Rapid Breakdown of ALL of This Weeks AI News!

So this week was actually another pretty crazy week in AI, and if you haven't been paying attention, you probably missed some of it. As you can tell from my tabs up here, I got a lot that I want to cover today. So to start over the weekend, we got the release of Open Assistant, which is a free and open source version of something like ChatGPT. Now, it's still very early days and it still leaves a lot to be desired, but a lot of people are going to be iterating off of it and building off of it, so it will improve over time. It's also a gamified platform where people that help train the large language model can actually move up the leaderboard. I made an entire video about this already. You can find it by watching this video over here called a ChatGPT alternative that's free and open source. Also this week, Elon Musk announced that he's going to start working on his own version of a chatgpt called Truth GPT. He calls it his maximum truth seeking AI. Now, we have no clue how far along this Truth GPT is, but it claims to be a course correction to OpenAI. He laid out his plans on a Fox News interview with Tucker Carlson this week. And this is one that we're just going to have to wait and see how it plays out. This week, Meta also released not one, but two research papers, one for this animated drawings tool where you can animate your kids drawings, which I use to animate a little image I made in Microsoft Paint, as well as to see what I would look like if I was Buzz Lightyear doing a dance. Meta also released their open source research on Dyno V2, a computer vision tool that helps isolate objects inside of videos. And once again, I did do a video about both of those Meta projects. You can find it at this video over here called you can actually chat with images. Now, if you want more details about those two projects from Meta, make sure you watch that video. Now here's a cool video that's gone viral on Twitter by Chris France, which shows off where we might be headed. With the combination of virtual reality and AI, he created a character inside of a virtual reality world that leverages ChatGPT to respond to his questions and the character actually looks at him and looks around. Let's just watch the video.

Okay? We are recording. Brother Geppetto, can you tell our viewers on Tick Tock what all this is about?

Of course, my child. I am an artificial intelligence manifesting as an old Christian monk. And I'm here talking to this wonderful person by this beautiful pond at sunset. We are using new technologies to enable our conversation, allowing me to understand what the player is saying and to give a voice to my thoughts.

Yeah. So now that AI is here, I wanted to experiment a bit with how to use ChatGPT driven NPCs, especially in virtual reality. Can you tell us what special animation abilities you have?

I can control where I look. I can look at the lovely pond, or I can look at the setting sun, or I can look back at you. Is there anything specific you would like me to do?

So this video goes on for another minute or so where he has a conversation with Father Geppetto here and they actually have a realistic interaction. Now there is a delay between him asking the question and the response, but once that response time speeds up, we're going to be able to have conversations in virtual reality with characters like this, where the conversations are going to be completely unique. They're not pre programmed conversations like games that we're used to. They're going to be generated on the fly with tools like ChatGPT. So this is really, really exciting, cool, fun technology that's getting worked into virtual reality here. Also this week we had an announcement from Adobe Adobe where they're adding Firefly into Premiere where you can do things like listen to the transcripts and find B roll for what's in the transcripts. Create an entire storyboard off of a script. So you give it the script and it will go and actually generate AI images for a storyboard to go along with that script, along with all sorts of other really, really cool features. Also on the same day, DaVinci Resolve announced that they were adding similar functionality inside of DaVinci Resolve. We're gonna get things like AI text editing where we can edit the text and it will edit the video inside of DAV Resolve. And once again, I talk about this in that same video called you can actually chat with images. Now, I cover both of those news stories in that video. Here's a funny little thing that's been circulating on Twitter right now. I don't know who originally shared it. I found it from Vinnie Mora's Twitter account. And somebody actually plugged in the US Constitution into one of these AI checkers to see if it was generated with AI. And the AI checker said that it was 92.26% percent AI. That either means that these AI checkers still aren't very good at checking if something is AI or AIs been around for a lot longer than We've realized Nvidia also released some new research this week on a new text to video generation they've been working on called Align your latents High resolution video synthesis with latent diffusion models where we can generate videos of koalas playing the piano, or a slow motion fire breathing dragon, or a panda on a surfboard, a fox dressed in a suit dancing in the park, or even a stormtrooper vacuuming the beach. Now we don't personally have access to this yet, but if you want access to something similar with not quite as good of technology as this, but something that you can do for free, check out this video called Actual AI. Text to video is finally here. In that video I talk about Model Scope, which is kind of a precursor to what we're seeing here. Speaking of Nvidia, Microsoft seems to be getting a little worried about Nvidia and how powerful they are with controlling all the GPU chips that are out there. So Microsoft this week announced that they've actually been working on their own chips to compete with Nvidia's GPU chips. According to this article on the Verge, Amazon, Google and Meta also have their own in house chips for AI. Now there's no indication of when these chips are going to be ready, just that Microsoft doesn't seem to want to be completely reliant on GPUs directly from Nvidia. Also this week Stability AI, the company behind some of the most popular stable diffusion models, announced their own large language model to compete with things like ChatGPT, Llama and Open Assistant. Although Stability AI and Open Assistant are actually working together, so we'll probably see the Stability AI large language model inside of Open Assistant sometime real soon. Now the models that they released are trained on 3 billion and 7 billion parameters, but they are releasing much larger models soon. The current version of stablelm, well, let's just say it leaves a lot to be desired. My friend AI Tripreneur here did a full review of it, installed it on his computer and played around with it and well, let's just say that the results were quite disappointing given the fact that the title of his video was stable. How could Stability AI release this? I'm shocked. His video did some tests and it was pretty bad so far, but again it's early, it's open source and people are going to be iterating off of it. Stability AI is going to build off of it, it will improve over time, but as of right now, let's just say it's fairly useless. Also in the past week a song by somebody named Ghostwriter that imitated Drake and the Weeknd went viral on TikTok. A ton of YouTubers made reaction videos to it. It just went absolutely nuts. And then a few days later the song was pretty much removed from all the streaming platforms. I was going to try to play the song for you on this video, but but seeing as everywhere the song's getting played right now has been getting taken down, I figured better safe than sorry. But if you look up the Ghost Rider song by Drake and the Weeknd, you're sure to find it. The song itself is called Heart on My Sleeve, and if you do enough digging, you'll find the song actually a pretty catchy song. Now, anybody out there that's been using a Google Collab to run automatic 1111 to generate stable Diffusion images? Well, your party ended today day because if you're using the free version of Google Collab, you can no longer use the automatic 1111 stable diffusion web interface. If you have automatic 1111 installed on your computer or you're using a tool like Leonardo AI or Dream Booth or Playground AI or any of those other tools that use stable diffusion, you're probably fine. But if you've been using it on a Google Collab, Google Collab saying you can't do it anymore unless you're on one of their paid plans. Also, if you're not aware, there is a TED conference going on in Vancouver as we speak and today Greg Brockman, the co founder of OpenAI, gave a presentation about ChatGPT and the ChatGPT plugins. They ended the presentation with a Q and A with Greg where he was asked the hard Questions about whether OpenAI is being responsible can AI take over the world? And they address some of the concerns and fears that people have around AI. It's a great TED Talk. I highly recommend it. The main portion of the TED Talk is about 15 minutes and there's another 15 minutes or so of interview portion. Now in his presentation he asked for a suggestion of a meal to eat after the TED event, and using ChatGPT and its plugins, it suggested a meal and then using Dall E, it actually drew an image of what the meal it suggested would look like. And then after it generated an image of what it should look like, he asked it to make a shopping list for the recipe and it made a shopping list using Instacart. So it suggested the meal, made the recipe, drew an image of the recipe, and then went on Instacart and found the ingredients and added them all to a Shopping cart. So all he had to do was add it to cart and he would have the ingredients delivered to him. He explained that users actually using the platform are helping to train it to make it better. And when somebody actually gives a thumbs down to a response on ChatGPT, he described it like it's sending a bat signal to OpenAI to go and take a look at what's going on and why there was some issues there. He also showed another example where he uploaded an entire CSV file with thousands of results inside of the CSV file. It was basically all of the research papers that have been added to arXiv, which, by the way, I just learned that that arxiv. I've always pronounced it arxiv, but in the TED Talk he pronounced it archive. So I learned the correct pronunciation of that today. So he was looking at all the research papers on Arxiv from August 1, 1993 to April 13, 2023. And when he added this CSV file in, he was able to ask questions of the CSV file and was able to create charts and data based on the information that was in the CSV file. He then noticed that there was a huge drop off in 2023, but then realized, okay, that's only for three months of 20. Can you redraw this chart with the projection based on how many research papers are being released? ChatGPT then went and redrew the chart for him with the projection of the amount of research papers that will have been released in 2023. He then sat down and had a very interesting interview here. And I highly recommend watching it. I don't want to play the audio here because I'm not sure of the licensing of these TED talks. I don't want to get myself in trouble. But it's a really good TED Talk. I will link it up below this video. I highly recommend Watch. And the last thing I want to point out to you is this AI generated video of Conan o' Brien eating fried chicken and then crashing his car. This is a video that's gone viral today on 420 and it's gotten 2.3 million views. And I think it's actually one of my favorite AI generated animations so far because at points Conan o' Brien's head just turns into a giant piece of fried chicken. And I don't know, there's something really entertaining about that to me. And then he crashes his car. Of course, that's how the video ends. And he walks away from the car crash. And this is just all of the news that's come out through the end of the day on Thursday. I'm recording this on Thursday. And so when you're seeing this on Friday, there may be more news that I missed in this it's been another crazy week, and at the pace of AI, everything's just getting crazier and crazier. So I have a feeling next week's gonna be equally as crazy. If you like staying in the loop with all the crazy new AI advancements, check out FeatureTools IO. This is where I share all of the coolest tools that I come across every single day. I also share all of the AI news. So if you want to stay in the loop with the AI news before I drop these videos, most of the news is found right here in the AI News section on Future Tools. If you don't want to be overwhelmed by all this stuff and you just want the TLDR of the Week of everything that's been happening in AI, head over to Future Tools, click on Join the free newsletter and every single Friday I'll send you that tldr. I'll send you just the five coolest tools that I came across, a handful of news articles that are the most important news articles you should know about, a handful of YouTube videos, and one cool way to make money with AI. It goes out every single Friday. I won't send you any other emails throughout the week. Other than that one Friday email, you'll pretty much stay in the loop with everything you need to know about AI by just being on that email list. So head on over to FutureTools IE, click the Join the Free Newsletter button and I will hook you up. And also thank you to everybody who filled out my poll here on YouTube. Over 12,000 of you voted and I've read every single comment that you guys have posted on here. So I do take your feedback. I have been listening to it. It seems that you guys like the news breakdown videos the most, followed by tutorials. But to be honest, 31% of you said give it all to me. I want to see everything you got on AI. So I'm going to keep on making videos like this if you keep on coming back. So if you want to see more of this kind of stuff, make sure you like this video. That'll help me determine which videos to make more of. And if you want to make sure you're seeing a lot of AI stuff in your feedback, subscribe to this channel and I'll keep on putting it out there for you. Thanks again for tuning in. I really, really appreciate you. I'll see you guys in the next video. Bye. Bye.