Most Intense Week For Practical AI Use Cases

All right, so I have another batch of AI releases that you can put to work today for you here. Matter of fact, we have 15 different updates, applications or brand new creations from all across the Generative AI space. This week has been a treasure trove of different apps that you can actually use. So let's get right into this because first things first, we got something that most people in the Genai space have been patiently waiting for, and that is a Sora competitor that is worthy and that you can actually use. So just a little bit of context, assuming you know what Sora is, we talked about it on the channel. It's OpenAI's video generator that looks incredible, but we never got access to it. Last week we got news out of China that they have a brand new model that you also couldn't use. You needed a Chinese phone number and even then it was very limited. But now we have something we can actually use. It's by Lunger Labs and they call it Dream Machine. And I think everybody in the space agrees it's the best AI video generator that is actually accessible today. So how does it work? Well, it's quite simple. If you log in on their website, you get faced with this interface and just right away, I typed in a cat with a hat surfing. I used a little check mark that improves your prompt and I got this. Okay, that's a good start. But people across the Internet have been using this for a variety of things. For example, here you can take a picture and animate it like so. Look at that. Some of this stuff is incredible. And you can create stuff like this yourself. I just logged in with a Google account. I didn't pay for anything at the time of this recording. You can just try it out like so. But be warned, so many people are using it that if you're using it in peak times, which is us waking times, this thing is slow on the day of release. People had to wait for hours to get one generation back. Okay, here's the little drone shot we got. First part, not so impressive, but the second half, I think this is completely usable. Look at this. That's a little clip you could include in a montage and I think most people wouldn't notice. Sure. If you look closely, you have the same water problems as with any AI video generator. I'm gonna run one more. But seriously, go try it out. If you see this, this is so fun. As the name suggests, you can turn anything that you dreamt of into little video clips. I wonder what it'll do with this AI. Generated Yoda relative that we generated on one of the YouTube streams. I'll just say Yoda. Here's the Yoda animation. This is actually a really good look at characters that actually excels. So yeah, uploading images and animating them seems to be one of the big strengths of this model. But this is above and beyond all other video generators we have so far, like Runway, Pika, Pixverse and others. Try it out for free as long as you can. Okay, so next up we have a Descript update and we barely missed this last week as our cutoff is usually on Wednesday. But descript came out with a major overhaul and they added in a bunch of features. This is similar to the canva update that we talked about recently. Canva also existing Graphic Design Studio. It's sort of a Photoshop for content creators. Well, this is the video editing software for creators who don't want to bother with the more advanced, intricate, convoluted programs like Premiere Pro or DaVinci Resolve. But here's the thing, they added so many new functionalities and AI integrations that actually pulled ahead of the big players like Premiere, DaVinci Resolve or Final Cut Pro when it comes to a few of these features. And all of these AI integrations come in form of a new assistant that they call the Underlord. And I appreciate the communication and the marketing on this because they're not claiming that this is some extremely capable AI that is just going to change everything. It's the Underlord. It's the opposite of Overlord. I kind of enjoy that. And it's accurate. A lot of edits that AI is going to propose are not going to be right. You need to double check what it does. That's just how most of these tools work these days. So let's have a look at some of the concrete functionalities. So I just uploaded a recording of a live stream and as you might know, descript is this text based editor. You have this feature in other software like Premiere Pro already. We actually use it regularly. So if I wanted to edit out this space, I just select it and remove it. This is kind of the very basics of descript. We're not going to go into everything here, by the way. I just want to highlight a few features. But all of These interesting new AI integrations are hidden under this Underlord 3 tab here on the right. So as you can see, there's a lot of options. These are the editing ones, These are the ones that affect the look of the video. And then here we can repurpose it for other platforms or create things like summaries or timestamps. So let's have a look. This is the beginning of one of my YouTube streams and I'm just looking at, chat and doing various things, sometimes looking at the camera, but mostly checking chat. So what I'll do here, I'll delete most of the live stream and I'll just keep the first part. And now we can try something like this new eye contact feature. And important to point out, this is not something we haven't seen before. It's just the fact that they integrate it into the tool that you might actually be using that makes so much sense. Hello, everybody. Welcome to another edition of the Other Livestream. There's been a waiting screen, so we don't need to waste any time. We can just dive right in and I spilled some water. That's okay. That's okay. So as you can see, this eye correction feature works best if you're looking slightly besides the camera, like so, then it can redirect your eyes back right into the lens. But if you have clips like this where you might be looking at this screen, this screen into the camera, you're doing stuff rather than consciously presenting straight to the camera. And in those cases, this eye correction feature won't do much. But if you have a long podcast where you might be looking at your screen for an hour, like so, it can fix your eyes and adjust them so it looks like you're looking into the camera. So this is just one of the features. Let's look at some others here. One of them is clipping things if you want to repurpose them for other social platforms. Also integrated here, you can translate to various languages. You can improve sound, remove filler words, create transcriptions, captions, green screen stuff. All functionalities that have been available all across the Internet, but you had to piece them together. Now they're in here and they're all integrated with this technology text based editor. But again, keep in mind that tools like this generally are something like the underlord, where they're not going to be perfect. You need to double check everything. It's an assistant to you. Let's have a look at some of the other features here. For example, if I want to edit this text that I have here, this first category allows you to do a few operations that you might want to be doing regularly. So here's one of these features. You can just say edit for clarity and it's going to make suggestions on which words should be removed and which words should stay If I apply these edits to the script, you'll see that these are edited out. Now shortening my 19 second video to a 13 second video. And look, it edited out the fact that I spilled some water and that that's okay. It took out the intro in the beginning. Hello everybody. Welcome to another edition of the live stream. There's been a waiting screen, so but as you can see, this isn't perfect. For example, this one I definitely would want to revert. So again it's just an assistant helping you. A very welcome feature set though. So yeah, for video editing, some very welcome updates to a tool that already has been great for people who don't want to spend time learning something like Premiere Pro or DaVinci Resolve. All right, next up we have some Serious releases from StabilityEye, the company behind Stable Diffusion and so many other models. And what they did is they essentially open source stable diffusion 3. Now it is sort of a limited license for very, very large companies, but if you're a small creator or a medium sized company, you can absolutely use this with a non commercial license. So Stable Diffusion Free is nothing new. We covered it before, so now they published a way. So if you want to run this locally or if you want to build the app on top of this, now you can, you have the full license and you have everything you need to do that. And this is significant because the quality of this is great. But it also has the ability to do consistent text as you can see in these examples here. So that's just a quick update of that. It's one of these releases that if you know what to do with it and know what to build with it, then this is very exciting. And for everybody else we have a bunch of other releases coming here. The next one being actually also from Stability AI, it's called Stable Audio Open. And this is very different from Stable Audio that has been released before. Stable Audio is sort of a weaker Suno and Yurio. I think that is a fair assessment. But now Stable Audio Open is not trying to compete with these music generators. Stable Audio Open is all about creating sound effects, foley sounds, background sounds or isolated instruments. So this is more of a creator tool and that's why I think it's very interesting they fully open source this thing and you can use it today via this hugging face space. So let's try the same prompt from last week that we ran for ElevenLabs new sound effects generator where it didn't work so well. Okay. After two tries and about a minute of waiting, we got this. Okay, so let me tell you, as somebody who has experience in video editing, you can totally use this because you can cut out the middle parts and make the duration fit your needs. Just start here. So is this the perfect engine sound for what you might be working on? I don't know, maybe not. But this crash is amazing and you even get extra sound effects here in the back. This is great. Totally usable. Free open source. If you're a creator of any audiovisual content, I think you should bookmark this one. So, as you might know, this weekly show is all about telling you what AI tools you can put to work today. But you're probably not going to learn everything you need to achieve your goals by watching one YouTube video a week. And you might already know, some skills just take a little more time and effort to acquire. And that's why I really like Brilliant. Org, the sponsor of today's video. Brilliant is an interactive learning platform that offers thousands of lessons in math, data, science, programing, of course, AI. What makes Brilliant different is their hands on approach to learning. Instead of boring lectures, you get interactive lessons that build your problem solving skills while you learn essential concepts. So in last week's video, I told you about the applied Python course and how useful it can be for people, people trying to make things happen with AI. Well, the next level to that same learning path is a course called Introduction to Neural Networks. It teaches you about the inner workings of neural networks, helping you understand how applications like ChatGPT actually work under the hood. In 15 lessons, this course breaks down advanced machine learning concepts into digestible interactive activities that make learning it manageable. And I'm telling you this to highlight that Brilliant's courses aren't one off lessons. Most of the courses link together in learning paths like this one, because that is the best way to learn. These skills really do stack, as I say, oh so often. To try Brilliant for free For a full 30 days, head on over to brilliant.org DaiAdvantage or click the link in the description below. Plus, if you decide to stick with it, you'll get 20% off an annual subscription. A big thank you to Brilliant for sponsoring this video. And now back to more AI news you can use. Okay, next up. And this one is going to be quick. It's just Asana implementing a new feature with AI teammates. Asana is a work or task management tool and they're implementing AI in a way that I think you should know of. And I'm pointing this out because I think it's interesting how they're Integrating it here I see more and more companies go with this route. They're not just saying, hey, AI integrations, they're saying, no, you're going to have some AI assistants, or in their case they're calling it teammates that you're going to be including in your workflow. And particularly I wanted to point out these use cases that they show off here in their announcement post because you can expect features and use cases like this across many other apps that help you manage something. Look, so for example, if you're in creative production, concretely an outdoor advertising company, you can use AI teammates to improve and fast track the creative request process so that other employees can spend more time focusing on creative flow and design. So what concretely could these teammates do here for you? Well, they can handle incoming requests and proactively gather missing information. They can assign work to specific people based on the context that you provided. And by the way, you get to provide it with this context. As you can see here, it automatically gathers it, but then you can also add or take away various projects or details that it might need to make the right decisions. Can help with client research and it can help with reporting if you have some data. Another use case would be this marketing campaign. It can create content or basically first drafts of content. You know, it's all the same with gen AI tools. It's an assistant, it's not a worker for you. Just like we saw with Descript a second ago here. So basically all of these AI capabilities are just now being integrated into the apps you already might be using. That's the common theme here. So just expect more of this to happen. I think the best takeaway is if you understand some of these primitives or whenever we show off a new hugging face base like the Stable Audio generator, it's just a question of time until some of that tech is integrated into the apps that you already using. Matter of fact, Apple's presentation this week might be the perfect example of that. I created a standalone video covering all the AI announcements that he made there. None of that is usable today. It's all going to come in fall. But most of that is them taking various features and various proven use cases that we cover on this show every single week and integrating it into their laptops, into their phones and iPads. If you want more details on what exactly they integrated, you can check out the standalone video. I think they're the best example right now of them integrating the most meaningful AI innovations by back into their products. But as you can see, many others like Descript Asana or even Photoshop are doing it on a regular basis now. Expect this trend to continue. And what we're going to be looking at next is actually some stuff that you guys created. I challenged you in a live stream to submit various GPTs that you have built in the AI Advantage public challenge for June, and this week I spent two and a half hours in a live stream on YouTube evaluating all those results, all the ones that had some learnings. Matter of fact, there was a total of 15 submissions that we analyzed and went into. So here I'm just going to show you the three winners and the one bonus winner of the contest. If you want more details and see me discussing these things more in depth, you can check out the livestream. They're always recorded and stored on the channel, but Basically we have four different GPTs that do unique things. Let's start with the winner here, which was Brendan with Wadslau the Vacation Planner. Very creative GPT. Again, I won't go into depth here. I spent about 10 minutes talking about this on the stream, but what Vaclav does is exactly as the name suggests. It helps you plan your vacation. And this implementation has been really great because after asking you some questions, it uses the code interpreter to create a heat map for you of where you should go on vacation, depending on various criteria that it comes up with as you answer your questions. Essentially, it's a very creative way of deciding where to go on vacation, which might just be very useful to you with summer essentially starting now. Another winner was Christian. With his Universal Translator GPT. He basically rebuilt what OpenAI promised with the voice assistant and the GPT4.0 announcement inside of today's ChatGPT. So this GPT is essentially set up for you to use your phone today with today's capabilities, without the new voice assistant, and for ChatGPT to act as a universal translator for whatever language you set it up to translate to and from. Some very smart implementations here like interruption handling and automatic language detection. So yeah, with this you can use your phone as a universal translator. Then we have Gerhard, who created a GPT that helps him craft stories for his children at night, and there's this interesting implementation where he actually limits it to not talk about specific topics. Very useful if you're creating prompts or GPTs for children. You can try it out over here. All links are in the description as per usual, and then in the bonus categories where there was no restrictions, just built an interesting GPT Jacob McClelland won with his debugging GPT So if you're building a GPT and run into some issue, you can add mention his GPT which is fully optimized and he's iterating on it on a daily basis, improving it with the community's feedback that now he has access to. Again, this GPT is freely available. You can just pin it to your sidebar and ention it whenever something goes wrong in a conversation with another GPT. Four fantastic use cases and now the GPTs are accessible to everybody, independent of you having a plus plan or not. You can go and try out all of these for free today. If you're enjoying all of this research and testing that we compiled into one video here for you, hit the like button on the video. It really helps out the channel. But now let's get back to the next piece of AI news you can use. All right, next up, we have yet another Suno update as they're shipping new stuff on a weekly basis now. And I really like this one because it allows you to upload your own audio files and it creates a song based on the audio you upload. So you already know what we're going to be doing here. I'm going to head on over to my Suno account, go to upload audio, say audio, and I'm going to give it access to my microphone this time. And now I'm gonna record some random sounds and see what kind of song we can create from that. All right, let's try something really simple here that works. Upload the file. It needs to be at least 60 seconds long. Up to 60. Let's see what we get from Suno here. From a very inspired creation. All right, we have a cover image. We can continue right here. I agree to the terms, of course, of course. Okay, then what you can do, you can click this extend button on this audio clip and I can turn this into a hip hop battle rap beat without lyrics and just say extend and it's going to go on from the 9th 2nd and continue. And you're going to be using SunOS AI generator to extend your clip. So after spending about 15 minutes with this, I can't seem to replicate what they shared on their socials. I follow these exact steps. You know, I navigate to library and then I have a clip and once uploaded I choose extend. I have a timestamp, I provided a genre and I did not include any lyrics. And the generations I get are battle raps, but they don't seem to have much to do with my original sound. Maybe someone in the comments can Enlighten me as to what I should be doing differently here. Nevertheless, I believe this is a useful capability that both Sono and Yurio have now. All right, so next up, we have some midjourney updates. Namely, there's a new personalization parameter that you can use. And basically, this is a custom style that midjourney creates for you based on all the previous images that you rated. Rated, where you might ask. Well, if you didn't know, there is a tab here on the midjourney website where you can rank different images and you can basically say which one of these two you prefer, like so or like so. I think you get the point. And then based on these votes, if you have at least 200 of them, it will create a personalized style for you. Now, a lot of people have thousands of ratings. Matter of fact, Dominique from our community that runs the midjourney events, has ranked over 9,000 images over the past few months. It's over 9,000. 9,000. There's no way that can be right. And here he's showing us the differences between the original prompt. If you just say beautiful woman in a 16.9 aspect ratio, and you use midjourney v6, you get this. So these are the results between the personalization parameter and these are the results after. Now, you can also see them side by side. Now, look, they're obviously different, but this personalized style is going to be different from user to user. So what you can do is you can actually share a little code at the end of it. Here's Dominique's personalized style. But as I did over 9,000 images before, this personalized style was a thing, it's not as focused. You probably want to do this with a new account while having this personalized style in mind. So if you have some other artists where you want to reuse their style, you can absolutely do that. Some more before afters from Dominique here. Here we have a cat wearing a fancy suit before and then after the style. And this is what a hates the AI artist on our team has been doing. And here you can see some of his generations in his personalized style. He has a completely different preference. Some of these results are wildly good. If you want to use this hate style in your own generations, you can use this quote that is on screen right now. And if you enjoyed this, you should definitely follow his Instagram for more extraordinary AI art creations like this. All right, then, the second and super quick midjourney update. It's Niji 6, and it's now better at representing Chinese and Japanese characters. Dominic went ahead and tested this. There's actually no choice between the previous and this model. It's just Niji6 and he reran some Niji6 prompts that he had stored away with the results with the new model, as you can see, there's barely any differences. And here's some brand new generations from Hades focusing on text. So this one is hard to judge without knowledge of the language. But what I can tell you is that Niji6 still is incredible. Look at this first person image with Niiji6. I've said this before, I'll say it again, it's a super underrated feature and style that hides within my journey. Very, very aesthetic in my opinion. Okay, and the last piece of news that you're gonna actually use for this week is gonna be Google's Illuminate. And they always experiment a lot with AI tools. I love that about them. They have this Google Apps thing that we talked about before. But here with Google Illuminate, it's all about condensing academic papers for you. And if you tried this before with something like ChatGPT, it's not that simple. It often loses key pieces of information, which is obviously essential when you're reading papers. So in Here you have 25 summaries of various papers related to the AI space. Attention is all you need is the original paper that introduced the transformer that makes all of these apps that we're looking at possible. And here you can simply click play and listen to a summary in four and a half minutes. Let's unpack. A paper titled attention is all you need. What's the core idea here? Well, the big idea in this paper is that we can build a really effective sequence transduction model without using recurrence or convolutions. And as you can see here, the key aspect is that it turns it into a conversation, not just some AI voice. Reading it off this makes it way more engaging and way more interesting to listen to. There's a wait list for you to try this on your own papers, but there's 25 of these that you can be listening to today. And I gotta say, having two voices read the summary is very refreshing and quite engaging as opposed to just one voice having you read the whole thing. This is especially valuable on heavier subject matters like summaries of papers. Plus the prompting that goes in the background is obviously very effective because it doesn't seem to drop the essential pieces of information in some of these summaries. You can simply join the wait list like so and listen to all the papers in here in a conversational format in under 5 minutes. Quite interesting tool. And that wraps it up for this week's AI News. You can use leave a comment below on which one is your favorite and which one you'll be diving deeper into and subscribe as I do this once every single week. All right, that's it. And I hope you have a great day.