Is This the Biggest Week for AI in a Long Time?

It's actually been a really busy week in the tech and AI world and I got a lot to cover in this video. So let's go ahead and break it down. Starting with the fact that this week Apple held their annual Wanderlust event where they basically announced the latest updates to the iPhone, the apple watch, the AirPods, and the fact that now everybody's going to have to use even a new cord to charge their iPhones. Most of the updates that Apple made were fairly marginal improvements. One of the coolest things that I thought about the event was when they talked about how the iPhone Pro was going to be able to actually film in 3D.

And there's one more unique feature we've created just for the world of spatial computing. With iPhone 15 Pro, you can now capture spatial videos. We use the ultra wide and main cameras together to create a three dimensional video. You can then relive these memories in a magical way on Apple Vision Pro. This is really important as your iPhone is with you all the time, so you won't ever miss capturing a special moment using spatial video. Like your family speech vacation, spatial video lets you feel like you're right back to that moment in time.

There's also some other cool features with the camera where can actually adjust the depth of field of your photos in post production. So if the foreground is in focus and the background's out of focus, you can actually go back in and edit the photo later and make the foreground out of focus and the background in focus. But overall it doesn't feel like a huge Upgrade from the iPhone 14. This week Adobe announced a ton of things as well. For instance, Adobe Premiere now has their speech enhancer functionality built in. So if you have sort of noisy audio, it will actually clean up the audio for you. And I'm a track coach. Hi, I'm Keenan Matthews and I'm a track coach. They're rolling out the text based editing inside of Premiere as well and it will automatically identify ums and us in the dialogue and help you quickly remove them. You can also do things like edit your video by directly deleting words from your transcript. And that little chunk that you just deleted from the transcript will also delete in your video. Something that we've had for a while in Descript, but now it's in products that might be more native to you, like Adobe Premiere. Adobe After Effects lets you edit 3D models directly within After Effects now. And they also added an AI powered roto brush so you can get that look of like a green screen and cut out everything behind you with AI directly from within After Effects Now. And these are available right now. In fact, if you have Adobe's Creative Cloud, you can access them by logging into your Creative Cloud desktop, clicking on Apps, and then down on the left side there's beta apps here. If you click into Beta apps, you can install Premiere Pro Beta and After Effects Beta and get those features right now. And speaking of beta, Adobe announced that their Firefly suite of products are now out of beta. So you no longer need to use the Adobe Photoshop Beta version to get Firefly and Generative Fill and Generative Expand and those really cool features inside of your Photoshop. Before you had to use that beta version of Photoshop like I just showed you in Adobe Creative Cloud. Now you just get it in the regular version. They are, however, increasing the plans on their cloud versions of the Adobe products. If you have one of the individual products, they're increasing by either $2 a month, $3 a month or $24 a year. And if you have the plans like I do, that gives you all of the Creative Cloud suite. They're upping the plan by $5, $7.50 or $60 annually. And still on Adobe News this week, Adobe announced that if you allow your images to be used as part of the training in Adobe stock, you'll actually get a bonus. They have a Firefly Contributor bonus. The stock contributor community of photographers, designers, illustrators, filmmakers and multimedia artists of all kind are vit process and to Adobe for both stock licensing and for Firefly training. To demonstrate that while the terms we already have with Adobe stock contributors permit Adobe to use stock content to train models, we're pleased to share two specific announcements focused on creating value for the community of stock contributors. A Firefly Contributor bonus. Adobe Stock contributors whose content was used to train the first commercial Firefly model will receive a Firefly bonus. It'll be based on the all time total number of approved images submitted to Adobe Stock the that were used for Firefly training, they'll be paying this bonus out once a year and it looks like it'll be starting in 2024. And now that you can use Adobe Firefly commercially, which is another one of the announcements that they made this week, Firefly generated works that meet our generative AI submission guidelines will now be eligible for submission to Adobe Stock. So if you generate images with Adobe Firefly and submit them to Adobe stock and they sell, you get commissions on the AI generated images that you're selling on Adobe stock. And speaking of stock Photography. I want to tell you about today's sponsor, which is Wirestock. Wirestock is a single website where you can submit your stock photography and AI generated images to one platform and it will redistribute it to all of the stock platforms for you. And it works with AI images for the websites that allow it sites, including Adobe Stock, which we just mentioned, 123rf, Freepik, Dreamtime, Imago and others. In the month of August, Wirestock saw a huge increase in sales of AI generated images. So not only can you submit your AI generated images, they actually sell pretty well. And one thing that I really love about Wirestock's website is they've got this Explore tab up here and you can actually see which types of images are the best selling images right now across stock platforms. And then you can go and generate more images with AI that are similar to the ones that are selling well. What's even cooler is you do not even need a tool like Midjourney or any of the other AI art generators. You can actually generate directly from within Wirestock's website right now. In fact, Wirestock sent me some images and said, show off these images as some of the ones that are the best selling images so your viewers can get an idea of what's working. Like this taco, this construction worker. They even told me that fantasy images like this one with two dinosaurs drinking coffee are doing really, really well right now on stock image sites. Here's another image of a tattooed woman and an image of a woman on a bridge and somebody standing on a mountaintop and a cat space suit thing and a colorful brain light bulb book. Again, you don't need a tool like Midjourney. You can go straight to Wirestock, generate the images, submit them straight through Wirestock, have them distributed to all of the stock websites for you that allow AI generated images and earn royalties when they sell. I put the link in the description and Wirestock said that the first hundred people that sign up through my link will get 30% off one of the premium plans. But don't worry, if you're not one of the first hundred, they're still going to hook you up with 20% off just for using my link. Once again, the link is in the description. And thank you so much again to Wirestock for sponsoring this video. Now speaking of AI art, let's check in with our good friend Ali Jules because she hung out on the mid journey office hours and gave us one of her epic recaps. Now from the recap, it doesn't seem like, there's a whole lot of news to report on from the Mid Journey camp. They're still working on 5.3. Version 6 is going to be a much bigger jump in performance than version four to version five was. It's going to do pretty much everything better. They're still working on the website, and from what I gathered, one of the big things that they talked about on the Mid Journey office hours this week was that they're working on 3D and video. I guess David really, really was talk about how excited he was for those 3D results that he's getting. So something to really look out for from midjourney. Sticking on the AI art theme here for a moment, a new fast diffusion image generation model called Burst Chin was released on Hugging Face this week. And if you're curious, verstion means little sausage or hot dog. Now, this one works slightly different than other diffusion models where it does the diffusion process on a very, very tiny compressed image as small as 12 by 12 before then upscaling it. This makes this diffusion process a lot faster than other diffusion models. And this one's available to use right over on Hugging Face right now. So I can enter a prompt like a flying wolf wearing a Superman cape. Click, run. And look how quickly this is actually generating. This is happening in real time right now. And here's the two images we got. We got like a Superman wolf here and one that looks more like a wolf with a cape. Here's what it generated with the prompt Astronaut in a jungle Cold color palette, muted colors, detailed 8K, not too bad. And then finally, in the AI art world, I actually got access to a sneak peek. I think this is the first time this is ever being shown. But Leonardo AI gave me a preview of their upcoming storyboards feature, and this is just an example, but it looks like it's going to be able to create multiple scenes from a prompt and essentially generate a storyboard for a film or for a comic or for whatever you want to use it for. Now, I can't see the whole prompt here, but it looks like the prompt is something like In a future Earth, silicon based life forms, remnants of a long extinct civilization, have seamlessly integrated with humanity as captivating headpieces. Advocates like journalist Leela hail them as I don't know the rest. But then look at the storyboard that it created. Right here's scene one or scene zero of Aeon City at night with a description of the city. And here's scene one, Modern home living room at night with Leela, scene Two, Modern home Living room at night with Sam, Leela and Sam. And we can get a sneak peek at some of the controls we'll have available to us, like pose control and changing the style per storyboard image. Scene 3, broadcasting studio in the daytime with Leela. And then we actually have the dialogue here. Ready, Korvax. Korvax. Always Leela. Today we have a special guest, Korvax, the spokesperson for our silicon based allies. So it's actually breaking out the images and the story and putting together what could potentially become a full little short film for you. Modern home living room, a coffee shop in the morning, another scene in the coffee shop, and then you can see, here's what it could look like with pose control turned on. You turn on pose control, and you can adjust the way the person is sitting in this scene specifically. Now, again, these are just preview shots. The final version might look a little bit different than this. Some of the features may not roll out immediately when this feature is dropped. I don't know the exact date when this is going to be ready, but they sent me a sneak peek. And me being who I, I wanted to share the sneak peek with you. So excited about that coming out from Leonardo AI pretty soon. All right, let's talk about AI Video, because at the end of last week, Runway made an Update to gen 2. And unless you've been on X and paying attention to these updates, you might not have noticed. But if I come to Gen 2 and I click on motion down here. Now, this motion scale came out a couple weeks ago. I actually played with that in a news video, like two news videos ago. However, there's now a button that says custom camera control. And if I open this up, I can actually tell the camera to move horizontally, vertically, zoom in, roll, and we have the speed change here. So I'm going to grab this random image. This is actually an image made with Leonardo that the people over at Leonardo actually generated as a kind of a cool example of what Leonardo was capable of. But I've got it on my computer, so let's go ahead and play with this image. I'm going to take this image, I'm going to drag and drop it right here, and let's actually have it zoom in while rolling clockwise. And it looks like you can kind of turn on two controls at once. Actually, it looks like you can do three at once if you're doing horizontal, vertical and zoom. But let's go ahead and do zoom and rol clockwise. Let's turn the speed up a little bit more let's go to seven just to see what it does. Let's click Generate. And here's what that looks like. It actually is zooming in a little bit with maybe a little bit of a pan. I'm not seeing the rotation so much this time. Let's do a left horizontal pan, so maybe it'll show off more of the tree while also zooming out. And here's what that looks like. It should be a pan left in a zoom out. And that one looks a little bit more like a pan left and a zoom out. This one, I feel like followed my instructions a little bit closer than the first generation. Now, not to be outdone, Pika Labs, a tool that has very similar functionality to Runway gen 2. In fact, Pika Labs, I believe, actually rolled it out first. They went and added pretty much the same features you could see on September 11th here. New feature. Control the camera movement using dash camera parameter. So zoom use camera, zoom in, out, pan, use camera pan up, down, left, right. You can use rotate clockwise, counterclockwise, anti clockwise. And you can also set the intensity of the motion just like you can inside of Runway gen 2. Now, Pika Labs is one that you have to use inside of their discord, but you can message the bot directly. So if I right click on the Pikabot up here, click message. So we can type slash, create. If we click on this one more here, we can select an image. Let's go ahead and use the exact same image so we can kind of get a side by side comparison. And then for our prompt, we'll go dash camera zoom in. And this one only lets us do one camera motion at a time. I can't do a camera zoom in and a pan up at the same time. I tried that and it gave me an error. There might be a workaround. I'm just not aware of it. So I'm gonna do dash camera zoom in and hit enter. And let's see what we get out of it. And here's what it created with our zoom. I actually like this one slightly better. You know, Pika does put the watermark on it, but just kind of analyzing this image. If you look at the guy down in the bottom, it actually looks like he's walking, and it looks fairly coherent. And look at the leaves right here on the tree. As it zooms in, you can kind of see the leaves swaying in the wind. There's just a details that I'm getting out of Pika that I wasn't really seeing with the Runway versions. One More time. I'm going to prompt it with camera, rotate clockwise and see if we can get more of the rotation effect from Pika than we actually did out of the Runway. Yeah, we're definitely getting much more of that rotation effect from this one. Again, it's only a three second short clip. Runway does have that feature where you can extend it up to 16 or 18 seconds or something like that and make the video longer and longer pros and cons to both Pika and Runway, but both of them have this feature right now. Pika has water, watermarks and much shorter clips. Runway doesn't have watermarks. You can do longer clips. But I also think what we're getting out of Pika looks slightly better. I guess the trade off is up to you which one you want to work with. At the end of last week, Roblox announced that they're going to add a conversational assistant inside of their AI game creation tool where you can just tell it what you want to see inside of the game scene and it will actually just create whatever it is you're asking for. You could change the texture on your polar bear, you can add bonfires, change the scenery, and all you do is type in instructions and it creates that scene for you. This will make game creation easier than ever for anybody. Now, it doesn't have the amount of depth and customizability as you'd get from something like Unreal Engine or Unity, which by the way, Unity is basically burning their business to the ground this week. But that's a story for a different day. This is going to be an amazing sort of beginner game creation tool where anybody can create whatever they imagine. It just might not have all of the customizability around exactly what players are capable and aren't capable of doing. It looks really fun and I've never used Roblox in my life, but I'm really excited to get my hands on this and put it through its motions. Now, as you probably know, I record these videos on Thursdays and this news came to my attention literally right before I hit record on this video. And I went, I need to share that. This is called V0 by Vercel and if we take a peek at it, it's basically a user interface designer where you prompt an entire user interface in video. They prompt a HERO for an email client app. It creates this landing page. Welcome to our email client with sign up and login buttons. It looked like it gave them an A B and a C option. They selected option B which actually has an opt in form on it. They select the main title tag here and then they enter the prompt. Add a gradient to the headline. It adds a gradient to the headline. They select the text again. Nope. Make it black and white. Makes the gradient black and white. Gradient should be top to bottom. Changes the gradient to top to bottom. Invert the gradient. It inverts it. Then they click on the code button and it shows them the code that they could copy and paste it and have it for their website. And if you head over to v0.dev, you can see some of the demo stuff that they've already created in here. A tweet ui. You click on this, you can see here's a tweet UI that it created. You can actually see the various iterations that they had built out and tested and changed. And you can see the prompt. Add a profile pic. That's not how you do 500 pixels in tailwind. Add more spacing on top. Replace tweet content with this specific content. A table of financial invoices. And here's a design of a table with financial invoices. And we can see the prompt history of how they got here over on the left. So if we look all the way at version 0, this was what version 0 looked like. You can click through the prompts and see each iteration of it until it came out with the final version, B20 right here. And if you like it, you come up to code, click here. Here's the code for this page. Pretty, pretty, pretty snazzy. I think right now this page just kind of shows you some examples, but I'm imagining that pretty soon we'll have access to be able to build stuff with this. And there's all sorts of examples that you can click around and play with on this site. In other pretty big news, this week, Stability AI announced Stable Audio, which is a tool where you can enter a prompt and get a song out of it. It's not quite the same as something like Chirp by Suno, where you can actually get a song that has the instrumental, the lyrics and actually the singing over it. It's more closely related to Meta and Google's version, where you're just going to get an instrumental song out of it or just sound effects. Like, for example, they have one of a car passing by.

An example of a drum solo.

Now, this is actually available for anybody to play with over@stableaudio.com however, every time I've gone to the site so far, it says we're seeing a lot of traffic. The launch of Stable Audio has generated a lot of interest and our systems are having trouble keeping up. I was able to generate one song so far, but then every song I've tried to generate after that, it sort of loaded forever. Never actually generated the song. So if you click on Generate Music here, let's see if we can get it to load for us. We get a text prompt box over here. Let's do an early 2000s pop punk anthem. Distorted guitars, power chords, fast drums. Right now you can generate durations up to 45 seconds. So let's go ahead and generate that. Maybe it'll work. This has been where it sort of broke down for me in the past. It just kind of loaded forever. It never actually generated my song. It says it's generating, so I'm going to give it a sec. All right, so let's hear our early 2000s pop puck anthem.

Not amazing, but maybe you can get some good stuff out of it. And it's the worst it's ever going to be. I imagine this will get better and better and better. All right, I got a lot more AI news to get through, so I'm going to sort of rapid pace, go through a bunch more of what's been happening this week. There's an article at the end of last week that ChatGPT is losing some of its hype as traffic falls for a third month in a row. However, I was also reading an article earlier this week, which I couldn't find again, that said ChatGPT was starting to see a slight uptick now that schools were back in session, that three months of slowing down might have been the summer months when people were out of school. And now that people are going back to school, maybe we'll see it turn a corner. But also, I think a lot of that is due to competition like Character, AI and Claude and some of these other tools that are coming out, and people are liking them just as much, if not better than ChatGPT. And in more bad news for OpenAI, another group of writers is suing OpenAI over copyright claims. Michael Chabin, David Henry Huang and other artists have filed a lawsuit against the maker of Chat GPT. They're basically claiming that Chat GPT is able to summarize and analyze the content written by the authors, saying it's only possible if OpenAI trained its GPT large language model on their works, and adds that these outputs are actually derivative works that infringe on their copyrights. This is the second lawsuit from authors in the last few weeks. The first one from authors like Sarah Silverman, Christopher golden, and Richard Kadri. Now, if I had to guess, OpenAI is probably going to win these lawsuits because the way that the content was used was transformative enough that it falls under fair use. But like I've said in past videos, I'm not a lawyer, still haven't gone and become a lawyer since the last video. So take what I'm saying as a grain of salt, but I actually do think OpenAI will come out ahead on these ones. But you never know how it's going to play out. In other ChatGPT news, this article came out this week about how the computers that ChatGPT was trained on are located in Des Moines, Iowa. And while these computers were running, it was reported that there was huge spikes in water usage. They estimated that Chatgpt gulps up 500 milliliters of water, about the same as a 16 ounce water bottle every time you ask it a series of five to 50 prompts. It measures indirect water usage, such as the water to cool power plants that supply the data centers with electricity. In other bad news for OpenAI, Meta is trying to develop a new, more powerful AI system, more powerful than the current GPT4. Now, this isn't necessarily new news, it's just an article was written about it in the Wall Street Journal. I talked about it in a news video a couple weeks ago that rumors were flying around that they want Llama 3 to be more powerful than GPT 4, and somebody at Facebook confirmed it in a tweet or a text message or something like that. But I wanted to point it out again because it seems to be sort of at the top of the AI hype right now. Although the news is kind of a couple weeks old, Wall Street Journal is just now kind of picking up on it and sharing it this week. Speaking of tools that are going kind of head to head with ChatGPT, I would argue that it's a slightly different use case, but character AI surpasses chat GPT in mobile app usage in the US now, character AI is about creating sort of characters with personalities. You can create a character that's supposed to be like Mark Zuckerberg and takes on some of the traits of Zuckerberg, or a character like Elon, or a character like Mega man or Sleeping Beauty or whoever you want, and then it kind of takes on the role of that character and then you can have conversations with it. It seems like it's being used more for entertainment purposes than actual like research. And as a tool like ChatGPT is. And it also seems to be something that younger generations are really using a lot and having fun with because they can and mess around and joke around and have fun with these fictional characters online. Now, character AI hasn't passed ChatGPT in usage on a computer, but on mobile phones. It seems that character AI is actually getting more use these days than ChatGPT is. EBay is rolling out a magical listing tool that harnesses the power of AI to make selling on ebay faster. Their magical listing tool uses artificial intelligence to extrapolate details about listings from the images. Sellers can list items with ease and buyers can access more information about potential purchases. So Basically eBay's using AI to make it easier to sell stuff, upload the photos of what you're trying to sell. It uses AI to try to generate some descriptions and some extra information so that the process of selling something on ebay just gets a little bit quicker. This week Amazon announced that authors must disclose the use of AI on Amazon Kindle. There's been a lot of books coming out on Kindle where people just generated them quickly using AI, uploaded them. They're very low effort books. In fact, there was a recent news article about how a book about foraging for mushrooms was generated with AI, but it was wasn't factually accurate and people could have potentially poisoned themselves by picking the wrong mushrooms because these low effort books were generated with AI. So now Amazon is doing something about that and saying if this book is AI, you must now disclose it before people buy it. They're going to put warnings on some of the books if they are generated by AI to let you know that this book was generated by AI. But what Amazon is not opposed to is using AI for your product descriptions. Because Amazon just launched a generative AI to help sellers write their product descriptions. I don't need to go too deep into this. It's pretty much exactly what it sounds like. You tell it a little bit about the product you're trying to sell on Amazon and Amazon is going to help you fill out the details. We saw this in Shopify already. Now Amazon's doing it too. Coca Cola is getting into the AI game. They just released Y3000 limited edition Coke Zero, in fact. One second. I'll be right back.

So apparently AI helped design the flavor of this Coke and they helped design the actual graphic imagery that you see on the can. And I've never tried this before. This is literally, I just got this can and I have no idea what it tastes like. I'm gonna try it out Right now. I'll let you know what I think. Kind of tastes like Coke Zero with maybe a little bit of extra fruitiness. I can't quite put my finger on the flavor. Definitely tastes like a diet soda, like an artificial ish, like cherry flavor or something. I can't put my finger on it, quite honestly. It's not bad if you like the taste of Coke Zero already. You'll probably like the taste of this. But yeah, Coke's in the game. Now. I've mentioned this in past videos and I want to keep this narrative going. I really think that AI is going to have massive implications on healthcare. And this story right here is kind of example A or I don't know D or however many examples I've given so far. A boy saw 17 doctors over three years for chronic pain and none of them were able to diagnose dose what issue he had. They plugged his symptoms into chat GPT and ChatGPT found the diagnosis. They took the diagnosis that ChatGPT gave them to a doctor. The doctor confirmed the diagnosis and provided the boy with treatment. Says after receiving the diagnosis, Alex underwent surgery to fix his tethered cord syndrome a few weeks ago and is still recovering. And in another story that came out this week, AI can actually detect eye disease and risk of Parkinson's from retinal images. Researchers have developed a model trained similarly to ChatGPT that can be adapted to evaluate multiple, multiple health conditions. I really think we're gonna see a ramping up of these stories where AI is helping with medical diagnosis, where AI is helping detect things early, and where AI actually helps discover new drugs. I think we're gonna see so much of this. And yes, there's a lot of negatives that could potentially come along with AI, but I think the amount of positives we're gonna get are gonna so outweigh the negatives. And finally, to wrap up the news for this week, many of the leaders in the AI world met with the US Senate in a private closed do discuss how AI should be regulated. Now, a lot of people wonder why it had to be private and closed door, but the result of it was that pretty much everybody that's developing AI right now wants some form of regulation. They want some sort of oversight on how it should be handled. Elon actually did an interview with the media after the event. And although Mr. Moneybags back here kind of steals the show, here's what Elon had to say.

It was very civilized discussion, actually, among some of the smartest people in the world. I think Something good will come of this. I think this meeting may go on history as being very important for the future of civilization. Senator Schumer, or leader Schumer did ask everyone to raise their hands in the room to see if they were in favor of AI regulation. And I believe almost everyone did.

Yeah.

So that's a good sign.

And then right before he got in his car to leave, he was asked this question, is AI going to kill us all?

Mr. Musk, I hope not. You should think of the future as a series of problems, probabilities, as opposed to certainties.

There is some chance that above zero, that AI will kill us all. I think it's low, but there's some chance. I think we should also consider the fragility of human civilization. And if you study history, there's a rise and fall to every civilization. Every civilization has sort of a lifespan and.

You know, so we want us to last as long as possible.

And that's what I got for you today. Thank you so much for tuning into this video. Hopefully you came across some AI news that you didn't hear about this week, because there was definitely a lot of it and if you weren't paying attention, you probably missed most of it. But that's what I'm here for. I'm trying my best to keep up with it every single day and bring it to you and keep you in the loop. Next week I'm actually headed out to New York for some sort of secret Microsoft event where they're going to be making some sort of announcement that I have no idea about yet. So I might have one or two less videos than normal next week, but I should have some really cool insights and info from that Microsoft event. So really excited to share that with you in a video within the next couple of weeks. And if you haven't already, make sure you check out FutureTools IE where I curate the latest AI news on a daily basis. I also curate all the coolest tools that I come across every single day on the main homepage here. And I have a free newsletter where a couple times a week I'll keep you up to date with all the coolest tools and all the latest news that I come across straight to your inbox. So, so many ways to stay in the loop with AI news. But if you like breakdowns like this and you just want that once a week, Friday or sometimes Saturday news breakdown, make sure you subscribe to this channel and like this video and I'll make sure more videos like this show up in your YouTube feed. Once again, thank you so much for hanging out nerding out with me today. So much cool news is happening right now. I know the hype is dying down in the AI world a little bit, but the news in the AI world is still ramping up and there is so many exciting advancements and I'm here for it all. I'm so excited to tell you about them. Even if everybody's not as hyped about it as much as they were a few months ago, I still think it's really exciting and I still think we're making some amazing progress. I'll be here. I'll keep making these videos for you, so hopefully you'll join me for the ride. And I just can't thank you enough. Thanks once more to Wirestock for sponsoring this video. I really, really appreciate that. It really helps me keep on making videos like this without having to turn around and sell a bunch of products and stuff. So thanks again to Wirestock and thank you for tuning in. Really, really appreciate it. Peace out. See you in the next one.

Sam.