AI Use Cases I Will Be Using Every Day

Every single week we have more and more AI tools coming out. So today we'll look at everything that came out that you can actually put to work. And I'll be giving you a live demo of pretty much all of these. Because Adobe released a new image generation model that is now in the new Photoshop. We have a new model from Microsoft that fits on pretty much any machine. Heck, this fits on most phones. And I found this little app that can de Clickbait all of YouTube for you and there's so much more. So as you can see, there's no time to lose. Let's dive right into this week's AI news you can use. First up, and most significantly, Adobe released a brand new image generation model. And this by itself happens regularly. Different companies release new models, but Adobe has the apps to add them to. Meaning we didn't just get Firefly free, we got a brand new version of the Photoshop beta where you can be using this. As you can see, it's even better at generating all kinds of images. Especially the hyper realistic ones are really good now. And even in the web interface, if you don't have Photoshop, they have this generative expand feature which is present in other apps, but it's one of the best features of these. You can reference structures. We looked at that already and it works together with Adobe Express or it can generate vectors on all this good stuff that is available across the Adobe cloud is now powered by this Firefly image free. So how does this work in practice? Well, first of all, you need to log in, but you can log in for free and get 25 credits for free. If you pay $5 a month, you'll get 100 credits for free. And if you have photoshop, you get 500 credits. And I personally have the Creative Cloud suite because I use Premiere and other apps from them. That plan gets a thousand generative credits. Okay, but is it any good? In short, yes. And what we're going to quickly do is I'm going to compare to the different versions of Midjourney. I know this is not the perfect test, but hey, this is just a first look. So in this resource we have 100 prompts that we ran for all six versions of Midjourney. By the way, this resource is something we created for the community, but I'll happily share it with you in this video. So again, I know this is not a perfect comparison, but let's run this through Firefly 3 as an effect, I'll only pick the fantasy one and maybe let's add the hyper realism back in to match this a little more as these are kind of hyper realistic. And there you go. That's pretty good. I personally would prefer Midjourney V6. If I were to subjectively rank this, I would say this is maybe better than Dall E3, but worse than Midjourney. It's a really strong model that you will absolutely want to consider. Don't forget that these Firefly models by Adobe are the safest in terms of copyright because they only trade up on data that they 100% own. And now let's look at the Photoshop features, because that is the really exciting part here. So as you can see, they added in Firefly free, but we can now also generate entire images, not just parts of images. We can use the reference image feature where we give it a backpack and then we map the structure of that backpack onto another part of the image. And then you can regenerate backgrounds and enhance details with the upscaler and a few more things. So let's just try our little demo here. I agree to the terms and I'll have to go with the obligatory. A cat with a hat and wings. Yes. We don't want a butterfly, we want something a little more exotic. I don't know about this. Okay, wait, let's try something proper. Let's do a neon colored butterfly. Okay, that's decent. And after reselecting it one more time, this is a really good result. I can already see the improved realism and you'll get to use all the other tools in here. So look, this is not a deep dive into the new Photoshop features. If you want that, we can absolutely do it. But from a first look, it is more realistic and the results are actually even more consistent than they used before. It could do little details, but something big like a butterfly that takes up this much space never really properly worked. You had to regenerate a lot and retouch it a bit. And I have to say Photoshop plus AI is one of the killer combinations that I just keep using regularly. It's that ChatGPT and a few others that I actually use on a daily basis. I'm so glad they upgraded this because for anybody who actually uses Photoshop, ever since they added generative fill, this is one of the features you would never want to give up again. It's just too good. And now it's even better. And as we're on the topic of AI image generators, we can quickly cover this. It's like Chatbot arena, but for open source image generators. Surprisingly Fun. You can just go in there and rate different images. Which one do you like more? And then it shows you which model did this. This is stable cascade and it generates another one with a random prompt. I like this Dragon more latent consistency model Stable diffusion excel Realistic vision. This way you can learn about the open source models really quickly. It's quite fun and it shows you the ranking here on the right. Who ranks highest? Super interesting easy use case. And talking about super interesting easy use cases, here's another quick one and I really enjoyed this. Now look, this one is independently developed by AJ here. Nevertheless I really want to show you because what this does is simply Describe it reskins YouTube in a less clickbaity manner and I think a lot of people would really appreciate this and I fully understand that need because YouTube just turned into this game of bombastic titles trying to one up others. It's just hyper competitive. So this is a fantastic solution. You can try it out for free and then you can also keep it for free. But he asked for a $1 payment which I would recommend you do. It's a great little tool. This is what it looks like on practice. So this is what happens when I disable the arrow and here enable it. I set it up so it picks random frames from the video instead of the thumbnail that the creator picked again before after. I don't know if it's necessarily better. It's certainly more unified and what it does to the title is that it essentially removes the capitalization. You can set it up to do different things here. I at first thought this used AI, but turns out it actually doesn't. Nevertheless, I did want to future this because a lot of you consume your AI content on YouTube and this is a way to remove some clutter. Great little app. I'll be turning this on and off over the next week and see if I like the results. Some of these AI tools we explore here sometimes seem like magic. A lot of times I see something new and it reminds me of this quote from Arthur C. Clarke. Any sufficiently advanced technology is indistinguishable from magic. But as with any magic trick, with enough knowledge about it, it turns into a skill. And that's where today's sponsor comes in. Brilliant.org because with Brilliant you can pick up new skills in no time. They have thousands of interactive lessons in math, data analysis, programming and AI. Each lesson is filled with hands on activities that teach you how to solve problems while you learn new concepts. And I think that's really important. You aren't Just memorizing and repeating information, you're becoming a better thinker while building your knowledge base. Plus, you can complete these lessons at any time that suits your schedule. Personally, I really like the scientific thinking course. It contains hands on lessons that teach you how scientific principles get applied to our physical world. And it doesn't require any math at all. So it's perfect for all levels of learners. To try out this course and everything else that Brilliant has to offer for a full 30 days for free, visit brilliant.org diadvantage or simply click the first link in the description. Through that, you can also get 20% off an annual subscription. Thanks to Brilliant for sponsoring this video. And now let's move on to the next piece of AI news you can use. All right, next up, we have a new open source model that actually pretty much everybody can run on their machine because it's so tiny. This is 5 free mini and it's a 3.8 billion parameter model that performs surprisingly well. The fact that it's this small means that it can run on most phones and therefore also on many laptops. I created another video this week. It's only three minutes long that shows you how to use Llama 8B or even an uncensored version of Llama. So if you want a step by step tutorial on how to activate that on your computer, watch that video. But pretty much I can go in here, change to the 5 free model, start a new chat, and this thing can perform some basic LLM tasks for you locally. No Internet required. Thank you, Microsoft. I guess. Okay, next up is a really fun one. It's just this random hugging face space where you can generate song ideas by humming them. I haven't tested this before. I wanted to do it live, so here goes nothing. You guys always seem to enjoy if I embarrass myself on here a little bit. You're God damn right. So let's do it. Okay. Always make sure to trim this because they explicitly state that you know, this is AI. Garbage in, garbage out. So just do that, keep all the defaults and then I guess it should generate a song from this. So as I catch this highlighter, surely the generation should be completed. Oh, it's done. Magic. Yeah, I. I don't know about this one. Maybe you'll get something better from it. Have fun playing. Next up, something that I consider a little more useful. This is a very simple use case, but basically you just write a prompt. It can be an emotion, it can be a small specific look, it could be pretty much Anything. And it's going to generate a color palette for you. For anybody who's not experienced or talented in graphic design, this might just become very useful. You're starting a new brand. You need a color palette. Let's just say excited, optimistic, and bright and see what this comes up with. As of now, this is completely free to use, and it should just give me color palettes based on my prompt. There you go. And beyond the color palette, it gives me mockups of different UIs, images, how my product could look. I think this is just one of these wonderful apps. It's so simple. And if you're designing something, what a great starting point. And guess what? If you don't like it, just regenerate. DeBlank. Good stuff. And talking about good stuff, fast food, what kind of transition is that? But Wendy over here implemented AI into their checkout system, and I just briefly wanted to share this with you. I figured that AI news you can use can spend beyond what just we can use. Wendy's using this to their advantage, and I figured you might just gain some inspiration from this. Look, they have this checkout counter where AI is transcribing the user's voice and then also and answering with a synthetic voice. Have a quick look. Welcome to Wendy's. What would you like? Can I have a chocolate frosty? Which size for the chocolate Frosty? Medium. Interesting. And this little Wendy use case might just indicate what will be happening in the near future here. Every customer support agent, fast food, restaurant worker that takes orders, heck, maybe even bank clerk at a certain point. I mean, they're pretty restricted in what they can do, right? There's always a thousand rules, and there's only a limited set of services they can provide. These technologies seem to be getting good enough quite fast. So a lot of these touch points with companies might be replaced by AI. Now, maybe banking is not the best example, but as you can see here with fast food, there's a lot of these use cases where, for a corporation, this automated solution might be just economically more efficient. But hey, as I always say, I think it's better to face these things straight up with open eyes and full awareness, rather than shutting yourself down to it and pretending like it's not happening. And arguably the best way to do that is to keep yourself informed and then take action whenever appropriate. And I hope that the show can contribute to that. All of these things you could be implementing into your workflows today, and I'd be super curious to hear which one of these excites you. The most. Which ones will you be using? And leave a comment below if you want me to dive deeper into any one of these tools. And if you're still here, you might want to check out this stream that we will be doing soon with some fellow AI servers. I'm super excited about this. We've been in touch in a private community since a while and now we're going to co stream and just discuss all the topics that were released throughout the last week. Should be a really fun time. So go and check out that livestream on April 29th here on YouTube. All the links are below and I hope you have a great weekend. Like and subscribe.