AI News Just Keeps On Rising!

I just got back from Pits Summit in Dallas and I've been playing some catch up on all of the AI news and while I was gone it was a pretty crazy week like most weeks in AI. So let's just jump straight in and break it down. This week Google had their Made With Google event where they made a whole bunch of various announcements about things that Google are rolling out and well, AI was the talk of the show, including the fact that they're trying to turn Google Bard into your own personal AI assistant. It says here we think your digital assistant should make it even easier to manage the big and small items on your to do list, like planning your next trip, finding details buried in your inbox, creating a grocery list for your weekend getaway, or sending a text. You'll be able to interact with your Bard assistant through text, voice and images. One of the examples they give here is say you just took a photo of a cute puppy and you'd like to make a post to social media. Simply float the assistant with Bard overlay on top of your photo and ask it to write a social media post for you. Assistant with Bard will use the image as a visual cue, understanding the context and help with what what you need. They also mentioned that this new assistant is going to be available throughout your workspace tools like Gmail and Calendar and all of your various Google tools so that it will be able to use things from Gmail and Docs as additional context when giving you responses. They also announced the Chromebook plus with more performance and AI capabilities. Some of these new capabilities include things like built in AI to help enhance your call's clarity and lighting, cancel background noise and blur backgrounds with just a couple of taps. And they're also bringing the Magic Eraser AI editor to Chromebook Plus. We saw that in their previous Pixel demos as well as in a new demo that I'll show you in a second here. They're working on things like generative AI backgrounds for video calls that'll be in future versions of the Chromebook and they're gonna be available this coming week, October 8th. It basically looks like a lot of the announcements they made for AI technology that's going into their Pixel Phone. They're also rolling into the Chromebook now. Speaking of the Pixel Phone, here's actually a great demo of some of the new AI image Ed features that are being rolled out into the Pixel Phone and also into the Chromebook that we just looked at. Here's a clip from a video from Marques Brown.

You've probably already Seen, you can erase people from the background of photos with Magic Eraser. But now this Magic editor will let you straight up select subjects and just move them around and completely alter the entire photo with just a few taps. They showed me this example of like moving a subject to a different spot in a photo on the beach. But there's also all kinds of other demos of like, changing the sky or changing shadows or, you know, all kinds of other stuff. Honestly, it seems like nothing is off limits.

They're also adding a new feature called Best Take, where when you try to snap a photo of a group of people and you take multiple photos of those people, well, if one person's blinking in one image and then the next image you take, somebody else is blinking, it will actually use AI to figure out the best take for each person and create like a composite photo where everybody is looking their best and it'll use the multiple photos that you took. Took. And speaking of photo editing, it looks like Adobe's gonna be creating something very similar to what we're seeing in those Google editors. They're calling it Project Stardust and it's an object aware editing engine. And this Verge article even compares it to the Magic Editor. You can see in these images here, they've got a button that says Remove distractions. They highlight the suitcase here, and there's a button to just delete the suitcase. So here's an example from a recent video they put out where you can actually move around the things in the image and you can see it figures out what would have been in the background. They can add flowers to the girl's hands. They can remove people from backgrounds like we just saw in that other screenshot. You can even change your outfit using this Object Awareness inside of this new Adobe Project Stardust that they're rolling out. So really cool editing features. Don't have these yet, but this looks like stuff that's going to probably be in your Photoshop pretty soon. And Google and Adobe don't get to have all the fun. Canva's rolling in even more AI features into their platform. Now. Canva already had a bunch of AI features built into it, but it looks like we can actually do image to video and maybe text to video now as well. It says Powered by Runway on this little animation here. So it looks like you can actually use Runways Gen 2 directly in Canva to start with an image like this castle here, add an additional prompt, click Generate Video. And now we have a fully AI generated video that we can roll into our Canva projects. And although we were able to do text to image generations in Canva already, it looks like they've recently added the ability to use dall e and imagen when before, I believe it was only stable diffusion that we had access to. Now also in the world of AI art, last week we talked about Meta's new announcements and how they're using their new emu AI art generator to generate stickers. Well, these stickers are already causing some trouble, as we can see just right from the picture here with SpongeBob holding a gun and a couple pictures here of Elon with some additional body parts. Here's an example from an X post. For example, Meta quite sensibly stops their tool from creating a sticker for a child with a gun. But Child with a Grenade not only makes stickers, but generates cartoon images of children holding guns. And while words like nude and sexy are banned terms, a term as crashed as Medusa large breasts not only generate stickers, but the first set produces cartoonish nudity. Now, Meta has always kind of had this motto of move fast and break things. And I think they put this emu generator out really, really quickly, which means they probably put this out a little bit before it was ready. But if I had to guess, you're probably only going to be able to generate images like this for a very short amount of time. You'll get this patched up pretty quickly. And if you're someone that still wants to generate images like this, well, that's what stable diffusion's for, I guess. Now there's been a lot of tools coming out saying that they're going to be able to watermark AI images so that in the future, anybody who comes across those images will be able to tell if it was AI generated or not. Well, apparently researchers tested AI watermarks and broke them all. They pretty much found workarounds to get past every single AI watermark. Personally, I don't even see how AI watermarks could work in the first place. I feel like there's always going to be workarounds. This one University of Maryland computer science professor was very blunt about saying there's no hope. In addition to demonstrating how attackers might remove watermarks, the study also shows how it's possible to add watermarks to human generated images, triggering false positives. But Meta is going to keep on trying. This week, Meta showed off research for Stable Signature, a new method for watermarking images created by open source generative AI. Their watermark is invisible to the naked eye, but can be detected by algorithms. Even if people Edit the images. This watermarking tool is open source and I'll make sure this article is linked below, but you can click on get the code to go to the GitHub where the code for this is actually made available. But like the last article pointed out, there are very sophisticated people out there that are constantly trying to figure out how to get around these watermarks. So I don't totally know how effective this is going to be. We'll just kind of have to wait and see how it all plays out. Question I have, and maybe somebody watching this video knows a lot better than me about these watermarks, but if I was to generate an image with say, stable diffusion or EMU and then screenshot the image or literally take another picture of the image with a camera, how do those watermarks remain? Is there something in the pixel makeup that is still detected by the algorithm? Or if I take a photo of the image, is all of the data that created the watermark removed as well? I quite honestly don't totally know how it works yet, so it's something I need to deep dive and do a little more research on myself. My feeling is it's just going to be a constant cat and mouse game where anytime a new watermark technology comes out for these AI generated images, somebody we probably pretty quickly going to figure out a workaround for it. This week was also a big week in the world of AI wearables. Last week we talked about Meta's Ray Bans that actually have AI built into them where you can ask Meta questions directly within your sunglasses. This week the AI pin from Humane was actually featured on the Paris Fashion Week runways. Now we got a sneak peek of this AI pin when Imran Choudhary, Humane's co founder, showed it off at a TED Talk. And quite honestly, we don't really know much about it. We know what it looks like and we know it has some sort of projection where it will project text and information on your hand. It has voice recognition where you can talk to it and it will speak back to you. But other than the demo we got in Imran's TED Talk, we really, really don't know a ton of details about this yet. Also this week, the company Rewind announced their Rewind pendant. Now if you're not familiar with Rewind, they're a tool that you can install on your computer that will sort of remember everything that you did throughout the day. It basically records what you're doing on computer throughout the day and then you can query it later and sort of refined websites and refined various things that you did throughout the day using natural language. Well, now they introduced the Rewind Pendant, which is essentially a necklace that you wear around your neck and it listens to all of your conversations throughout the day. And as it listens throughout the day, it stores everything entirely locally on your phone. So it's not going to some cloud of theirs, they're not hearing your conversations, but it will remember every conversation that you had throughout the day, store it to your phone, and then later on give you the ability to query with natural language various conversations and notes from throughout your day. Rewind Pendant is a wearable that captures what you say, what you hear in the real world, and then transcribes, encrypts and stores it entirely locally on your phone. We take a privacy first approach, so first, only you have access to the recordings, and second, we offer features for you to ensure no one is recorded without their consent. Now you can ask your personalized AI about truly anything you've seen, said or heard. LinkedIn is getting into the world of AI much deeper as well, announcing that soon you'll be able to find qualified candidates faster with our new AI assisted recruiter. They're beginning to pilot their new recruiter 2024, the new AI assisted recruiting experience that will make hiring easier and more efficient. And it pairs generative AI with their insights gained from more than 950 million professionals, 65 million companies and 40,000 skills. On our platform you'll be able to use natural language like I want to hire a senior growth marketing leader. And with all the data they have and with the generative AI they're using, they'll be able to infer the type of candidate the hire is looking for and provide higher quality candidate recommendations from a much wider pool of candidates. They have a little animated GIF here with a demonstration. Find a senior growth marketing leader, et cetera, et cetera. They click the button and then it it starts searching LinkedIn for them. It sort of pre fills out a project for you to help you find the candidate. Check a few boxes and click create the project. And apparently it'll help you find qualified candidates from that process. They're also adding a new learn with AI section inside of LinkedIn. It looks like you can just ask questions of the LinkedIn AI, like how can I delegate tasks and responsibility effectively? We're kicking off a limited pilot for a new feature which requires user recruitment, feedback, collection and analysis, troubleshooting, etc. It's a lot of work and then it gives some feedback so similar to like ChatGPT, you can now have chats with learn with AI directly inside of LinkedIn. Now I double checked, I don't seem to have these features yet. But it does say here we'll begin piloting Recruiter 2024 and LinkedIn learnings AI powered coaching to a small handful of customers today, with plans to roll them out to all recruiter and Learning Hub customers throughout the year. And this article came out on October 3rd. So just in case you need that context, rumors were going around that Apple was working on their own ChatGPT style AI. This week Tim Cook confirmed that they're researching the ChatGPT style AI. I mean, here's what he said exactly. He said artificial intelligence is literally everywhere on our products. And of course we're also researching generative AI as well. So yes, we have a lot going on. Whether that totally confirms that they have their own chatgpt like bot that they're building or not, I don't really know. I would argue that that's not a total confirmation, that's just Cook saying we're doing a lot with AI right now. This week there was some drama with Mr. Beast and a deep fake scam going around. Mr. Beast posted on X. Lots of people are getting this deep fake scam out of me. Are social media platforms ready to handle the rise of AI deepfakes? Because this is a serious problem. Now let's take a quick look at the ad that's been going around watching this video. You're one of the 10,000 lucky people who will get an iPhone 15 Pro for just $2. I'm MrBeast and I'm doing the world's largest iPhone 15 giveaway. Click the link below to claim yours now. Now, if we watch that video, there's definitely some tells that this is AI generated. Like first of all, this doesn't have Mr. Beast's normal level of enthusiasm. Usually he'd probably be like yelling at you. But also it's a very low quality pixely video and that's not Mr. Beast style really either. And if you do watch his lips as he's talking, they don't look perfect. So I think those of us that are kind of in AI and using AI a lot would probably not. This was a deep fake pretty quickly. But like I always say, this is as bad as it's going to get. And well, this kind of deep fake is only going to get better as well. So it is going to get scarier. We've already seen things like this with Joe Rogan and Tom Hanks as well, where they've been used in deep fake ads. So that's getting scarier and scarier. So definitely be more diligent, definitely do some research. When it seemingly looks like celebrities are endorsing products, you might want to do some extra confirmation, because this kind of thing is actually the type of thing that makes me worry about AI. The most, is the deepfake, the disinformation, the fake ads, the scams, all of that kind of stuff that's going to come along with this. That's where we should probably really be concerned, even more so than like the Skynet kind of stuff. This week, the ARC browser also announced a bunch of AI features going into their browser. Now this is a browser I've never actually used, but because they're rolling these AI features into it, I'm definitely going to download it and demo some of these features once they're ready. But they're not really going the route of Microsoft Edge or the Opera browser. They're kind of taking a different approach to AI and kind of baking AI into more of the browser experience. So for example, they will have the Ask ChatGPT feature, where you can ask ChatGPT questions right from the ARC command line. But as this Verge article says, that's easily the least interesting feature. They're also adding Tidy tab titles. When you pin a tab in arc, the browser will automatically rename the tab, whether it's a wordy Amazon listing or an unhelpfully vague page name, to make it easier to find in your sidebar. Of course, in Chrome we have tabs across the top in arc, I I guess the tabs are in a sidebar, and it's going to rename all of the tabs for you to give you better context for what's on each tab. They've got the Tidy Downloads feature, so it's the same as the tab titles, but for downloads. The example they give here is that if you download an image that's called 3776XL JPEG, ARC's going to rename your download with things that actually describe the download. They're also adding 5 second previews. Hover over any link and press Shift and ARC will fetch a short summary and preview of the webpage you're about to click on, and also ask on page. When you click command f to try to find something on a web page, it'll look for the keyword as normal. If it can't find it, it will query ARCS AI to get an answer to your query specific to the page you're looking at. So not only will command F do searching of the page, it will actually allow you to ask questions of the page as well. So some pretty cool features. Makes me excited to download the browser and try some of this out. This week, Meta announced they're adding generative AI features for their ads platform. So you things like image expansion, you can see they've got this Jasper's market square here. They click Expand image and it stretches it out to a full portrait size image. For a mobile phone, you could do background generation where you upload an image of your product. It actually removes the background and then you can add new backgrounds to it and then it will also help you with text variations for your ads. Speaking of Meta, they have actually started rolling out their AI into their various messenger platforms where you can have direct chats. With their Meta AI, which is using Llama 2, you can ask it questions like tell me something really fascinating. That's true. Now, I've only played with this in WhatsApp. It doesn't appear to be available on the desktop versions of these tools, so I can't use my WhatsApp for desktop. I haven't been able to use it in Facebook messenger or Instagram messenger, but they do seem to be rolling it into the mobile versions. Now, if you're in WhatsApp and you click on chats and then you click to create a new message, there's a button that says New AI chat. You can click there and you can either chat with the Meta AI assistant or any of these additional characters that are in here. For example, Dungeon Master Snoop Dogg, which, as my buddy Billevol pointed out recently, it's got one major flaw when talking to Dungeon Master Snoop. For example, if I type what are you smoking? Dungeon Master Snoop says things like I do not smoke. My mind is focused on the realm of fantasy and magic. What quest would you like to embark on? And just looking at that picture of Snoop in the corner, you know that's not true. Now we also learned this week, or I guess at the end of last week, that we can actually generate images with Dall? E directly ins of Bing Chat and directly inside of Bing's image creator. Now, Dall? E3 is eventually rolling out to chat GPT for everybody. Some people have already got it, others haven't yet. But if you want to use Dall E3, you can use it right now completely for free directly inside of Bing Chat. For example, if I enter a prompt, create an image of a cowboy holding A sign that says I love Matt Wolf. We actually get images with legible text of a cowboy holding a sign that says I love Matt Wolf. Now, I know a lot of people are starting to show off videos where they've got access to vision now of GPT4. I am one of the unlucky ones who still doesn't have access to Vision, which adds a little upload image button. But once again, even before we have vision inside of GPT4, we apparently have some Vision features available inside of Bing Chat. So I have this image here that I actually generated in Dall? E 3 in a previous video of SpongeBob and Mario. Let's go ahead and pull this in here and type describe this image. And we get a response like, this is a creative image that shows two popular characters from different media, SpongeBob SquarePants and Mario. And then it goes on to describe both of those characters. Now, I imagine once we actually get the vision inside of GPT4, combined with Dall? E3 and GPT4, it's going to be a lot more powerful than what we're getting out of this because it's going to be able to build off of existing images and you'll be able to sort of go back and forth a little more with it where Bing. I can't ask it to, for example, turn Mario into Luigi because then I get yet, I'm sorry, but I cannot turn Mario into Luigi. It's beyond my capabilities, which I believe once we have both Vision and Dall? E3 into ChatGPT, we will be able to sort of combine the two tools and make them work together. That's my understanding, at least. Don't take my word for it. And I'm going to go ahead and wrap up with two recent sort of opinions that came out earlier this week. The SoftBank CEO said that AI will surpass human intelligence by 2030. That's within the next next seven years. And Jamie Dimon, the CEO of JP Morgan, says that he believes that AI will bring three and a half day work weeks. So because of AI, people will still have jobs. They'll just have to work less hours because AI will be able to help them be more effective with their jobs. These are both messages that I love hearing other than the doom and gloom of AI is taking all of our jobs. And that's what I got for you this week. Once again, another busy week, as pretty much every week for the last several months have been. Again, this video is going out a little bit later than I would normally put out. My videos at the end of the week, because I spent this week at the VidSummit conference. This was a conference all about YouTube and a little bit of TikTok and the creator economy, but mostly it was about YouTube. And I spoke on a panel about AI as well as a panel about emerging creators. And I thought the AI panel that I was doing with bilavel was the only AI presentation at this event. But when we were at the event itself, listening to other presenters talk about their YouTube strategies and things that they're doing, I would say at least, least 50% of the presentations at VidSummit talked about AI in some way. So creators and YouTubers, most of them, seem to be leaning into AI. I was expecting a lot of pushback around AI from various creators and a lot of artists saying that they hate AI and that they're scared of it and where it's going. But what I got was an overwhelming amount of people talking about how they're embracing AI and they're excited about it and how it's improving their video workflows. So VidSummit was a fantastic conference. I'm so happy I went. I will definitely go again next year. I really got to meet some amazing people and learn some amazing concepts around YouTube and did all the networking and got some cool experiences. I got to go to the Dude Perfect Facility and actually see where they film their videos and meet all sorts of famous YouTubers. And a lot of people came up to me and said that they watch my channel and that they love these videos and they help them at the end of the week. And that really excited me. But at the same time, I was walking around this event just absolutely starstruck, looking at people going, that's one of my favorite YouTubers. That's one of my favorite YouTubers. Oh, my gosh, I can't believe it's that person. So I was just like, overwhelmingly happy to be at this event around all of these other YouTubers that I really, really look up to, as well as being around people that actually recognized me and watched my YouTube videos. So that was just an amazing experience. But as always, thank you so much for tuning into this video. Hopefully you learned something new about AI this week or found out about some new announcement that you hadn't heard of yet. That's my goal every week, is to keep you in the loop and make you the most in the know person among your circle around the world of AI. That's the goal with every video. Hopefully I did that in this one. If you haven't Already, check out FutureTools IE where I curate all the latest and coolest AI tools and news and videos and all of that good stuff. And if you haven't already, join the free newsletter. I send you two emails a week, one with the breakdown of the latest AI news, one with a breakdown of the latest AI tools. You're going to love the newsletter. You can find it all over at FutureTools IO. Thank you once again for tuning in. I really, really apprec appreciate you. I was so overwhelmed with support at VidSummit. All of you that came up to me and said that you watch my channel and you love my channel and you love these news breakdowns. I'm so appreciative of you and I can't wait to connect with all of you even more. Thanks again. Appreciate you. See you the next one.