OpenAI’s Roadmap Revealed – Major Changes Coming!

Well, it's been another intense week in the world of AI with lots of drama and a bunch of cool product and feature updates, and it's my job to sift through all the noise and try to help you uncover just the important stuff that might actually impact you. So without wasting any more time, let's get into this week's AI news breakdown, starting with what's probably the most talked about news story of the week in that Elon Musk attempted to purchase OpenAI for $97.4 billion. A team of investors led by Elon Musk submitted the 97 point billion bid to purchase OpenAI on Monday. Now, if you're not already aware, Elon Musk is already suing OpenAI because they've been straying away from the mission in which he originally invested. Elon said it's time for OpenAI to return to the open source safety focused force for good it once was. We will make sure that happens. Sam Ohman's already replied to the bid on X by saying no thank you, but we will buy Twitter for 9.74 billion if you want. Which was a little jab at Elon because, well, Elon initially bought Twitter for $44 billion. Elon came back calling him a swindler and I really like Kevin Rose's take here where he says, I'm hoping you both go on a spiritual journey and heal all the things. You'd be unstoppable together. I agree. I would love to see a Sam Altman Elon Musk combo project because I do think whatever version of AI they combine on would just be insane. But I have a hard time actually seeing that play out. A lot of people are speculating that Elon Musk doesn't actually want to buy OPEN AI. He just wants to sort of muddy the and make it more complicated for OpenAI to transition to that for profit that they've been trying to do. The thinking behind it being that if OpenAI was to transition to a for profit company, it essentially needs to buy out the nonprofit and the valuation that the nonprofit was going to be bought for is much lower than what Elon offered, therefore requiring the nonprofit to sort of take his bid seriously. But because it's a nonprofit, they don't necessarily have a fiduciary responsibility to shareholders. So it's real sort of muddy and confusing. But here's what we do know about the offer from Elon. There's a deadline of May 10, 2025. The deal is an all cash deal, 100% of the purchase price would be paid in cash. Elon and the buyers would want full access to the books and personnel to do their due diligence, which is another potential reason Elon is doing this. Just to sort of uncover information and then say, based on our due diligence, we're not going to purchase it. According to the letter of intent, they need to see assets, facilities, equ books and records. While this is a normal part of due diligence, this could give Xai, which is an OpenAI competitor, access to sensitive internal information. And once they've seen it, all their diligence could also provide them with a reason to withdraw their offer. And Musk claims that he'll withdraw his offer if OpenAI decides to stay a non profit. In an interview by the Pulse here, Sam Altman was asked basically what he thought about the Elon Musk offer. Here's what he had to say about it.

Elon Musk making a bid, apparently for Open Air.

Your response?

You're turning him down for 97 billion? I mean, look, OpenAI is not for sale. The Open Air mission is not for sale. Elon tries all sorts of things for a long time. This is the late, you know, this week's episode.

You take it seriously at all?

What do you think he's trying to drive out with this? I think he's probably just trying to slow us down. He obviously is a competitor. You know, he's working hard and raised a lot of money for X and they're trying to compete with us from a technological perspective from that, you know, getting the product into the market. I wish he would just compete by building a better product, but I think there's been a lot of tactics, you know, many, many lawsuits, all sorts of other crazy stuff. Now this, we'll try to just put our head down and keep working. Does it make it more difficult to move from that nonprofit model to that for profit model for OpenAI? We're not moving to a for profit model. We're. I mean, we're not sure we're going to do it all. But no matter what, the nonprofit will continue to be extremely important. It will drive the mission. It will continue to exist. The board is looking at lots of options about how to best structure for this next phase, but the nonprofit is not changing it anything or going anywhere. Okay.

So essentially he doesn't seem to be taking Elon too seriously. He sees it as Elon just trying to slow things down, trying to compete a little bit dirty instead of competing by creating the best product. Now moving away from The Elon Musk Sam Altman drama OpenAI actually did make some announcements this week. They shared their roadmap of what they're planning on doing. Plus they actually rolled out some useful new features. So let's start with the roadmap here. One of the things Sam Altman is commenting on is he says, we want AI to just work for you. We realize how complicated our model and product offerings have gotten. He's referring to the fact that when you use chat GPT, especially if you're on the pro plan, you've got 1, 2, 3, 4, 5, 6, 7, 8 model choices. Plus with these choices you've got additional options within the model to turn on things like search and deep research and image and canvas and it's just too complicated for people. He says we hate the model picker as much as you do and want to return to magic unified intelligence. He says a top goal for us is to unify O series models and GPT series models by creating systems that can use all our tools, know when to think for a long time or not, and generally be useful for a very wide range of tasks. So what he's saying is you shouldn't have to pick GPT4O or ChatGPT03 or ChatGPT01 with Pro Mode or Reasoning model or Deep Research. You shouldn't have to pick all these things. You should just be able to enter your prompt and then the chatbot that you're using when you're using Chat GPT, it decides which model to use based on the question. Okay, this one requires some more deep thinking. Let's go ahead and use one of the OH1 models. Okay, they're clearly asking for some deep research on this. Let's go ahead and turn on the Deep Research feature. Oh, this is a really quick, easy one to answer. Let's go ahead and use our GPT4O mini model. That is the lowest compute model. The ChatGPT tool should pick for you which model to use. In the same post here, he also talked about how the models are going to be released in the future. He did say we have GPT 4.5. That's the model they called Orion internally and that's their last non chain of thought model. So that's the last model that are going to release that doesn't actually show it sort of thinking through things. All of the rest of their models after that are going to be built around test time, compute or when you enter a prompt. It actually does a lot of processing at that point to think through the response and give you an answer. In both Chat GPT and our API, we will release GPT5, a system that integrates a lot of our technology, including O3. We will no longer ship O3 as a standalone model. So up until, you know, last week, the road map was they shipped O3 mini, O3 mini high and they were going to ship O3 the non mini model, but now it looks like they're not going to do that. It's just going to be like part of GPT5. He also says the free tier of Chat GPT will get unlimited chat access to GPT5 at the standard intelligence setting. Plus subscribers will be able to run GPT5 at higher level of intelligence and Pro subscribers will be able to run GPT5 at even higher levels of intelligence. And then Chubby here says any ETA for GPT 4, 5, GPT 5 weeks months and Sam basically said weeks slash months, which is pretty much a non answer. Weeks slash months could be any time within the next year really, but theoretically we'll have GPT4.5 fairly soon. I would imagine within the next several weeks. That's going to be the replacement to the GPT4O that you're using now that doesn't have the sort of chain of thought thinking and then when GPT GT5 comes out, from then on out, I think every model is just going to have test time compute. It's going to have that thinking ability at the time you enter the prompt if it needs to do it, because it's going to start to learn. Should I really think this one through or is this an easy one for me to answer? OpenAI did also release a few other small updates this week. If you're a Pro user and you want to use Deep Research on mobile, it's now available on all of the platforms. So you can use it on iOS, Android and even on the desktop platforms Mac OS and Windows. And they also announced that OpenAI 01 and 03 mini now support both file and image uploads in chat GPT. Before you couldn't actually add things like PDFs and ask questions about the PDF with 01 or O3 mini. Apparently you still can't with 01 Pro. I tried it earlier today, it didn't work. They also raised the O3 mini high limits by 7 times for plus users to up to 50 per day. So if you're on the plus plan, the $20 a month plan, you now have 7 times more usage limit on O3 mini high. And finally in the last bit of OpenAI news that I'm going to share. I know there's been a ton this week. Sam Altman and Kevin Weil spoke at the University of Tokyo center for Global Education. And most of what they talked about is the same kind of talking points we always hear Sam Altman talk about. But I thought the most interesting part of this was when he was asked about open source earlier.

You mentioned making intelligence accessible to everybody. And with the recent release of China's Deep Seq AI, along with the open publication of their weights and architecture, the validity of OpenAI's shift away from open source has been called into question. Are there any plans to reconsider that stance and what would be the justification? Yeah, we're going to do it. I don't know exactly what we're going to open source or exactly when. It does seem to us like the direction of travel is towards more open models available and that society is willing to take the trade offs, at least for now, that come with that. And you know, I think we've made good progress on getting these models to be generally safe and robust in open source uses. They want all be used that way, but they will be most of the time. Yeah, I'd expect us to move in that direction.

So very interesting because I started this video today talking about how Elon Musk wants to buy OpenAI so that he can take it open source and get back to the true original goal of what OpenAI was. And now Sam Altman is going around saying we need to do more open source stuff. He said it in an AMA last week and this week he's saying it at the University of Tokyo. So it does sound like they are going to make some sort of pivot into open source. Whether they're going to open source their biggest models, I find it kind of unlikely, but I can almost see them open sourcing like two generations ago models. I don't know. We'll see. You know that feeling when you're working on a creative project and you can picture exactly what you want in your head, but bringing that vision to life feels like trying to teach a cat to swim. Maybe you need some B roll footage for your video, or you're trying to create the perfect background for your thumbnail, or you're working on motion graphics but just can't get them quite right. And let's be real, sometimes you end up spending hours searching through stock footage sites or trying to create something from scratch, only to end up with something that's, well, not quite what you imagined. And that's why for this video, I partnered with Adobe to tell you about what's new in Firefly like Generate Video. Powered by the new Firefly video model, this is seriously game changing stuff for creators like us. Adobe Firefly is the first generative AI for video that's actually designed to be commercially safe, meaning everything it creates is trained on licensed content and public domain work. No more wondering if that AI generated asset might get you into trouble down the line. Now let me show you what makes this so exciting. The Generate Video modules let you create completely original video clips just by describing what you want. Need a cinematic drone shot of a mountain landscape? Done. Want to add some cool atmospheric effects to your existing footage? You got it. The best part is you have full control over the camera angles, motion and visual style. It's like having a professional VFX team right in your browser. But it's not just about video. Adobe Firefly is basically your Creative Swiss Army Knife. You can generate custom images, create animations, design backgrounds, and even translate your content into over 20 languages automatically. Everything on the web app integrates seamlessly with the Adobe Creative Cloud apps that you already use, so you can take your Firefly generated content straight into Premiere Pro, After Effects or Photoshop for final touches. And what I particularly love about Adobe Firefly is how it respects creators. This isn't about replacing human creativity, it's about enhancing it. It's about taking those tedious, time consuming tasks off your plate so you can focus on the creative decisions that really matter. And unlike other AI tools, you know you can actually use what you create commercially without worrying about any of the licensing issues. Right now you can try Adobe Firefly for yourself by heading over to firefly.adobe.com trust me, once you start playing with these tools, you'll wonder how you ever manage without them. Whether you're a professional creator or just getting started, Adobe Firefly gives you the power to bring your creative vision to life in ways that just weren't possible before. So check out firefly.adobe.com to learn more and start creating with Adobe Firefly now. We're going to dive in and talk even more about Adobe Firefly in just a couple minutes here, but there's a few other things that I want to talk to you about first in the world of large language models, starting with Anthropic here, we've been getting a lot of news lately. Out of OpenAI, out of Google, out of Mistral, out of Deep Seek, out of Alibaba. All of these companies are making big splashes and Anthropic's been kind of quiet. Well, Dario, the CEO of Anthropic, has been speaking up more and more lately. Apparently they're taking a different approach. They've developed a hybrid AI model that includes reasoning capabilities, but the model can also handle simpler tasks quickly without the extra work by acting like a traditional large language model. So basically they've been working on what I just mentioned Sam Altman wants to do next, a model where you give it a prompt and it decides whether it should use deeper reasoning with the model or to just give you back a quick answer. Anthropic's new hybrid model gives customers the ability to control how much computational resources it spends on a query, in other words, words, how long it will reason about that problem now one thing that really excites me about this new Anthropic model is specifically for coding. Because to this day, most people that I know that use AI for coding are still using Claude 3.5 Sonnet. And when I mess around with code and AI myself, I'm still using Claude 3.5 SonNet, even more so than things like 01 Pro, mostly because O1 Pro doesn't have API yet, so it doesn't connect up with tools like Cursor and Windsurf and some of the other tools that I use. But CLAUDE still seems to be the best fit for those tools, and we can see here Anthropic's upcoming model has also made especially strong strides in coding, which has emerged as one of the strongest applications of generative AI in business. While OpenAI's reasoning models are better at more academic problems like competitive programming questions, Anthropic's model is more adept at the sort of coding tasks engineers working in business might do. We don't know exactly when this model is coming out from Anthropic, but they did say in this article the company plans to release it in the coming weeks. So maybe sometime February or March, we'll see. But I'm really excited about this one. Specifically for the coding implications, we got some updates out of Perplexity this week as well. Perplexity now offers file and image uploads with an expanded context window of a million tokens, and it's free for all signed in users. You just got to make sure you're set on the auto mode in order to use it. Perplexity also rolled out a new large language model that they're using, which is actually based on Llama's 3.3.70B model, and it uses the new Cerebras inference infrastructure, which means it loads really, really fast. It actually loads at 1200 tokens per second. We can see in this little comparison video here that Gemini Flash runs at about 140 tokens per second and Claude 3.5 sonnet is about 75 tokens per second. So 1200 tokens per second is really freaking fast. And some other interesting kind of weird news about Perplexity. They actually put out a blog post on February 13, the day that I'm recording this video called Introducing Perplexity Deep Research. You can even see it's still indexed on Google right here. And in the little description it says, today we're launching Deep Research to save you hours of time by conducting in depth research and analysis on your behalf. However, if you click into the article, it gives us a 404 page like they took the whole thing down for some reason. So I don't know if they prematurely posted about Perplexity Deep Research or if they're gonna rebrand it as something else. Since Both Google and OpenAI use the branding of Deep Research. Here's one of the screenshots that shows the simple QA benchmark which basically asks it a whole bunch of questions and figures out how factual and accurate a model is essentially. And you can see that it scored like 95% this perplexity deep Research, where the next best model was OpenAI01, scoring under 50%. Now this image here also doesn't include, you know, 01 Pro 03, DeepSeek R1. A whole bunch of the new models that people are talking about right now isn't included on this, but this was the screenshot they were showing on their original blog post. Here's another one showing how fast Deep Research was. This was a screenshot from that same blog showing that it took about three minutes to do some research, used 25 steps, used 42 different sources. And then I believe the output that it gave back was 1,321 words. So it was a pretty like in depth write up. And then finally there was also this screenshot that showed Perplexity side by side with Perplexity's Deep Research, showing that a typical query on Perplexity looks like this on the left here. And one that used Deep Research just shows off a lot more information. If I go to my Perplexity account, there is no Deep Research option. So my guess is they might have released the blog post before the feature was rolled out. But it looks like Perplexity is about to have its own Deep Research feature. If you're a Gemini advanced user, they just rolled out the ability to have memory from past chats. So when you talk to Gemini, it can actually look in previous conversations that you've had with Gemini and use information from those previous conversations in response to your new conversation. And it does say you are in control over what information is stored. You can easily review, delete or decide how long to keep your chat history. You can also turn off Gemini Apps activity altogether, so if you don't want it to remember stuff from past conversations, you don't have to leave that turned on if you don't want A few weeks ago we talked about a cool new feature inside of Google Labs called Whisk. It's a feature where you can blend multiple images together, multiple styles together, and it will give you an output based on all of the inputs that you gave it. And well, this week it's gone global, so now it's available in over a hundred different countries. YouTube teased out some new AI features that they're working on. The CEO Neil Mohan wrote up a blog post about his four big bets for YouTube in 2025 and number four on his big bets is AI will make it easier to create and enhance the YouTube experience for everyone. They say that they're investing in tools to help creators in everyday work of creation things like coming up with new video IDE titles and thumbnails. They're also using AI to help creators find new audiences. Basically they're going to do that by rolling out dubbing so that if a video is recorded in English, I can put an automatically AI generated Spanish version up and an automatically AI generated German version up and that'll make it easier for these videos to be consumed in those countries, helping the videos find more audiences. And since we're talking about YouTube, they're actually rolling out their VO2 model inside of YouTube shorts. The article here says says today Dream Screen is getting another major upgrade with the integration of Google DeepMind's newest video generation model, VO2. And we're also now launching a new capability powered by V2, generating standalone video clips that you can add to any of your shorts. Here's an example short they shared of what that could look like. Yeah, for sure.

No, no, no.

I'm just getting old now. Yeah, yeah. No, that's crazy. No, no, absolutely not. Yeah.

Yeah.

Oh, I have to take my dog. You want a treat? Good girl, good girl, good girl. So I haven't played with that yet, but apparently you can do that straight from within the YouTube app. While creating Shorts now. And as I mentioned earlier, Adobe released Firefly video model this week and it's actually quite impressive with a lot, lot of control that you can do. Inside of the video we can see some examples of like aerial footage of a majestic snow capped mountain reflected in a calm lake. A flickering flame on a black background only camera is steady and there are no objects in the video other than the flame. Cloudy mist on a green screen, background only camera is steady and there are no objects in the video other than the mist. I think it's really cool that you can create really good green screen and like black background video with this. We've got a lion here, we've got like a giant whale. Really, really cool videos. In fact, let's jump in and play with it for a second. If we go to firefly adobe.com we can see we have an option for video at the top and we've got text to video, image to video, translate video and then coming soon, speech enhancer and text to avatar. But let's go ahead and do text to video here. We can play with the aspect ratio, widescreen or portrait shot size. We can do extreme close up, close up shot, medium shot, long shot, extreme long shot. Camera angle. We've got aerial shot, eye level, high ang glow angle, top down. We've got options for the motion like zoom in, zoom out, move left, move right, tilt up, tilt down. Static handheld. Let's do our typical wolf howling at the moon and then I'll do a high angle shot for. No, let's do a low angle shot on this one, maybe with a slight zoom out and let's generate it. All right, and here's what we get with our wolf howling at the moon. Low angle shot, zoom out. We see the moon and it zooms out with past the wolf, right under the wolf. I actually think it's a really cool shot because it kind of comes in under the wolf's chin a little bit. Definitely an angle you don't really see see often when you want to get a wolf howling at the moon. Never really get the camera coming from underneath like that. Now let's try something totally different. Let's try drone footage flying over an African desert landscape with elephants. I'm going to switch this to an aerial shot and then I'm just going to remove the motion here and let's generate that. And that first generation took maybe about a minute and a half to two minutes. Somewhere between 90 seconds and two minutes. And here's what we get from that. Definitely looks like a drone Flying over an African desert. Desert. Even the camera movements, how the camera's sort of like jerky at the end there really reminds me of like a real sort of drone shot. See how it kind of like pivots a little bit, like they're adjusting the camera. All right, I'm actually going to try to modify one of the prompts that they showed off in their page. Right here. I'm going to try a truck on a windy road that is driving towards the sunset made of clay. Claymation. On the horizon you can see a rocket taking off, leaving a trail of rainbow smoke. Let's see if we can do it. All right, here's what we got from that. We see the truck driving down the road. Claymation. There's our rocket. It with the rainbow smoke. I mean, it kind of nailed it. Let's test another one here where we give it a ton of detail. They have this prompt where they're doing a lion, but I want to change it to a different animal and see how it comes out. I'm going to change it from a lion to a wolf. So a majestic male wolf yawns while surveying his domain from a rocky outcrop at sunrise. Golden light illuminating his. I mean, I guess main kind of works there as well. I'll just leave it shot with a Nikon D850, 600 millimeter, etc. Etc. Let's go ahead and leave all the rest of the prompt that it had before and see how it does when we swap it to a wolf. And here's what we got out of that one. We got a wolf. And then it looks at the camera the way it gets up and turns. I don't know, something a little funky there. But this is the worst is going to get. And for our last little test here, let's try the image to video module inside of Adobe Firefly here. I want to try giving it this image of like a firefighter spraying water on a fire. Although this is an AI generated image, you can tell because it looks like the guy's wearing a bucket on his head. But I want to see how it performs with this image here and see if it can animate the fire and the water and that sort of stuff. So we'll go ahead and add a first frame here. We'll pull in this image a man spraying a water hose on a fire. Let's just see what it does. And here's what we get from that. I mean, actually fairly realistic looking water, especially right in the beginning here. And the fire, I mean, kind of looks like a explosion from a volcano or something, but it's still a pretty cool looking visual. And one thing that I do really appreciate about Adobe Firefly is that it is designed to be commercially safe. So when they trained it, they trained it only on videos that they had licenses to and public domain video. So that if you're someone that is concerned with the ethics of how some of these video models were trained or maybe some of the copyright implications of how some of these video models models were trained, well, Adobe Firefly is probably the one you're going to want to use because you know the origin of the training material. So thank you so much again to Adobe for sponsoring this video and I'm going to move along to some other video news. This week Luma rolled out their new Ray 2 image to video model. They already rolled out Ray 2 text to video, where you can enter a prompt and get a video. Now you can start with an image and turn that image into a video as well. And from the examples I've seen, it looks pretty impressive. Impressive. I haven't had a ton of time to play with this one yet, but I'm sure we'll explore it in a future video. If you're somebody that likes to use Bolt new to develop apps, well now you can develop native mobile apps. They say you can now go from idea to app store just by prompting Bolt. No coding necessary. Definitely something I want to play with in the future. I've built various AI tools using AI before, but they've all been desktop or web based tools. I haven't tried to build any mobile app, so this should be a fun rabbit hole to go down in the future as well. While we're on the topic of AI coding, Windsurf this week released Windsurf Wave 3, their newest version of Windsurf. It now leverages model context protocol so you can bring in more tools and it can find context from more things around the web. They added Tab to Jump, a turbo mode to run terminal commands for you, which is really, really helpful. Lately I've mostly been using cursor and one of the reasons I really like Cursor was that it would just run terminal commands for me. Well, now it looks like you could and Windsurf as well. Drag and drop images. This is another really cool one because if you're getting an error on whatever you're trying to build, you can literally just screenshot the screen and drag and drop the screenshot in and say this is what I'm running into. Help me solve it so that makes things a lot easier as well. So some pretty cool new quality of life updates to wind surf. This week was also the AI Action Summit out in France and the US Vice President JD Vance was there to talk about how basically the US was going to stay dominant in AI and that he urged other countries to not be so restrictive with AI because it's really going to slow down progress. I'm not going to play the whole thing, but here's a couple little sound.

Bites I'd like to make four main points today. Number one, this administration will ensure that American AI technology continues to be the gold standard worldwide and we are the partner of choice for others, foreign countries and certainly businesses as they expand their own use of AI. Number number two, we believe that excessive regulation of the AI sector could kill a transformative industry just as it's taking off and will make every effort to encourage progress, growth, AI policies. And I like to see that deregulatory flavor making its way into a lot of the conversations this conference. Number three, we feel very strongly that AI must remain free from ideological bias and that American AI will not be co opted into a tool for authoritarian censorship. And finally, number four, four, the Trump administration will maintain a pro worker growth path for AI so it can be a potent tool for job creation in the United States. And I appreciate Prime Minister Modi's point. I really believe will facilitate and make people more productive. It is not going to replace human beings. It will never replace human beings. And I think too many of the leaders in the AI industry, when they talk about this fear of replacing workers, I think they really miss the point. AI, we believe, is going to make us more productive, more prosperous and more free.

And for the most part, he goes on to elaborate on those four main points that he just made. The auction house Christie's announced that they're going to do an AI art auction. We can see in their X announcement here, Enter the world of augmented intelligence. Christie's first auction solely offering artworks created with AI, featuring works from Rafique Anadol, Claire Silver, Pindar Van Arman, Holly Herndon and Matt Dryhurst. And there' list of a whole bunch of other artists here that are participating and they're actually going to auction off these AI works of art. Now, most of these artists, I don't know all of their backstories, but I do know Rafiq Anadol and I do know Claire Silver. They are not just going and prompting something and taking that prompt and submitting it to auction to make money. No, these people are using AI as a creative tool and doing tons and tons and tons of work with AI to get the exact output they were looking for. So it's not just like, hey, prompt in image out, submit it for auction. And because people know my name, it's going to go for a lot. No, these are actually works that they put a ton of effort into and AI was used as a tool, as a piece of the puzzle to create that art. Yet people aren't very happy with the fact that Christie's is doing this auction and a petition has been started where almost 6,000 people have signed it as of the recording of this calling to cancel the Christie's AI art auction. Their argument here is many of the artworks you plan to auction were created using AI models that are known to be trained on copyrighted work without a license. These models and the companies behind them exploit human artists using their work without permission or payment to build commercial AI products that compete with them. Your support of these models and the people who use them rewards and further incentivizes AI companies. Mass theft of human artists work. Again, like I just mentioned, a handful of the artists that I am familiar with, Arthur, Thieves aren't trying to steal other artists work, aren't out there trying to get rich off the back of other people's hard work. Maybe some of the artists in the auction are, I don't know for sure, but the ones that I'm familiar with definitely put a lot of time and are just as much of artists as anybody who didn't use AI to create this stuff. And finally, to wrap this up, let's wrap it up with a fun one. Apple engineers created an expressive Pixar like table table lamp with AI capabilities. So check this out. They built this lamp that actually can convey emotion just with its movements. And it's all using AI to do it. So you can see how it's sort of looking at the block and just based on the angle of the lamp, you could kind of infer an emotion from it. It also seems like this lamp could be used. So if you're like reading a book and you need to move the book around or move the lamp closer, it'll just automatically do that for you. Or if you want the lamp to point a different way and light up something else, it can do that. So it looks pretty interesting. Clearly inspired by that Pixar lamp, it says here the purpose, like any ordinary lamp, is to provide light via a bulb. But it does so as if it is aware of the needs of its owner. It can Move its face closer, for example, or move further away via pushing gesture, or even by responding to oral requests. I just thought that was a fun thing to end with. I always like to end with robot type stuff or just cool gadgets, and I thought that one fit the bill for this week. But that's what I got for you. A lot of interesting stuff still happening in the AI world. Some drama, some new updates, some new cool tools rolling out. It's never a dull moment in AI and I do my best to stay in the loop with it all. Keep up with the news on a daily basis, play with all of the different tools and see what they're capable of, and then turn around and make videos. Videos to try to keep you looped in so that you never feel left out on what's going on in the AI world. That's exactly what I have this channel for. And if you like stuff like that, make sure you like this video and you subscribe to this channel to make sure more videos like it show up in your feed. Also, be sure to check out FutureTools IE where I curate all the cool AI tools I come across. I've got over 3200 in there right now. They're easy to sort for the exact tool you're looking for by using these little toggles here or by searching exactly the type of tool you need. The AI News page is updated daily, so even news that doesn't make these videos is updated on this list. And I've got a free newsletter. If you join the free newsletter, I'll email you twice a week with all the most important news in the AI world and just the coolest tools I come across. And if you sign up, you'll get free access to the AI Income Database, a database of various cool ways to make money using some of the AI tools that are available on future tools. It's all totally free to sign up for over at FutureTools IO. And one last really cool thing I'm going to be doing. I'm going to be giving away an Nvidia RTX 5090, their most top of the line highest end GPU. And I'm going to be sharing more details about how you can win that GPU starting in videos next week. So make sure you tune into this channel to learn more about how you can win that. It's a $2,000 GPU. If you can even find one right now. And I've got an extra one that I'm going to be giving away on this channel. I'll give you a hint. You do have to be subscribed to the newsletter and subscribe to the channel. And there will be one other easy, easy step and I'll share that with you next week, but you can do the first two now. Anyway, that's all I got for you. Once again, really, really appreciate you nerding out with me. Thank you so much to Adobe for sponsoring this one. Hopefully I'll see you in the next video. Bye bye.