A New Era of AI Has Just Begun

When it comes to AI, we've officially entered a new era. We're now in the era where all of these cool AI advancements and all of this cool AI technology that we've been seeing released over the last 9 to 12 months is now just being integrated into all of the tools that most people use on a daily basis. We're moving into this era where you're likely no longer going to be using a whole bunch of different AI tools. Instead, the big companies that you're already used to working with are just going to have all of the AI tools available to you. And this week is the week that that became the most apparent, and we're seeing most of that start to happen. And I know I'm starting to sound like a broken record here, but the last four weeks in a row, every single week, has been bigger than the last. And this week is no exception. This week was a monumental week in the world of AI. Just so much stuff happened. And I'm going to do my best to break it down as quickly and in as helpful of a way as possible to bring you up to speed, starting with the fact that ChatGPT can now see, hear and speak. For example, you'll now be able to take pictures of, for example, a bike and then say, help me lower my bike seat. And it can actually use the picture for context, understand what it needs to do, and then give you instructions based on your query and what it sees in the image to help you lower your bike seat. They take another picture, they circle the thing that they think they need to fix. They ask ChatGPT again, and then ChatGPT basically clarifies whether or not that's the proper lever or not. So ChatGPT, similar to Siri, will now actually be able to listen to you when you give it your query.

We want to hear a bedtime story. Tell us a story about the super duper sunflower hedgehog named Larry. Start with telling us a little bit about him. Larry was a unique hedgehog unlike any other. He had bright sunflower petals instead of spines.

Will then obviously respond to you similar to the way Siri will. And they've got several voices that you can choose between. They've got juniper.

Once in a tranquil woodland, there was a there's sky mama cat named Lila Cove.

Any day she cuddled with her playful kitten Milo under the ember under the shade of an old oak tree and.

Breeze began her voice soft and gentle.

So whatever voice you prefer, you can have Chat GPT respond to you in that voice. Now these speaking features are going to be available in the mobile app. Let me just double check here. Nope, I still don't have it in my app. And when I check my own Chat GPT here, I also don't have the ability to upload images yet. However, they made it sound like they're going to be rolling it out in the very beginning of October, which at the time this video is going live is in the next couple days.

So.

So hopefully we get all of these features rolled out in the next couple of days. Now, there are a handful of people that have had access to it already and some pretty fun examples of how this can be used have already started circulating. For example, this video posted by McKay Wrigley here, they posted an image of a human cell here directly into ChatGPT eating this diagram of a human cell into GPT4 vision. And I'm simulating, hey, I'm a ninth grade biology student struggling in school. Help me understand this. Check this out. It then went on to break down every aspect of that cell image. He then basically asked it to dumb it down for him and it explained it in simpler terms. Peter Yang here shared this screenshot saying kids will never do homework again. Where you could see he just uploaded a math assignment with all of these math problems and then it just says certainly let's solve these addition problems. And then one by one it just put the answers to all of these questions for him right into the chat. And then here's an example from Patrick Gota where they used two chatgpt next to each other to have a total bro fest conversation.

How's it going bro? I've been hearing some buz about your voice capabilities. Fill me in, would you? What's good, bro? So here's the deal. I've been upgraded to have this voice capability where users can chat with me using their voice. And guess what bro? I can respond back with a supernatural sounding voice. It's like we're having a legit convo, just you and me vibing. It's next level stuff, bro.

Now, I cut out some of the delay just to keep the video moving a little bit, but there is a little bit of a delay between when a question is asked and when the question was answered. I do record these on Thursday. If you're watching this on Friday, there's a chance you got it already. I don't know. But one thing that chat GPT did give us, well, gave us back at least, was that we can now browse the Internet once again. So if you're in your chat GPT account and you hover over GPT.

4.

You should have this Browse with Bing option now and if you don't come down to your username down here, click on the three dots, click on Settings and Beta. Click on Beta Features and make sure Browse with Bing is turned on and you too will have the Browse with Bing feature, assuming you're a ChatGPT plus user. With that turned on, you can ask more current questions and it will once again, like it did months ago, actually browse the web for you again. This week Meta also held their annual Connect event where they made a whole bunch of announcements. They started with announcements about the new Meta Quest 3 which looked pretty awesome but not super AI based, so I'm not going to go too deep into that, but they did make a whole bunch of AI announcements. I don't want to spend too much time on the AI announcements because I can literally make an entire video just about Meta's AI announcements from this event. In fact, I actually did make an entire video just about Meta's AI announcements from this event. You can can find it here. It's currently called Meta just blew everyone's minds. And you can look for this picture of me wearing AI sunglasses. That is an entire breakdown of Meta's event. So I'm going to move through the Meta news in this video really quickly because I got a whole other video that you can go check out that breaks down all of that news as well. But for the super quick rundown, they announced their emu image generation model and the first use case of this emu image generation model is to create these custom stickers that you'll be able to create directly inside of your Facebook, Facebook Messenger, Instagram messenger, And inside of WhatsApp, they're also releasing an image editing AI that's going to be part of Instagram and it uses their Segment Anything model. You can do things like turn pictures into watercolor pictures, remove backgrounds from images, and then make yourself surrounded by puppies like that. They also announced Meta AI which is going to be an AI powered by llama2 and it's going to live right inside of all of your messenger platforms. So whether using Facebook Messenger, Instagram messenger or WhatsApp, you're going to be able to chat with Meta AI which is their sort of response to chat GPT and they're also creating a whole bunch of custom AIs that are sort of trained on specific things and they use celebrity faces to represent the various areas of expertise. So some of the examples are Charlie d' Amelio as Coco the dance enthusiast, Chris Paul as Perry the pro golfer, and Snoop Dogg the dungeon master. So you can have conversations with llama2 chatbots that are custom trained on these specific niche conversations. They also introduced the new Ray Ban Smart sunglasses, which have a camera built into them. They can actually hear you speak. They've got headphones in them so you can hear what's going on inside of whatever you're listening to. But they've also got AI built into them. Now, basically the idea is you put these sunglasses on and then while you're wearing them, if you ever have a question that you want answered, you can say, hey, Meta, what is the square root of 7362? And then in its speakers that are right by your ear, it's going to be able to answer the question. Now they also talked about how these sunglasses will be multimodal and they're going to get an update in the coming months and the update is going to make it so the glasses can see what you see. So you can literally walk up to something and say, hey, meta, what is this thing that I'm looking at? Because it's got a little camera on it. The glasses will be able to see what you're looking at and then respond with details about what you're looking at. Or if you're traveling and you see a sign that's in Japanese and you don't know how to read it, you can say, hey, meta, what does this sign say? It will read the sign and translate it for you. So when these sunglasses become multimodal, they become infinitely more usable and powerful because now you can get information using AI and the vision about anything that's in your surrounding. Again, I did an entire 19 minute breakdown of the whole event. There's a lot of details, a lot of nuances of what they talked about. Some of it's really cool. Most of the comments on that video are, I hate Meta or I hate Zuckerberg or whatever, but that's besides the point. I personally think the technology that those guys are putting out right now is really cool. And I really like that Meta is taking this more open source approach, but that's just my personal opinion. Next up, I want to tell you about Today's sponsor, Coda IO. Coda is an AI powered work assistant. Inside Coda, you can create docs and collaborate with other team members, but it's also got the power of AI built in. It's also got thousands of app integrations so that you can pull data from any of the tools that you're already using and work with them all in one place. It's built around four main features. You've got your AI assistant that can help you create docs, outlines, first drafts, and more. You've got the AI column, which can essentially help you build spreadsheets and then populate those spreadsheets using AI. You've got the AI block, which helps you find critical information fast by creating summaries or finding action items from scattered notes and data. And they've recently added AI Chat, which is this side panel down here where you can have conversations with your documents and data. The user interface is super clean. You can keep all of your documents organized. And when you need to find information from a document, it's super easy to just press this little chat button and type things like who leads new hire onboarding? Do a quick search. It's going to search this document. Here at our company, New Hire Onboarding is led by Rocky, our onboarding expert. That information's somewhere in this doc. Here we go. Rocky, our onboarding expert. See it? Found it for me. Now, here's an example table of customer call notes. You can see there's the full set of call notes here. If we take a peek at this, it looks like it's the transcription from the entire call. And then over here, I did a quick summary of what they talked about on the call and added some tags based on the conversation. This was done with AI. Here I could jump over to chat and say, based on all of the calls, what was the most common feedback based on the customer call notes? The most common feedback from customers was the request for more color options for buttons. Within the ACME platform, Bob, Emma, Sarah, and Samantha expressed the desire to have a wider range of color options to align with their branding and personal preference references. So Coda I.O. is a great platform to organize all of your notes, organize all of your data, have AI actually help you summarize and further organize it, and then actually chat and query any of the data and any of the content that you've created. And if you go to Coda IO, Matt, they've offered to give viewers of this channel either 3 months free or 25% off. So check out Coda IO for a very familiar, very easy user interface to keep all of your notes organized, automated and easy to query using new AI chat feature. Again, that's Coda IE, Matt, for either three months free or 25% off. Thanks once again to Coda IO for sponsoring this Video. Next up, let's talk about Windows. Now. Last week I was at the Windows event in New York where they announced that Windows 11 was getting an AI upgrade where everybody who has Windows 11 is just going to have AI built in. Well, this week they finally rolled it out. I actually just installed it on my computer. Now I'll probably make another video where I deep dive into it a little bit further, but once it's installed on your computer, it adds this little button down here that says Copilot Preview. I click on this or if I hit Windows C, it brings up this copilot over here. Now at the moment, it pretty much does the same things that the Bing Chat does inside of Microsoft Edge, with also the ability to open up some apps on your computer. Sort of. You could pretty much ask it any questions that you can ask Bing Chat. You can also have it generate images with Dall E. So if I say create a watercolor image of a wolf, I get some watercolor images of a wolf. Now I will say it does run fairly slow. If you've used Bing Chat instead of Edge, it definitely runs faster in Edge, but it also can do some very simple things like open up apps for you on your computer. So if I type in open the calculator app, it brings up a dialog box asking if I'm sure that I want it to open up the app and then I can click yes and then it'll pop open my calculator. However, I kind of find that a lot slower than just typing in calculator and pulling open the app right here. So I don't know how much time that really saves you. My computer is currently in dark. I usually leave it in dark mode. I can say switch to light mode. It asks if I want to turn off dark mode, I can click yes and then it switches it to light mode. For me, this process would honestly be a lot smoother if it didn't ask me if I'm sure. I do think that this sidebar still leaves a lot to be desired. This is the very first early iteration of it. If you watch the keynote from Microsoft from last week, they did show off a whole bunch of features that are going to be coming into this sidebar that aren't quite ready yet. So while it's cool that we're seeing this kind of stuff just being added directly into Windows, I don't see myself actually using this a whole heck of a lot yet. Speaking of operating systems, macOS Sonoma rolled out this week. Actually, it rolled out the exact same day that Windows rolled out their new update. Not sure if that was coincidental or not. Now, not too much AI features to talk about in this. There is one very, very, very minimal minor AI mention inside of their announcement in that their keyboard, the autocorrect is receiving a comprehensive update and it's using state of the art on device machine learning language model. That accuracy, you're getting a little bit more accurate of an Autocorrect here with this new Apple update, but other than that, not really much AI to talk about there. This week Amazon and Anthropic partnered up. If you've ever used Claude, then you've used Anthropic's tool and in my opinion Claude is a lot better at a lot of things over Chat GPT. For example, summarizing long articles, Claude is so much better than ChatGPT at that at the moment, but this announcement was pretty big. Amazon will invest up to $4 billion into anthropic AWS. Amazon's Web Services will become Anthropic primary cloud provider. Now Amazon does have their own model called Amazon Bedrock that they've been working on. I find this really interesting because Anthropic actually already has some backing from Google as well, but it definitely feels like Google is kind of going more all in on their Palm 2e and their Gemini from DeepMind and not really putting a ton of focus on Anthropic. So it sounds like Amazon and Anthropic are teaming up to go against the Googles, the Microsoft ChatGPT OpenAI llamas. Amazon wants to be in this game as well. They likely saw that getting together with Anthropic was the best way that they can actually become more relevant in the large language model space. Speaking of Amazon, they also announced the general availability of their Amazon Bedrock. Amazon's Bedrock is going to help larger companies fine tune their own models for their own specific needs. And with this wide rollout from Amazon, it's going to make it a lot easier for companies of all sizes to start creating their own fine tuned AI models not to be left behind in the AI race. Cloudflare, a popular CDN content delivery network, is launching their own suite of AI tools to help customers deploy and run models. So it sounds like Cloudflare has the infrastructure because of their content delivery network and all of the online websites that sort of pass through their servers. So they're using some of their GPU capacity at Cloudflare to let people run AI models. At least that's sort of what I'm gathering from this. This week, Snap partnered with Microsoft on ads in its My AI Chatbot feature. Basically, if you were to use Snapchat's My AI, you can ask questions like where should I get dinner near me? And it will give recommendations of places to eat near you and those recommendations could be sponsored by companies. Well, with this announcement, it was announced that Snap and Microsoft will be partnering on these link suggestions. So essentially it sounds like if you want to purchase ads and have this My AI recommend your company during chat conversations, you should be able to do that through Microsoft's advertising platform. And Microsoft will sort of act as that middleman between you, the business owner and Snapchat's AI platform. Also this week it was announced that Spotify is going to clone podcasters voices and translate them into other languages. So if you have a podcast, pretty soon Spotify subscribers will be able to listen to your podcast in their native language and it will actually sound like your voice. This week Getty made their own AI generator that's only trained on Getty images that are licensed to Getty with tools like Midjourney and Stable Diffusion and now Dall E3 and Adobe Firefly and all of these other tools just getting better and better and better. I'm not sure how much demand there really is for a Getty Images version of the same thing. And if you check out their website, you actually can't even use it. You have to request a demo and fill out a long form to be able to use it. So there's a little bit of a barrier. They want a lot of data on you just to be able to use their model. But yeah, we'll see how that plays out. Also this week, Leonardo AI announced that you can now use Loras or Low Rank Adaption models. If you're not familiar with Loras, it's not necessarily a custom trained model. It's almost like you can use a model and a Lora at the same time. A Lora is almost like additional effects or additional information that you want to provide when training an image. So for example, if I just keep this same prompt in here that generated these cool Wolf images last time and then click on Add Elements here, you can see that there's a handful of Lora models available that will give specific sort of styles to those images. So if I want to add the Inferno Lora to it, I can select this, but it also lets me select up to four different Lauras so I can also add the Tiki Lora to it. Click Confirm. Let's say I want it Weighted a little bit more on the inferno, a little bit less on the tiki. And let's click Generate. And there you go. You can see it kind of added in those tiki sort of elements with maybe a little bit of inferno with some of this explosion going on in the background. But it applied those loras and added that extra little effect to it. We saw some cool new text to video stuff roll out this week. Pika Labs now actually lets you add messages inside of the videos that you generate. So check out some of the examples that they shared. Here they got the words green inside of a forest. Here the word fresh inside of fruit. So basically, if we jump into Pika Labs here, let's do our little SN slash and let's select encrypt text. And for the message, let's put wolf. And for the prompt, let's put a wolf howling at the moon. And here's what we got out of it. Now, the motion here is subtle. You can see the wolves moving a little bit, but you could see the word wolf in the background. And as of right now, Pika Labs is still 100% free to use. Also this week, Genmo AI released their replay effects, which kind of has this cool kaleidoscope effect to it, where you can have the face sort of be stable, but then what's going on around the face is a lot of movement and motion and extra stuff going on. So pretty cool new AI video feature there. You can check that one out by going to genmo AI, come down to your little effects. You can turn the effects plugin on, and there's a whole bunch of effects libraries here, like rings moving inward. Let's put the motion in the middle somewhere. And we'll also have it loop. And let's say a wolf howling at the moon. And let's just see what we get out of this one. And it says generation complete. And here's what we got from that one. The circle around the wolf kind of shrinks and and expands and then shrinks and expands again because it's looping through twice. Definitely something I need to play around with more. I haven't messed with the settings myself a whole lot yet, so I haven't really seen what it can do. This was literally my first time testing this, and the result's pretty cool. But I definitely want to go through some of those other effects that are down here and to see what I get when playing with some of these. This week, the app Polycam announced that you can actually do Gaussian splatting inside of their app. Now, I have talked about Gaussian splatting in a few of my past videos. If you're familiar with nerfs, it's kind of a similar idea where you can take a whole bunch of pictures around a scene and then it's going to sort of stitch all of those pictures together into a 3D image. Now, that's a complete oversimplification. I've explained it better in past videos, and there's also a lot better people to explain it than me. But really, it's a way of taking a whole bunch of pictures similar to like photogrammetry or LIDAR or NERFs, where you take a whole bunch of pictures around it and then they all get stitched together into a 3D image. However, the difference between Nerfs and Gajan splatting is that Gaussian splatting actually looks a lot. In my opinion, a lot of the video footage that comes out of it looks clear, but it's also a lot faster. So the processing of all of these images into this sort of 3D object that you create, it actually processes that a lot faster so you get your result back quicker. And then you can make videos that look almost like drone shots where you're like flying around inside of a scene. And you can now do that directly within this Polycam app here. Here's a shot of a woman standing, and you can see all of the cameras around her. And I can actually move and zoom in and zoom out. And if I wanted to, I can actually create a video where I keyframe various shots and make sort of an animation moving around this person standing here. And a lot of these files can actually be imported into tools like Unreal Engine and Unity and actually used as game assets or game backgrounds. So you can check that out over at Poly Cam Splat. And they do have step by step instructions here on how to create a splat. And. And it seems pretty simple. Upload between 20 and 200 pictures of different angles of the image, and then it will do the work to sort of stitch them all together for you. This week also saw the end of the WGA strike. The Writers Guild of America, they finally came to terms and the writers are going back to work. And one of the interesting terms that they ended on was that Hollywood studios can train AI models on writers work. The argument from these large studios being that all of these works are going to get trained into AI anyway, and there's very little control they have over that. Might as well let the studios do it and have the control over how the AI is used. We also saw some cool video from Tesla's Optimus robot and that it can actually sort objects autonomously. It just knows how to sort these blocks by color fully autonomously. And it does it while dealing with the dynamic reality of our world. So as somebody is there trying to mess him up and move pieces around, you can see the robot still figures it out and still sorts them properly. Now, I know I'm probably going to get a lot of hate in the comments just for mentioning this, because so much people waste their energy on hate for people like Zuckerberg and Elon Musk and stuff like that. I don't totally get it, but I thought this was really cool and so I'm sharing it. But Lex Friedman and Mark Zuckerberg had an interview conversation in the Metaverse. So these are their Metaverse avatars that we're seeing on the screen here. This isn't actually their real face. They actually scanned in their faces. If you think of Mark Zuckerberg's old version of the metaverse, it was cartoony people that didn't have legs that just sort of floated around. Well, you can see the advancements are making some headway. And that wasn't meant to be a pun, but they were able to scan in and have this conversation in the metaverse. And the first, like 10 minutes of this conversation is Lex Friedman just being blown away by how real and intimate it felt. Zuckerberg and Lex were in two different locations, but they kept on commenting about how it truly felt like they were side by side during this conversation.

Incredible. The realism here is just incredible. Where am I? Where are you?

Mark?

You're in Austin, right?

No, I mean this place where we're shrouded by darkness with ultra realistic face, and it just feels like we're in the same room. This is really the most incredible thing I've ever seen.

I love the idea of feeling like I can have conversations with somebody, like they're in the same room with me. One of my best friends lives in New Zealand and we talk once every other week or so. And to be able to do it in the metaverse and actually see each other's faces and expressions and arm movements and body language, that to me is just going to be game changing for conversations and that sort of additional connection that you just can't get over, like a zoom call. And finally, to wrap up, let's talk about some good news in the AI world. 78% of companies say that AI has created more jobs. Now, this comes from a website called Digital Information World. I don't really know this site super well, so I can't really speak to the site's credibility. But the report actually comes from the Institute for Future of Work. And in the report that surveyed over a thousand different firms, 78% of them said that the use of AI and the incorporation of automated processes into their workflows directly contributed to new jobs being created. Created. 47% of the people that responded to the survey stated that AI had removed certain jobs that used to be available. But 67% reported that new positions opened up. So with all of this fear of AI taking away jobs at the moment, the data supposedly says that that's not actually the case, that new jobs are being created as a result of AI. Now, there are definitely functions in companies that no longer need to be done by humans, but that's always been the case, as technology's advanced. Humans are always going to try to figure out how to automate workflows and make things easier to do and simpler and cheaper and faster and better. But with that innovation, the roles that humans play have always just kind of shifted around into different roles. And that seems to be what is happening in the AI world right now, is, yes, AI is taking away jobs, but new jobs, new career paths, new opportunities are also opening up as a result. I choose to believe that this technology and these innovations and where we're going with all of this stuff is going to lead to a better life in a better world for most people. And so when I see articles that sort of validate that belief, I'm going to share them with you. Because this AI revolution, this AI technology, it's coming whether we want it to or not, whether we're ready for it or not. Might as well play in the playground of being optimistic about it and being excited about the potential and the opportunity that it'll provide for all of us. That's all I got for you this week, and by all I got for you, I know that was a load of information and news this week. This week. And I happen to know that there's quite a bit more in the pipeline coming over the next couple weeks as well that I'm not allowed to talk about yet, but excited to share when they do come out. So it's going to be a fun ride. The rest of the year is going to be exciting in the AI world. And if you want to make sure you're staying in the loop with it, maybe give this video a thumbs up, subscribe to this channel. I'll keep you in the loop with all the AI news. I'll keep you up to date at least once a week. I'm going to do my best to be helpful and help usher as many people into this new reality that I can. And if you want to stay on top of the news on a daily basis, check out Future Tools IO. I update the AI News page every single day with all the latest news. I update the homepage every single day with all the coolest tools that I'm coming across. And make sure to subscribe to the newsletter that'll send you all the latest AI news and all the coolest AI tools directly to your inbox. So once again, thank you so much for tuning in. Thanks again to Coda I.O. for sponsoring this video. Go to Coda I.O. matt for 25% off or your first three months free. Thanks again to them for sponsoring. And thank you once again so much for tuning into this channel. I really, really appreciate you. I really, really love that you tune into these videos and love nerding out with me because I'm having fun making them. Hopefully they're still enjoyable to watch and things are ramping up, so more exciting videos in the works. Thanks again. See you in the next video. Bye. By.