Update: The AI News You Missed This Week!

Here's a breakdown of the AI news that you probably missed this week. This came out after I recorded last week's video. Google announced that they're starting to roll out AI. Inside of Google Sheets we can see in their example here they type agenda for a one day sales kickoff event with session description and status. The AI does its thing over here and then generates an entire plan for the conference. Here's another example from a preview they put up on YouTube. They type personalized messages for our clients. It looks at a spreadsheet of all their client information. They they just drag and drop down and then it writes personalized messages for all of their clients. Now they don't actually tell us when it's going to be available for everyone. It just says rolling out now in Google Workspace Labs. Also at the end of last week this article was released on the MIT Technology Review website. The people paid to train AI are outsourcing their work to AI. It says many companies pay gig workers on platforms like Mechanical Turk to complete tasks that are typically hard to automate, so such as solving captchas, labeling data and annotating text. This data is then fed into AI models to train them. The workers are poorly paid and are often expected to complete lots of tasks very quickly. They used AI detectors after getting some of the work back from some of these Mechanical Turk workers, and after running it through these AI detectors, they estimated somewhere between 33% and 46% of the workers had used AI models like ChatGPT. They say that using AI generated data to train AI could introduce further errors into already error prone models if they generate incorrect output that it itself is used to train other AI models. The errors can be absorbed by those models and amplified over time, making it more and more difficult to work out their origins. Essentially meaning that if these AIs are continually trained on outputs that were generated by AI, it sort of becomes this, for lack of a better word, incestual process, where potentially hallucinated baked facts are then further trained into the AI model, making things worse and worse and worse over time. Now the article does say I don't think everything will collapse, but I think the AI community will have to investigate closely which tasks are more prone to being automated and to work on ways to prevent this. Now here's something that Adobe's doing. If you use one of their Firefly Art generator platforms, Adobe says they will be offering enterprise customers an IP indemnity, which means that Adobe would protect customers from third party IP claims about Firefly generated outputs. So right now There's a lot of fear going around that, that if I generate images with AI and I use them in my business, well, the data that it was trained on may come from unknown origins that the trainer of the data may not have had the rights to. And is that gonna cause legal issues down the road for the companies that use the images generated with these AI platforms? Now, Adobe is telling their enterprise customers, you don't have to worry about that. If you're using our Firefly products and somebody sues you and you lose will cover the cost for you. They're that confident that they own the rights to the data that was trained into their AI model. Adobe feels comfortable taking the stance because it has trained Firefly on Adobe stock images, which it has broad permission to use, along with openly licensed content and public domain content where the copyright has expired. So all of this is basically saying that you should be able to trust Adobe on how their models were trained. And any images that you use that were generated with one of their Firefly products, like Photoshop, Generative Fill or directly from the Firefly platform, you should be barely safe to use those images without any future legal consequences. And if there are legal consequences, Adobe's got your back on it. Now to jump forward in the news cycle a little bit. ChatGPT, on the other hand, is facing a lawsuit right now over how it used people's data. A law firm is launching a class action lawsuit against OpenAI, alleging the artificial intelligence company that created the popular chat bot, ChatGPT, massively violated the copyrights and privacy of countless people when it used data scraped from the Internet to train its tech. Basically, the lawsuit is saying that when people posted things on Twitter or Facebook or blog posts or wherever they posted them on the Internet, they were never posting them with the intention that that information would be used to train AIs in the future. And therefore, everybody whose data it was trained on should essentially be compensated for the use of that data for profit. Now, OpenAI's defense on this is to be that that data was available to the public for anybody to use. And because of fair use policies, OpenAI used that data in a transformative enough way that it really had no negative effects to the people that originally created the content. It's going to be interesting to see how this one plays out in court, because, like, with a lot of AI we've seen lately, there's going to be a lot of legal battles and a lot of discussion around how the ethics of all of this should be handled in the future. Earlier this week, Google Deepmin CEO says its next algorithm will eclipse ChatGPT. Now, I read this entire article on Wired and there's nothing really in here that substantiates that claim. They basically are saying that they're going to use some novel methods that combine what we're seeing with tools like ChatGPT as well as using techniques from AlphaGo to make their AI systems. Now, AlphaGo was a really powerful chatbot that learned how to play a very popular game around the world called Go and actually managed to beat a Grand Master, which nobody thought AI would ever manage to do. So it's using technology from that AlphaGo, plus the same type of large language model we're used to with something like Chat GPT to create what they call Gemini. Now, other than that information that they're just kind of combining some technologies to make something bigger and better, they don't really tell us how it's going to be better or what we can expect to be better about it, just that, hey, it's going to eclipse ChatGPT and just wait and see. That's kind of what this article says. In a really fun announcement from Unity this week, which, if you're not familiar with Unity, it's a popular game engine which developers use to actually build a lot of the most popular video games. They announced Unity Muse in their example video. They actually managed to question the documentation for Unity and learn how to do various things inside of Unity. While it searched the documentation, they also demoed creating character animations directly from text. They type, do a backflip and the character does a backflip. They also demonstrated the ability to quickly create 2D sprite assets. They took this image that looks like it could be out of like an Age of Empires style game and told it to change the color of the roof. They were able to generate realistic textures instantly. Beach, sand, debris. It generated this texture and overlaid the texture on a map. This Unity Muse, they say, is now enclosed beta. One cool thing about Unity as well as Unreal Engine is they're both actually free platforms to use. So if you've ever been interested in game development, you can use tools like Unity and Unreal Engine and develop games for free. And I imagine all of these new AI features that they're adding into Unity are going to be available within the free platform as well, which is going to make it even easier for anybody to create games. Now, when I say it'll be easier for anybody to create games, I'm not saying it'll be easy for anybody to create Good games. I still think there's a lot of storytelling and thought that goes into good games, but anybody could create fairly basic games. They also announced Unity Sentis is now in closed beta as well. If you remember my previous conversation with Pernindu from Convey, this actually uses the same technology to give characters unique dialogue within games, but it's built directly into Unity.

What hairstyles would you choose if you had hair?

If I had hair, I would probably go for a messy bun or ponytail. I like how they look on other people and I think they're easy to maintain.

What's the secret to life?

The secret to life is to find something you love to do and do it every day.

You can see right here below the video on their announcement, they actually say you can now find professional quality AI solutions from providers such as Convey in World Layer, Leonardo and more. And Convey is actually the company that I demoed with their CEO in a previous video. Speaking of game assets, here's another cool tool that I came across that is also enclosed beta, but I'm excited to play around with it. It's called csm, which is short for Common Sense Machines. And this is a tool that can take 2D images and turn them into 3D assets. Theoretically, we'll be able to take images that we generate in tools like Midjourney or Leonardo, pull them into this tool, and turn them into a 3D asset that we can use inside of a game, inside of a movie, inside of a virtual reality or augmented reality world, or whatever we want with them. In fact, here's some examples that they show on their YouTube channel of some of the assets that were generated. Here's the 2D input, here's the 3D output that it generated. Here's the 2D input of Iron Man. Here's the 3D output that it generated as a character. Here's some other examples. You can see where it compares the input on the top and then what you get from their output, their Common Sense machines output compared to what's existing right now with the OpenAI shape E, which, you know, obviously still leaves quite a bit to be desired and what that's capable of. I can't wait to get my hands on something like this and show off what it's capable of. Also this week, the tool Wonder Dynamics, the tool that I use to make this crazy robot shooting back basketball, came out of closed beta, where just a small amount of testers were able to use it. And now it's in open beta where anybody can get access, albeit for a little bit of a cost. Now in other news this week, OpenAI's Chat GPT app can now search the web, but only via Bing. Now, I don't know why they added but only via Bing, because if you've used Chat GPT anyway, it always says Chat GPT with Bing. So even the desktop version is searching the web via Bing. But basically, if you remember a couple weeks ago we talked about how you can now use ChatGPT directly on your iPhone. You now have that web browsing feature available inside of ChatGPT+ on your iPhone. And speaking of ChatGPT, in this article from Gizmodo, they ask, is it starting to get a little too crazy with what we're putting ChatGPT into? They talk about this E bike which now has built in ChatGPT, and there's probably very few scenarios where you're actually going to need ChatGPT built into your bike. I mean, if you're out for a bike ride and you happen to have your mobile phone on you and maybe like Bluetooth headphones in your ears, don't you get the exact same thing? Now, there's not a lot that I agree with from Gizmodo these days. However, this is one where I do feel like, all right, this bike company is just throwing AI in, thinking that if I sprinkle some AI on this, then it'll make more sales of my bike. This one's an odd one to me. Also in semi AI news this week, hundreds of actors are ready to strike if SAG Aftra doesn't secure a truly transformative deal. A ton of actors in Hollywood are trying to renegotiate the minimum pay that actors earn on jobs. And as part of this potential strike that might happen, they're also concerned about AI, it says here. The letter also called for more control of the degree to which studios are able to make actors record themselves during the casting process, and pointedly called out the industry's adoption of AI tools as something SAG Aftra should be thinking about very seriously during this current round of negotiations. We think it is absolutely vital that this negotiation protects not just our likeness, but make sure we are well compensated when any of our work is used to train AI. Now, in last week's video, I referenced the Black Mirror episode Joan is Awful, which again, not going to spoil it for anybody, but that kind of touches on this issue and what that could look like for celebrities if this all goes horribly wrong for them. But it does make me wonder if a ton of actors refuse to work because they're afraid of AI taking their job. Couldn't that potentially cause AI to take their jobs even sooner? Suhail from Playground just announced a new feature called Mixed Image Editing. This seems to me to be very similar to what we get from Adobe's Generative Fill. Where we can see in their example, they generate an image of a chocolate cake with fruit. Then they sort of mask out a certain area and erase the fruit. They mask out a spoon and say make it a fork and it replaces that spoon with a flat fork. They take the cake and they add birthday candles. And then they mask out the top of the candles and say light the candles and it adds flames to the candles. Also, the big news of last week was Mid Journey Zoom out feature stability. AI also recently released Uncrop. Well, it now looks like you can do that same kind of thing in Playground where you can just stretch out an image, click Generate, and it will basically uncrop, Zoom out, whatever you want to call it to that image. And what's really cool about Playground is that you can actually still generate images for free on Playground. Playground right now, if you want. So let's write a prompt here. A wolf in a snowy forest. Click Generate. I get something that looks like this. I can click the button that says Edit in Canvas. Now if I come up to the generate image up here, expand it outside the box a bit like so, and just click Generate. Just like Uncrop or Generative Fill or Zoom out, it will fill in the missing gaps here. I could select this little pen up here, make this a little bit larger. If I select this tree here and say make the tree green and click Generate. I now have a green tree that's called Mixed Image Editing and it's available now inside of playgroundai.com and I don't actually pay for Playground. This is actually free to use right now. If you're a fan of Midjourney, Midjourney just announced a new command that you can try called weird. So check this out. If I enter a prompt like a woman in a colorful dream like world, I get a pretty cool image that looks something like this. If I do the exact same prompt but add dash dash weird 2000, it goes between 0 and 3000, I get something like this. So just a fun new command that you can try out. And if you're looking for something really wacky and crazy, try ramping up the weird. Also, this week, Microsoft announced that they're launching their new AI skills training and resources as part of their Skill for Jobs initiative. Now, this is online coursework. It's also going to give the first professional certificate of generative AI, so you can actually go through their course and tell people that you are certified in generative AI. The coursework will be unlocked and available for free through 2020. This week, the Verge also released some research. They pulled 2,000 people on how they're using AI, and this was actually very, very interesting. Although 2,000 people isn't a huge data set to really be pulling from. It is interesting because you can see on the left side, these are people that have used or heard of it, and these are people who have not. And if we hover over here, 57% of people in the U.S. according to this poll, have actually used or heard of ChatGPT. 43% haven't Bing Chat. Only 46% of people have heard of it. My AI from Snapchat. Only 45% of people have heard of it. In my opinion, that's probably really high because even among AI circles that I run in, I don't think a lot of people have heard of Snapchat's My AI. Only 38% of people have heard of Google Bard. Only 25% have heard of Midjourney. I would have thought for sure more people would have heard of Midjourney than bard or my AI. And then only 23% have heard of stable diffusion, which that tracks for me. Millennials seem to use it the most, followed by Gen Z, followed by Gen X, followed by Boomers. That tracks right along with the data that I see in my YouTube analytics of who watches my videos. It's probably mostly Millennials and Gen Z, so that seems about right to me. Also in their poll they asked which of these technologies will have the biggest impact on society. 74% of the people said that AI will have a big impact on society. 69% electric, 60% virtual reality, 52% augmented reality, and only 34% NFTs. If I had to guess that VR and AR over time are gonna flip, I think more people are gonna see AR as the future more so than VR. I just think VR is a little more in the movies with Ready Player one and stuff like that. But so many people are already using augmented reality without even really thinking about the fact that they're using augmented reality. For example, Snapchat and Instagram filters are all using augmented reality. I'll up to this article below. There's some other interesting information they came up with. Like what Most people thought AI did a better job than they did at what people's biggest concerns around AI have been including 70% of people saying that they believe artists should be compensated when AI copies their work and how people feel about regulation for AI. They also found of the 2,000 people they polled, 51% of them believe that AI will become sentient at some point. And 64% of those people are not opposed to companies producing an AI that is conscious or sentient, which I think with a wider data set, I honestly believe that number would probably shrink. But it is what it is. I'll link up the article below if you're interested in learning more about what they found. And finally, I wanna end this one with some comments from Mr. Beast, one of my personal favorite YouTubers about his thoughts on AI from this amazing interview that was on the Colin and Samir show. Again, another one of my favorite YouTube channels. If you're a YouTuber or creator yourself, Colin and Samir, they're the best. You gotta check out their channel. Here's Mr. Beast thought thoughts on where AI is going.

How do you think about the difference between human intelligence and artificial intelligence? I don't think it'll destroy things. I think it would just free up people to do more of what they love and like focus on creating. And they'll have that for inspiration. And like there'll probably be a period where everyone just plagiarizes and uses that and like it's gonna get more homogenous at one point and then it'll normalize and then the people who break through are gonna be the ones that use that as inspiration and level up on it. And you know, it just kind of is what it is.

Again, as I've said in so many videos in the past, that is the way I think people should be thinking about AI. I don't think it's going to replace people. It's going to give people that use it superpowers and free up their mind space and their time to focus on other things that could benefit them and their growth even further. And So I love Mr. B's thoughts on this and curious what you think about it, but that's what I got for you. That's a breakdown of the AI news this week and I'm recording this on Thursday. So there might even be more news that's dropping on Friday that didn't make the video. But if there is, I'll make sure it's in next Friday's video. If you haven't already subscribed to this channel, maybe now's a good time to do it. To make sure you see the next AI News video that I put out next week. And if you do like this video, maybe give it a thumbs up that'll make sure you see more videos like it in your YouTube feed. Thanks so much for tuning in. Check out Future Tools IO. This is where I will keep you up to date on the news in near real time as well as tell you about all the coolest AI tools that I came across. And if you don't want to keep up with it on a daily basis, join the free newsletter and every Friday I will send you a TLDR of the Week in AI. The high level overview of just what you need to know and just the coolest tools that came out goes out every Friday. You can find it over at Future Tools IO. Thank you so much for tuning into this video. I really, really appreciate you. There is so much going on in the world of AI and I am just absolutely in love with it and I'm going to do my best to keep on keeping you up to date and making tutorials and finding new and exciting ways to make fun content around AI and tech that I think you'll enjoy. So you watching this video is helping me get closer to that dream of just getting to live in this world all of the time and talk about AI and tech all of the time. So honestly, when I say I appreciate you, I really mean it. Thank you so much for tuning in. I'll see you guys in the next video. Bye Bye.