know a lot because they've been taught using massive amounts of information. And they can handle, of course, a variety of tasks like writing stories, answering questions, or I don't know, even making jokes, but that's what she said. Yeah, there you go. And going back to my conversation with Jeremy the other day about small language models. Yeah. Again, I did a little bit of research about the small language models. And it's fascinating because these are the little siblings of the big models. And they're not as powerful, but they're quicker and needs less energy to run. And they're great for simple tasks or when you don't need the full power of a big model, you can use a small language model. And like a fun fact, let me find it somewhere in here in my notes. I was finding something yesterday. Oh, okay, so in here. So ChagGPT, because we're talking about ChagGPT, right? Right. ChagGPT cost over 700K, like dollars per day. And it's running on, because it's a large language model. And it's running on 285K CPU cores and 10K GPU cores and network connectivity of 400 gigabytes per second per GPU server. And yeah, this is creating a giant carbon footprint and raising concerns about the feasibility and further scaling. So yeah, this is absolutely nuts if you're thinking about it, right? I never thought about this one in this way until yesterday, when I've stumbled upon this fun fact. Yeah, so I just bought a new laptop because literally this laptop that I'm using is dying, but I have 30 GPU cores. You said it like 10,000 GPUs, that's insane. So like also the question there is like, okay, well, what's more useful, right? Because like blockchain also requires a lot of power and a lot of money to run. And then it's like, well, AI also is the same. I mean, that's also why they decided for Microsoft to invest, right? Because they needed the capacity, right? They needed someone to fund this experiment as it was a nonprofit before open AI. So anyway, that's probably me sidetracking, but it's like, well, which one would you rather have? Blockchain or AI, right? And it seems to me that AI would probably be the way to go as it's providing way more value to the average human than blockchain is. But anyway, okay. So there are a bunch of language models, right? Yeah, like vision models. And yeah, there are a lot of language models, but yeah, so these notes that I was taking like for a little bit of context, let me tell you first things first, how I started to get into this AI thing. And maybe, I mean, of course, you know, but for our audience, maybe some of you guys know that I wrote a book, chat GPT for designers, everything you need to know about parameters and prompts. And I wrote this book. I started to wrote this book last year and I've published it like officially in February this year, 2023. And I started to wrote that book like from a pure curiosity of mine, I just stumbled upon chat GPT. And after that, I was testing a lot of stuff. And I didn't know what I was doing. I had absolutely no clue to be honest, but it felt fascinating. This new technology for me, it was mind blowing like last year. And I started like a personal journal and that was the beauty of it. I started as a personal journal on Notion and I was taking notes. Okay, so today, this is what I'm testing today, this, this and that. And based on my test, it worked out pretty well or it flopped. And after a while, after like a couple of months, I stumbled upon also parameters. And I was understanding that you can fine tune the outputs from chat GPT even more. And I was in that second, I was mind blown. And I was, bro, I need to do something and maybe to create a book or put all these, these informations of all my tests together, like in a comprehensive manner to make sense for others, not to go through the same struggle that I'm doing, these back and forths and fails and whatever. And right now, right now, it's super important. I don't have the link right now but I have it somewhere saved. I need to find it. OpenAI just released a new blog post with like the official ways of prompting. And I'm super happy to see this because what I was talking like one year before about how to use prompting and what are the difference between prompt design and prompt engineering and different type of prompt structures and parameters and so on and so forth. They're saying the exact same things that I was saying like one year ago. So my brain works the right way, I think, you know? And this is why I'm super happy about it. But we're gonna send to the audience as well, that link, they need to check it out. It's official from OpenAI. So they know what's up and they need to check it for sure to learn how to get better responses with AI. Yeah, so like that's actually a good segue to what we wanna define because the conversation here is like defining fundamental terms that help you kind of understand this world, right? Like demystifying all this stuff. And you just mentioned two words that people hear and throw around all the time. The first one is prompt engineering. I think that one is the more popular one. And then there's prompt design. And I didn't know there was a difference. So is there a difference? Why is a prompt engineering, why is one's why prompt design? Me as a designer, I want it to be called prompt design. So why is the engineering one more popular? Anyway, just tell us what it means and let's dive into that. Okay, so the difference between prompt engineering and prompt design. Let's talk about, I'm gonna read like the boring one. And after that, we're gonna discuss it in simple terms. The boring explanation, okay? First, and we're gonna start with prompt design. So prompts involve instructions and context, prompt design involves instructions and context passed to a language model to achieve the desired task. So in simple terms, this in simple terms, think of a language model as a chef, like in a kitchen ready to cook something, okay? And you as the customer need to tell the chef, in our case, charge you PT, what do you wanna eat? So you give the chef a recipe or an order, like please make a cheese pizza. Let's put it this way, okay? And this order is what we call a prompt. It's like a recipe or a set of instructions for the chef. In our case, the language model to follow. And the term design counts from how you create or put together this set of instructions. And it's about deciding what to ask the chef to cook, just like you decide what to ask the language model to do. This is the prompt design. Okay. And moving on to prompt engineering. And I'm gonna read the explanation, the definition of the prompt engineering as well. And after that we're gonna explain it in simple terms. So prompt engineering is the practice of developing and optimizing prompts to efficiently use language models for a variety of applications. And now going back in simple terms, you suppose, like suppose you learn that the chef makes better pizzas when you give more details. For example, like please make a cheese pizza with thin crust and extra mozzarella because we love more mozzarella, right? Okay. And this process of figuring out the best way to give the order, to get the best pizza is prompt engineering. And it's about refining and optimizing your instructions. Or in our case, our prompts. So the chef, Chad GPT understands exactly what do you want and can make it perfectly for you. Got it. Okay. So prompt design and prompt engineering like are closely related concepts in this natural language processing. And like in simple terms. And both involve processes of creating a prompt that is clear, concise and informative. Okay, okay. Well, okay. Let me ask you this and maybe a contrarian and put you in a difficult place to be, which is. Okay. Then it seems to me that design is like the term design is oversimplified, right? Because it's more like a general concept. And prompt engineering is more like a specialized concept. Right. Okay, okay. Got it, got it, got it. And so it's closely related like product development, usually both are hand in hand to produce this. So yeah, what I know, look, so I can put it the other way. So while prompt design is essential, prompt engineering is only necessary for systems that require a high degree of accuracy or performance. So in prompt design, you can go wild and whatever, whatever. But if you want like super high degree of accuracy and performance, like in our field, in UX, we need something like, you know, super, super specific. We need to know a little bit of prompt engineering as well. And from my point of view, I see some like amazing results. If you know how to structure your prompts the right way, how to fine tune your parameters the right way and stuff like that, you know, it will help you like tremendously with with, with charge APT at least. Yeah, yeah. Like it's the age old argument, right? Like should a designer learn to code? Like you don't need to, but you'll build better products. Yeah. And about that, about that, like making a small parenthesis about the, should a designer learn to code? Yeah. Yesterday, also yesterday I found a new tool because we got to dive deep into some tools later on yesterday, but I want to discuss a bit about this one. It's called GPT engineer dot app. And everyone can search a super simple GPT engineer dot app. And it's a rapid prototyping with speed you haven't seen before this is their tagline. And you can build an actual app with code from prompts. And it's kind of mind blowing. I want to test it out. This is what I don't want to go super in depth with this one. I didn't have the chance to properly test it out. But from what I've seen, I find it absolutely fascinating because I don't know how to code either. Unfortunately, I know a little bit of HTML and CSN, but super basic not to die. But other than that, yeah. I don't know how to code. Yeah, I think, I think what I was trying to get at was that it seems to me that, I don't know, maybe it's an engineering heavy field. Like you're talking about deep learning, right? You're talking about... No, I know what to think. That's the problem. Wait, wait, wait, wait. Just, just... Sorry, I got excited. Sorry. But like, you know, as a designer, right? Like sometimes that design gets overlooked because we care about the technology more, right? So therefore, like with the differentiation between the product design and engineering, it seems like, okay, you just tell the machine what to do. And it'll do it. But then with the engineer, it's like, no, you need to like give it parameters. You need to like really fine tune that. To be fair, that could also be called prompt design, right? Like spirit of iteration, testing what works, what doesn't, right? So like, it's just funny how these, I don't know, who, like, because it's a relatively new field, who named this? Was it an engineer, right? Like, or was it a designer? So to me, that is a quite a good discussion topic because within tech, usually engineering is that thing that sort of people focused on first, like Google's engineering heavy, whereas Apple is different design first, really. So, but then now both of those, I guess, specialities or expertise are now considered very much next to each other, right? But I reckon like prompt design and prompt engineering still don't feel like it's like level playing field, but that might just be my nitpick of like, you know, being a designer. But anyway, do you agree or you like, no, they're so like similar, there's really nothing to it. Or do you think like, do you see the sort of gap between design and engineering? Or do you think they're the same shit? From my point of view, at least right now, because that's the beauty of it. Since it's a super new field, it's continuously evolving. And we might not be, because I was looking the other day of some of my older posts about ChargePTNA, I am prompting and stuff like that. And most of them are right now, not like irrelevant, but they don't have like that bigger impact because everything is moving so fast. And what I was explaining like one year ago, right now it's like something super basic, but back then it was like mind blowing. And I feel like right now AI, it's this new field like internet was in the late nineties, early 2000, you know? And we're still exploring this new field. And yeah, going back to your question, as all I can say right now from experience that I have until this moment, yeah, prompt engineering and prompt design from my point of view are kind of similar. But who knows, maybe in the future we're gonna change our minds. So but until now, yeah, I feel the same. Like they're not that different. Yeah, they're not that different. Yeah, yeah, and maybe I'm just being one of these designers who really care about terminology. And but anyway, let's move on. What else did you wanna talk about in terms of prompt design and prompt engineering? So the prompt design, I guess we pretty much covered the differences in simple terms about prompt design and prompt engineering. And let me think, maybe we can discuss about why to use AI in UX. Yeah, that's a great topic. I think loads of people would know like why and then maybe how afterwards. But yeah, why, okay, that's a good question. Why would you use AI in your UX work or workflow? Yeah, so a couple of days ago, I do have the link that we gonna post it in the chat as well. I stumbled upon this amazing article from the Jacob Nielsen itself, the one and only on Substack. So it was something like, I don't recall the exact name, something getting started with AI for UX designers or AI in UX, something like that. Anyways, and I was finding pretty fascinating because these are the same three things that I was thinking about and I was like using them as well for myself. And Jacob Nielsen is also saying this in this article. So it's like matching what I'm doing with this article as well and I find it like fascinating. So there are three main reasons to use AI in UX in our field. One is to increase our productivity as designers. Second one is to improve the quality of our work. And the third one is to enhance our current skill set. So that's pretty much it, at least for the moment. Or at least this is what I'm using like AI tools for right now. Okay, then the question is, how the hell do we increase our productivity, improve the quality of work and enhance our current skill set? Yeah, so how do we do that? The first thing that I absolutely love is the ideation part. And the ideation part like things it used to take days, you can almost get these ideas instantly with AI. And bro, come on, we're living in the future. I mean, I was struggling for days to find the better ideas and whatever concepts and so on and so forth. And right now, boom, boom, boom, I have a solid start. I mean, not like a super, super solid starting point, but it's pretty solid, you know? And it's a start at least. And you have like some directions to explore. And for me, it's come on. I'm learning so much time right now. That's true, like, yeah, because like we're all like looking at random sources on Google, also when we're trying to research for the other part, like ideation, I love that. Like you can just be like, give me 10 ideas for X, right? And it will just spit it out versus you kind of doing crazy eights or whatever. And then research is also a good part of it, right? Where you're saying, okay, well, like, how much is the MBA worth? Who are the most paid players? What about which, who are the teams? Who are the own, like you can basically just keep on asking it questions without actually having to go into each individual link and source check. But there's also one caveat to that, which is like sometimes it might just kind of be delusional, right? You know what, you know what about that? Sorry to interrupt you, but this is super, super important. Today, two days ago, I did a post. Everyone can check it out on my LinkedIn. And I was like searching randomly, searching on Reddit, like whatever. And I've stumbled upon in the chat GPT subreddit, these posts that blew up, chat GPT stealth release GPT 4.5, amazing, whatever, whatever. And nobody tried to actually only one person in that, in the comment section, tried to actually fact check this one, you know? Because of course, it's super simple to ask chat GPT, hey, what's your exact model? Tell me what's your exact model? And yeah, it can hallucinate a lot. And that's the problem, chat GPT is a super good liar. And that's why it's super important to always, always single goddamn time to fact check everything. Super, super important. And that guy, and this is why I found it fascinating. And nobody else at least tried to read most of the comments to see the pros and cons of this release in quotes. That one, it was- Wait, one follow up question there is like, can you ask chat GPT to fact check itself? You can apply like it's called emotional pressure. Okay. Like it's a new concept, I'm still testing it out. I've just stumbled upon recently about this one and I find it like super nice. I also have like a research paper that I also can share to the audience about emotional stimuli and emotional pressure and how to apply emotional pressure to get better outputs from chat GPT. And one thing that you can like try to do this with emotional pressure is to ask, are you sure? You know, and you'll receive, but the thing is because this is super important. I don't want to talk most only about chat GPT because it's super important as a designer to know what other tools and what is the best tool to use for a specific task because chat GPT might not be like the best for every single thing that you need. And for example, I'll go with perplexity if I want to fact check stuff and receive links from internet or Bing AI, Bing chat or whatever else. But yeah, a chat GPT is good for other things. For example, Claude, Claude AI. It's good for long form content to write better LinkedIn article post hooks, whatever name it to generate ideas. And yeah, maybe it depends. Every single tool has its pros and cons. And this is why we always need to be aware and not to overly rely on a single tool. Yeah, what are your, I guess, favorite ones within the, yeah, someone actually asked, right? What tools slash platforms enhance your work process and results? Like what are some typical ones that people can use or that you've used in your workflow? Like chat GPT is obviously one of them. But yeah, you mentioned it. Another one that I'm heavily using it lately and I'm talking only from my point of view, but it is not only from my point of view. Most of the people can use it as well. For example, I'm using this tool is called fireflies.ai and be careful not to be mistaken with firefly from Adobe, the image generation one. Fireflies with s.ai. And it's an AI not taker to transcribe, summarize and analyze your meetings and conversations and chats. And it's like a personal assistant, bro. And it's mind blowing for me. It's absolutely mind blowing. I mean, you don't need to like, you can focus on your actual conversation and to focus 100% on the discussion and have like an AI to summarize and transcribe everything and make a summary and have more than that. And the good part it's free for 10 conversations, but the thing is the trick in here, what I'm doing not to pay another subscriptions because I have way more subscriptions than I need and want. So I'm trying to minimize this kind of things. I forgot the word in English. Let me one second to get it back in my head. To sync, yeah, that one. I'm syncing fireflies with notion at the end of each conversation. Oh, so this is free. This integration is free. You don't need to pay anything at all. And at the end of each conversation, you can integrate it directly from fireflies or use Zapier. And for 100 zaps, it's also free. So this is also nice. And you can transfer all your summaries and notes from fireflies after you end up the conversation directly in your notion second brain. And for me, this is game changing because after that you can export them and import them in chat GPT. And for example, I had a lot of conversations lately, more than 20 the last month with amazing content creators and we're discussing about strategies and all that stuff, how to write better content on LinkedIn and so on and so forth. And all these conversations, I can take them, upload them into chat GPT data analysis and start playing with it. Tell me what are the best things that I can improve to whatever, whatever, you know? And so this is how you can use fireflies, integration with Zapier, push them into Notion, from Notion export them, put them into chat GPT, upload them into chat GPT and boom. This is like one workflow that I'm using, for example. Right. And there is like a whole March to AI, right? Leveraging open AI or other companies, large language models and like, for example, right? Framer has their AI tool, so prompt to a website. Miro has Miro Assist, FigJam has their own AI tool to sort sticky notes. So they're actually now companies are racing to embed AI tools within your workflow. So sometimes you don't need to change your workflow. You can just be like, I'm using the same tool, does it have an AI feature? Like Butta has this recap that is exactly what you're saying, where it's transcribing everything, but Butta is also a workshop tool that is like video conferencing tool, records it, it transcribes everything, and then there's like actionable steps. So I think that's gonna be like more and more prevalent. So instead of us natively looking out, like maybe Slack will have an AI tool soon and it'll just give you a recap of like, what happened in the team this week, right? In the chat or something like that, which I think will be super interesting. For sure, but like truth be told, and this is only my like personal, I don't know how to say it, personal battle with Framer. Don't get me wrong. I absolutely love Framer from the bottom of my heart. My website is fully built in Framer. I'm a paid client and all that stuff, but I feel that AI, it's like right now it's a gimmick. And from my point of view, again, you can do that much with that AI tool. And they should focus for example, on the CRM. Oh, CMS, sorry, not CRM. Sorry, my bad, CMS. They're CMS with the blogs. Right, the content management system with the blog. Sorry, sorry, that makes no sense. Also analytics as well, right? Yeah, I mean that they have like some big gaps that they need to fill in. For example, why I'm building a website. This is again a small parenthesis in this entire conversation. Why I'm building a website to get traffic and promote my services and whatever, get some clients and make money, right? And most of the cases, most of the cases. And for example, right now, my problem with Framer is that I'm trying to write articles, write SEO optimized articles, and they don't have a lot of stuff to help you out with the SEO part. But they focus on this AI thing that is kind of worthless from my point of view. I'm sorry, Framer, bro, I love your product. But from my point of view, I don't think this is the right focus right now. Right. At least try to improve it to make more sense and be more useful. Because right now I use it once and after that I was like, okay, game over. It's not for me, it's not helpful at all. I'm gonna do it like the old way. And yeah, but I love AI. And this is why I'm telling, some of the companies are trying to do this just to jump on this trend and promote these gimmicks that are not kind of useful. So again, we need to take everything with a grain of salt. Right, right. That makes sense. Have you tried stable diffusion and or mid-journey? And is there one that you, which one do you like more or find more useful? This is a good one as well. I have a good friend of mine. If you don't know it, guys, you definitely need to follow. He's working on Snapchat and creating like he's a mid-journey wizard, David Barona. Shout out to my boy, David. And he's super talented and he's working with stable diffusion and mid-journey as well. And I've learned a lot of stuff about mid-journey and how to properly prompt on mid-journey to get the best results from him. And right now it depends. With stable diffusion, in some cases, yeah, you can get way more better results and have this continuity. For example, with the same faces, you can use masks and all that stuff and to replicate the same face and the same models and whatever, so on and so forth. But for some quick task, and this is why I wanna discuss like what I'm actually using in my day-to-day life as a designer, you know? Yeah. So for example, myself, I'm using more, more also Dali and mid-journey. It depends. When I don't like what Dali generates, I'm going to mid-journey back and forth because it's much more simple and convenient for me not to, I mean, yeah, stable diffusion is better but more technical. And you need to have more time to play and fool around with stable diffusion to get way better results. Yeah, that's the part, but you're wasting time. But yeah, if you do have some time, go and have some fun with stable diffusion as well. But for me, when I'm writing a blog post and I want some images super fast and slap them in there and upload everything and move on to my other billion tasks that I need to finish, you know? I'm trying to, yeah, not waste that much time and use either charge-EPT, Dali inside charge-EPT or mid-journey. Okay. At least for the moment. Maybe I'm gonna switch in the future, who knows? But right now, this is what I'm doing. Yeah. So Dali and the mid-journey for sure. Okay. Let's go to some audience questions. I've seen some being dropped below and then we can circle back on maybe another topic that you wanna touch a bit later but let's go to this question here. How can we design products that make it easy for people to work together with AI, ensuring that AI feels natural and enhances our everyday experience? You wanna give that a shot? Cause that's a tough question, right? Like what do you mean? I don't think I got the question. I'm sorry about that. I don't think I got the question. What do you mean by like to enhance, like to use it as a designer to use AI tools or to implement AI tools in the actual product? Yeah. So I think, yeah, to implement tools in actual product. Like a chatbot, like a custom AI chatbot or stuff like that. Yeah. So from the designer's perspective, right? Let's just say me and you are building something AI, right? How do we ensure that it feels natural and enhances the experience of the product, right? How do we make it easy for people to work with the AI that we're building, right? I can't be more specific. Sorry, can't be more specific. Like we're building what? Okay, let's just say we are building a portfolio tool. Let's just say like, we're gonna try and help designers crack the how to build a portfolio tool, right? So there's like, okay, well, how do we apply AI into this? And how does it, how is it natural for people to just use this tool without thinking about prompt design, prompt engineering and deep learning and all this other stuff, right? So they can just kind of pick it up and just use it, right? And this is a question from the audience. So I'm just kind of adding color. But yeah, what would you like, you know, any thoughts in your head? Like, have you thought about designing AI products and how do you make it like approachable for like everyday people? Let's just say like a 21 year old that's just like come out of university that doesn't learn much, right? They want to learn more about how to use AI tools to help them out in their like day to day. No, they actually, no, they just actually want to use a tool, but the tool, our tool has an AI feature, right? So we have to make sure that our tool is approachable and accessible to them without them doing the heavy lifting. I think this is what the person is saying. Does that make sense? Okay, let me try and answer this question then. If I understood it correctly. Yeah, sure, please go ahead. And I'll repeat the question just so we both on the same page. How can we design products that make it easy for people to work together with AI? Ensuring that using AI feels natural and enhances our everyday experience. So in my perspective, the way we speak to AI now is fairly natural. It's natural language. We are texting. That's kind of the most used interface ever, right? I think to make it even feel more natural and enhance everyday experience, you could design products that are embedded in the home and you can speak to it, right? That's the probably the most natural thing like me and you are doing. We don't have to type, right? Because inherently that isn't natural. Like we didn't do this before, like the computer and the phones. So it's like talking, hand gestures, looking at stuff, right? Like is probably the most natural way to do things, right? Or it could be if we're going super futuristic, a thought, right? I think neural link is probably gonna research something like this, which is like if you put that thing in your head by you thinking, you can tell the AR to do stuff, right? But if we actually take it to an example that's real, like every day, I think it's actually not about telling people necessarily you're pushing AI features and having people build prompts and stuff like that. Just really tailoring sort of the thing that they need basically kind of predict what they need before they need it, right? And having like maybe slightly less annoying notifications and be like, oh, did you wanna Google how to get home today, right? Or like, oh, did you know your three of these meetings you need to follow up on? Like that would be interesting way to use AI, right? Like imagine it scanning your inbox and saying you didn't reply to these three messages and they seem quite important. Yeah, yeah, something like that, you know? There's probably a lot more ideas but really designing AI that meets the user where they are versus having them prompting and learning about machine learning and stuff. I think that's probably like a very general answer. I'm not sure if I understood this person's question but that's what I got from it. Any take on what I just said and what would you add, Tibi? Yeah, the thing is I was thinking about this one and since we're talking about what's natural and what's not natural. And right now, and I might be biased on this one so take everything what I say with a grain of salt. For me at least, let's talk only from my perspective right now. For me at least, how do I feel it? Right now I feel like more natural somehow to write. Maybe this is because I'm not a native English speaker, I don't know, it might come into account this one as well. But for me, I feel more like natural to write than to speak, you know? Because I feel like I have more time in my brain to try to understand and develop the whole concepts and whatever. And yeah, we're talking about maybe this is more natural in the future to simply talk with tools like that, you know? But for me, for example, I'd feel more natural to write in the future. What about speaking in your native tongue? I mean, AI can do that now, right? You can just say whatever language you want, man. It's cool. Yeah, I don't know, I'm telling you. I feel like I feel more comfortable right now to write than to speak. I don't know why, maybe it's only me and maybe I don't know. If this is, we need to do some research. If this is the new natural, you know? Tell us in the comments. Yeah, I'm super curious, drop down in the comments. If it's more natural for you, are you feeling more comfortable, like to write or to speak? How do you feel like more comfortable? Super curious about this one. Like that. Okay, that's fair enough, yeah. And maybe in the best of both worlds, they'll be both, right? Like writing as much, writing is already there, in my opinion. The speaking element is not, right? We're like, Siri sucks, like. I mean, yeah, right now we're having ChargeGPT on phones and we can speak directly with ChargeGPT. And for example, like a fun fact, yesterday I bought like a whatever thing and I received the instructions only in Polish. And I was, bro, you're kidding me. Only in Polish, what is this? Bro, what is this? How I'm gonna build this thing, you know? And I was, yo, TV, you have ChargeGPT, bro, on your phone. Snap up, take a picture and put it to translate it into English. And boom, like that, easy. Right, right, right. So, yeah, it's amazing what a time to be alive. Yeah, exactly. I have a couple of, another question here was like, any thoughts about voice AI products replicating human voice? How do we know whether it's real or not? Any thoughts about that? I know, I mean, I'm aware. I don't know. I'm aware a lot of AI tools that are replicating like the human voice and stuff like that. And I saw a couple of weeks ago, some guy who was mimicking, I don't know, a famous rapper voice and trying to sound exactly like him with AI training that voice. And yeah, it's mind blowing. I don't want to say more about that because I didn't like dive into this, like super deep. But yeah, it's interesting what is going on. But I played around with some deep fakes and that shit, bro, it's super scary to be honest. I took a Tupac video, two of America's most wanted and slapped my face on Tupac's body. Bro, literally two or three minutes to do this. And it was like almost flawless. For two minutes, two minutes of work to do a deep fake that good. And to combine it with voice, that's some next level shit. And yeah, this is why I'm scared to be honest about this direction of AI. Because as internet, internet has its good parts and bad parts and AI will be the same, actually the same, the goods and the bad. So yeah, we need to be super careful how we're gonna use it and what to do with AI. But let me tell you something else that I completely forgot to tell when we're discussing about the prompting and whatever, how to write better prompts to get better outputs. Super fast. I was fooling around the other day with the, with chat GPT and I've stumbled upon this amazing thing. It's called markdown. You know what's markdown, right? To use both the italics and bulleted list and whatever. Yeah. So using formatting, like formatting elements like markdown can positively impact the chat GPT conversation. And chat GPT can pay closer attention to what you put in bold. So two stars, you add two stars, add your code or your word or whatever and two stars at the end. And you're gonna bold that one and chat GPT will pay closer attention to that part or you can write something in caps, all caps and as well. And this is super fascinating. I mean, you don't need to say, be, please take, have a closer attention to this one. Instead, you can use markdown super fast and boom, chat GPT will highlight and focus more on that part. And this is another way how to get better outputs. And about prompting techniques, I'm gonna get into a list like super fast type of prompts that will help you out. Like, this is what are the type of prompts that I'm using the most in my conversations with large language models, in our case, chat GPT, right? I'm using exploratory prompts, refinement prompts and we're gonna discuss briefly about each one super, super fast. Exploratory prompts, refinement prompts, contextual prompts, iterative feedback prompts, concept validation prompts, and information synthesis prompts. And these are the six prompting techniques that I'm using right now when I'm starting conversations. So the first one, exploratory prompts, you begin with a broad or open-ended questions to explore a range of ideas or gather general information about a topic. These are the, you explore, right? You're exploring with open-ended questions. The refinement prompts, you start with the general idea and after that you progressively narrow down the focus through a series of more specific questions. So you start broad and then get more specific and you refine your query, right? After that you have the contextual prompts and these prompts include the detailed context or background information. This is why they're called contextual. You include the context and helping to frame the question where a specific scenario or a set of constraints. After that you have iterative feedback prompts and you seek iterative feedback or critique. These are like the critique prompts adjusting the queries based on the responses you received to refine or improve the designs, for example. And the concept validation, you present the concept and seek validation or opinions or whatever suggestions for improvement. And the last one, information synthesis. These prompts aim to gather and synthesize. This is the word in English, right? Synthesize, I'm saying it correctly. Okay, cool. Synthesize information from multiple sources or viewpoints for a comprehensive understanding of the subject. And this is why it's super important in your custom instructions on chat GPT to write in there when giving me, when I'm receiving answers or the outputs, please talk also about the goods and the bad parts. Because most like chat GPT or any other tools we'll talk about only the good parts. And you need to have like the best of both words, the good and the bad. Right. Okay, that makes sense, that makes sense. But yeah, those were some instructions as well. By the way, as a hint. Right, right. So exploratory prompts, good for ideation, refinement prompts, start with an idea, refine it over time, contextual prompt, give background scenarios, iterative feedback prompts, that's I guess... Seek feedback or critique. Yeah, yeah, that's self-experimentary. Constant validation, ask for recommendations, opinions, suggestions, and then information synthesis prompt. Basically like affinity diagram this and give me the summary or whatever, right? Something like that. And about that, another tool that I wanna share with everyone, it's free, it's from Google, amazing. But if you're not from the States, you need to use a VPN. But you can find a free VPN and that's the end of story. And you don't need a phone number to register an American phone number or whatever. You just need a VPN and that's all. It's called notebooklm.google. And this notebooklm, it's a virtual research assistant and you can upload documents. And from that, you can receive information that most matter to you. And this one is super nice, notebooklm.google. Yeah, let me share that. I recommend it and it's free. Okay, what do you use this for specifically in your workflow? How do you use it? Yeah, that's a good one. So for example, I'm building my own website, right? Yep. And I'm trying to improve it every single day. That's the point. Yeah. And one tool that I'm using is St. George reviews. And I'm getting reviews from like actual people who purchased my products from my website. And I have over 100 and something reviews. And one thing that I'm doing, I'm exporting everything from Senja all my reviews. And I'm importing them into Google LLM, for example, but you can do the same in Bing, in chat GPT with data analysis, but let's take on, in our example, the Google LLM. And I'm asking, okay, so based on these reviews, what I'm doing good and what I'm doing bad and what I can improve based on the user feedback. And this is super important because I'm trying to understand what the actual users like on my products or they don't like or, and I can improve them. Yeah, get better. And actually, yeah, on the long run, improve my products. So this is another way, actual way you can use this kind of tools. Right. Yeah. That makes sense. So like the whole take of this, like ask me anything. Yeah. I want to tell people that you need to be curious. You don't need to be an expert. I mean, it's one year and something. The AI technology, it's super young, it's still evolving. It's just only the beginning. And this is the best time to be and jump straight in with your head and be curious, go and fool around, try all the tools, see what works for you, what don't. You don't need to be an expert. You don't even need to know all the terminology. Yeah, it's helpful. And it's okay to know the terminology, but go with your head right in, see what happens. And yeah, AI shouldn't sound, you know, shouldn't sound. That's my problem, bro. AI, most of the people make AI sound super complicated. What is this? What is why, why, bro, why? Tell me why. To be fair, I think we added to the confusion where we were like, okay, before we do AI, here are five definitions, but really you're right. You could just try it all. Come on. We didn't see what happens. You don't need to understand deep learning, generative AI, you can just do it. Be curious. Yeah, yeah. Be curious. That's the whole take of it. And yeah, it should be fun. It should be fun. Discovering new stuff. You know, come on. So like, you know what happened with open AI the other day? And I really want to hear your take on this, which is like, the CEO got like, you know, ousted outside the company, kicked out the company basically. The board basically fired him. What was your opinion around that situation? Cause obviously you're following this field really closely. Did you, were you shocked? Were you like, oh no, Microsoft hired him. Like, did you follow that closely or? No, I didn't. Like to be honest, to be honest, I didn't. I was trying to do some other stuff. I had a lot of my plate and I'm going to be honest. I don't like to talk what I don't know. You know, that's the thing. That's right. Yeah, it was just crazy. It was just crazy. Yeah. On the long run, I don't know how this, I mean, yeah, it will affect in a way or another, but I don't think it will be like a super groundbreaking thing. If Sam Altman, it's the CEO or not, or whatever, from my point of view, at least. I mean, he's only one person. Yeah, one super important person in this entire thing. But yeah, it's, I don't think it will change that much from my point of view. It might be wrong. I don't know. If you want to hear my take. Yeah, yeah, of course, exactly. That's why you're all right. I think without him, the things are moving forward anyway. So we should focus on what we can actually help us to in our day-to-day job, like learning how to prompt better, finding new tools, whatever, you know? Yeah, yeah, yeah. No, he is the CEO at the moment, but there was this whole debacle that he wasn't, and then Microsoft hired him, and then people were going to quit. So basically, OpenAI being half of a nonprofit turned into a for-profit. But anyway, everything solved. And there's also rumors that he saw something so revolutionary from the team at OpenAI, that he was trying to push it commercially, but the other folks were like, nah, bro, you can't release this. What exactly, like it was, do you know? No, no, there was just like some whispers, because basically this was the whole news for like a whole week. Like everywhere, every single podcast, every single article was like new information coming out, and it turns out to be nothing. But I can't wait, like even if, like, you know, there was talk about, what is it, singularity, like that's kind of like the level of like, are you sure that this is not sentient? Like is this AI, you know, is it AGI? But anyway, that was just kind of the industry. Yeah, about this kind of rumors, I was reading, again, a couple of days ago, something like Chagipiti is getting lazier in the wintertime. And he's acting, you know, like an actual human being somehow and he's getting lazier in the winter. Yeah, it's also fun. Okay, so let's end on some like future, do you have any predictions for AI or like what would you like to see? Not predictions, like let's not be like experts. Yeah, like what would you like to see in the next five to 10 years that AI can help us or help you or like some interesting stuff, like anything you like? Bro, to be honest, I don't know what to say exactly. I'm trying to take every single day as it is and to try to focus on what I can do the best today. And how I can get the most out of this technology today and focus only on present because I guess this is a personal problem that I'm fighting with it lately. I was locked in future for such a long time. If I'm gonna do this x amount of money, if I'm gonna be whatever in this place in the next whatever years, if I'm gonna whatever. And I don't want to think in the future. I wanna stay locked in present, not in the past, not in the future, only in present. And to do whatever I can do the best today, you know? And this is why I don't think I want to do like predictions anymore. Let's try to focus on what we can find today and get the best out of it and hope for the best. If because- Okay, okay, let me ask you a different thing. If I gave you unlimited money to build an AI tool, what AI tool would that be? And it has to be something AI. Like that maybe is a bit of framing for, what would you like to see in the world, right? Cause you can do it- I don't know, I would build, like this is a thought that I had, but I've like right now, I was jumping around from here and there. I'm thinking that would be a good idea to have like a dedicated AI platform to teach other people about the right way to curate the contributors in the platform with courses but with the top notch, the best AI courses for everyone. And you will find only, only, but only AI courses in there. And courses related to design to, because right now, yeah, you have one the interaction design foundation. You have it on the Coursera, you have it on whatever, but they are scattered in multiple places. As far as I know, as far as I know, I might be wrong. I don't know if you guys in the comment section, you can send us, if you know, like a dedicated place, a platform only, only with good AI tools, AI courses, sorry. Because I also know there are a lot, unfortunately, because yeah, a lot of people are trying to like ride the wave. Yeah. And I are doing this just for the money and are trying to push a lot of like, from my point of view, trash AI courses. And this is why I didn't want to go with and create at least not for the moment, an AI course. And I wrote only this book because I feel like I'm not kind of ready to create like an amazing AI course right now. And it's kind of hard to be honest, right now with the informations and all that stuff. And this is why I'm waiting a little bit for the technology to evolve and all that stuff. And I'm gonna update the book, but I'm not prepared to create a course, at least not yet. Okay, okay. Well, I'm sure folks on the call is waiting for that. And you know, maybe taking your own advice TV, just get started. You know, you're learning as well. The course will get better over time. You'll use AI to refine it. So... And you need to have fun. And yeah, you can hear from a lot of people, bro, you're not doing this correctly. You're using it wrong. You're, bro, leave me alone. Let me do more stuff, you know? Let me try to be happy and explore and see whatever. And if I don't know stuff, I'm gonna go and research because that's the part of the game. I'm getting smarter and all that stuff I'm evolving. And yeah, I guess at the end of the day, this is how we get better and how we get like the experts in maybe five years or 10 years or name it. You know? Right, right. You just need to be curious and genuine. And if you don't know something, assume that it's totally fine. You don't need to know everything in the world, but you need to keep this curious mind, to have this curious mind and try to explore and see what happens. Learn and write everything down. This is super important. And I just wanna say it. You need to write everything down. You don't need to remember anything in your brain. Your brain is only to have ideas. And not to store them. Use a second brain, a notion second brain, please. Yeah, and if you like speaking, use voice notes. There's AI voice notes rules these days. So yeah, okay. So let's wrap it up here. Anything that you wanna plug to the folks listening? Anything at all? Yes, I wanna plug. So for a while, I did this announcement to my LinkedIn a couple of days ago. I'm gonna take a break. I feel anxious and overwhelmed a bit. It's not my best period, but it is not all about that. I have some resources, like free resources in the first place. I don't wanna try to sell anything or that stuff. I do have some paid products as well, but I wanna focus more on my free resources. You can check out on my LinkedIn or on my website, tbdavid, t-i-b-i, david.com. You have a lot of free resources that can help you out to kickstart this, your AI journey. And yeah, other than that, have fun and explore this amazing field that will change the world for good, yeah. Amazing, man. Okay, well, thank you for coming on and you guys can follow Tim on LinkedIn. I just left his website in the comments, so go check that out. Go get yourself some free resources to download and learn about AI and I'll probably be doing another LinkedIn Live very soon, potentially probably next year. So let's just say happy holidays to everybody and you can check out my stuff where you know. But thank you again, Timmy, for having you on. Bro, you guys are amazing. I love you. Heart, heart, heart, yeah. And thank you for having me, Chris. Oh good, it was fun. All right, peace guys.