OpenAI REVEALS New PRODUCT! , Ellevenlabs New MUSIC Tool, Anthropics New Prompt Engineering Tool

So with an absolutely incredible day in artificial intelligence, let's take a look at some of the stories you most likely missed. So coming in straight away, I'm not going to waste any time getting into this story. OpenAI plans to announce Google search competitor on Monday, sources say. Now this is probably the most fascinating piece of news because I'm actually going to analyze a lot of the things that are spoken about in this article, as well as some other tweets that were just recently released. And I think combined with the video that I made recently on OpenAI's Search Thing, I think we're really in for some interesting week in artificial intelligence. And not this week now, but I think next week is probably going to be pretty, pretty crazy. So yeah, enjoy the weekend while it lasts because things are probably going to go up another level. So essentially, and I'm going to show you why this is pretty crazy, it says. Reuters says that OpenAI plans to announce AI powered search product on Monday, according to two sources familiar with the matter, raising states in its competition with search king Google. And essentially, this is pretty crazy, okay, because this is something that was not that crazy until I saw some recent tweets. So essentially, okay, what actually just happened now is that OpenAI recently just tweeted that we'll be streaming live on OpenAI.com on 10am Monday, May 13th to demo some ChatGPT and GPT4 updates. So OpenAI here are basically stating that there's going to be some secret updates, but here's what actually made this post really interesting. Sam Altman actually just tweeted that this is not GPT5 and not a search engine, but we've hard but we've been hard at work on some new stuff we think people will love. Now here's what he said that I think is even more interesting. Okay? And maybe I'm just someone that's a bit too excited or a bit too intrigued by this, but he did say, feels like magic to me. Monday, 10:00am Pacific Time. So I think that right now this is pretty strange because they are actually stating that this is not GPT5, which I don't think that was a surprise. I think everyone knew that GPT5 is probably just going to be released in maybe two months or so. And that timing of the release is pretty coincidental when, you know, OpenAI's traffic usually drops. So that one actually does make sense. And then essentially he does state that this is not a search engine. So I'm guessing that maybe the previous claims about what it was is probably untrue. But I think I now know what this is because if Sam Altman is stating that this is not a search engine and we've seen previously that there have been a lot of different things regarding that. So for example, when I was doing the video on what I thought the search engine was going to be due to the a lot of different things being spoke about, of course, Reuters, they did say that there was going to be a search engine. So considering the fact that Reuters is a reputable source, I think maybe there might be a search engine, you know, sometime in the near future. The fact that they're stating that it's not a search engine means that either one or two things happened. Either one, the sources completely got it wrong or either two, it was a search engine. But because. Because they delayed it last week. So there was supposed to be announcement this Thursday, but because they delayed it now, they probably delayed the search engine announcement to something else because it's probably not ready and they're about to announce something else. Now, of course we already know what's coming, we know there's going to be some ChatGPT and GPT4 updates, but I think that this is either going to be one of two things. I think mainly that this is where we're probably going to see the new GPT2 model in action. I'm not exactly sure what the GP2 model model is, but I did previously recently speak about this and I basically spoke about how I think GPT2 is much likely a smaller model for on device purposes and how we've seen smaller models get more and more efficient. If you haven't been paying attention to Phi 3 and Llama 3 models are pretty much the same size but just as efficient as other models, which is pretty incredible if we talk about how effective they are. And also OpenAI are developing a new product like a new LLM that could run on personal computers and other devices and this is something that was reported on with the information. So I wouldn't be surprised if they maybe announced some kind of LLM. But I really just think that if they're going to, you know, demo something that feels like magic, I think there is the potential for it to be agents. The only reason I don't think that this, you know, think isn't going to be agents is because they actually spoke about how agents is probably going to be coming with the release of GPT5. So I think, I think it's going to be interesting because what they says is that it feels like magic. We don't know what it is, but clearly, clearly, clearly what this does show us and something that I've been telling you guys before is that OpenAI is coming for the full stack. And the full stack means that they are coming for voice, they're coming for video, they are coming for text, they're coming for agents, and they're coming for personalization. So anything you can think of in terms of AI that is a product, coding, whatever, they're pretty much coming for it. It's not just like a chat GPT. OpenAI has audio, video, images and stuff like that. So I'm guessing that, you know, with regards to the GPT4 updates, maybe it's probably just the new image generator because we also do know that secretly and this is something that you don't. If you don't know. I'm going to tell you guys right now, there's actually a new update to ChatGPT. Like the image generation software is just incredible. I don't know what they did to it, but it's actually good. So Dall E3, it's not Dall E3 anymore. It's. It might be Dall E3, but it might be a super version of Dall E3, because I use Dall E3 all the time and what's going on in GPT4 is not the same at all. So it might be like a way for us to get more Images out of GPT4 because that's something that they stealth updated. There wasn't like a blog post about it, but I know 100% because things are photorealistic now. And yeah, so my honest bet is that maybe it's going to be something about GPT2. There are of course going to be updates about GPT4 that could include things like the new enhanced Dall E3 or potentially it could be agents because that is something that OpenAI did talk about. But I'm thinking that, you know, another thing I do want to state is that the timing is very, very important because like I said before, if you don't know, it's a big week for AI, and the reason it's a big week for AI is because not only are OpenAI doing something on the Monday, which is pretty, pretty crazy, okay, Monday is going to be the 13th. Okay, now why is that so crazy? It's because literally the day after that, Google's IO is on Tuesday 14th May 2024. So we have a situation where there are literally going to be back to back AI events which are pretty Pretty intense and I think that OpenAI, whatever they're doing okay, they've clearly changed their thing and they clearly want to show the world that they are ahead of Google. So the fact that they are doing this one day before Google's IO is pretty, pretty competitive, which shows us that we're, whatever OpenAI do release. They're probably just going to try and blow us out of the water again. I don't know, maybe it's going to be something pretty bas crazy, maybe it's not. But I think those two days are going to really show us where we are in terms of AI because I always think the opening eye going to surprise us and they usually do. So if they've been able to do that with video, I, I wouldn't be surprised if they weren't able to do something crazy. So my best bet is something with GPT2, something with agents or you know, something crazy in terms of, you know, better usability in terms of chat GPT because they did say updates are going to come to that website. But whatever it is, it might surpass Google's IO, which I'm still excited for.

The current generation of models that we're seeing out there is still in $100 million range. You know, I think all of this can scale to the $100 billion range at the end of that range with some difficulty in terms of power and chip provision. But we're going to, we're going to get there in a few years so that, that's a factor of a thousand more and then I think the algorithms are going to get more efficient and we're going to see things, some things on top of post traffic training, at some point we're going to run out. But, but I think there's, there's lots of things stacked on top of things to keep making models.

So that was Dario Amade, the CEO of Anthropic actually talking about how models are going to get a lot more expensive and there's going to be a lot more investment. And of course, you know, he's, of course, you know, he spoke about how the, you know, 100 million dollar range, which is pretty insane when you think about any kind of, you know, traditional piece of computer technology that you're kind of building. I mean, you know, just when you actually think about the sheer amount of numbers in terms of financial backing that is required to run these systems, it's actually pretty, pretty crazy. And I think this is why a lot of people are starting to realize that in the future it's Only going to be a handful of companies, maybe even three or two that can really, really, really afford this. And unless you've got billions of billions of dollars like Apple does and some of these top companies and you make billions of billions of dollars a year afford the compute, you really, really can't compete at the level that these guys are playing at. So I think that, you know, this is why a lot of people are stating that, you know, in the future the, the, the big companies are only going to absorb the smaller ones and they're only going to extend their lead. Because you got to think about it like this, okay? If we have, you know, we look at the, the Fang or we look at the, the Nasdaq and we look at, you know, the top 10 tech companies, whatever, these guys have billions of dollars, okay? They're not in debt. They can afford to spend a few hundred million dollars, you know, training the next generation of AI models or generative, you know, systems, whichever ones are going to be the most profitable or most impactful. And I think when these guys are training models that are like, you know, $400 million and we start to get to the billions of dollars of models and you know, things start to get really, really crazy, I think, you know, the other companies just won't be able to catch up at all and their lead is just going to be surpassed because with every model that is released after that, it's just going to get better and better and better and the lead just, just carries on increasing. So I mean it's pretty incred incredible because there was also a recent Reddit post And it says AI godmother. Stanford's NLP lab only has 64 GPUs. Academia is falling off a cliff in terms of AI resources and it's pretty important to understand that like, you know, this is something where governments should be investing in because these things are really, really important in terms of understanding the nature of AI and how quickly we are moving. So I think there's going to be continually that journey giant disconnect in terms of these giant AI companies and where we are in terms of academia. Because you know, yes, whilst these giant companies are mainly focusing on efforts that are just profit based, I guess you could say, in the sense that they want to make models that, you know, customers can enjoy and can impact humanity in a way that's profitable. Academia just doesn't really have the financing to be able to get any of the compute that they need to conduct meaningful research some of the times. Now continuing to talk about anthropic or Claude 3, we know that they recently released something called the Prompt Engineer. So it says you can now generate production ready prompts in the Anthropic console. Describe what you want to achieve. And CLAUDE will use prompt engineering techniques like chain of thought reasoning to create more effective, precise and reliable prompts. And I'm not surprised that they released it. But one thing that I've slowly started to realize about Anthropic is that they've started to start shipping more. That sentence was pretty awful, but they've literally started to ship a lot more products recently. And you can see right here that this is something that's really, really effective. Now if you don't know why prompt engineering is so important, I can guarantee you that the average person is using AI in the wrong way completely. And I got to be honest, when I started using AI, I was actually using it in the wrong way. Like, you know, going up to chat GBT and asking a basic question. If you have a basic question, that's fine. But trust me when I say using different mental models, using different prompting strategies and all the kind of different ways that you can interact with the AI system just dramatically change your results, especially like five shot as well, things like that. This dramatically change your results. So essentially you can see here your generated prompt. So you can see someone say classify chat transcript into categories using our content moderation policy. Then you can see generate prompt and then you can see here that it's able to generate this prompt for you and then it says you'll be acting as a content modern modifier. Here's the policy you should enforce. Carefully review, yada yada yada. If there's no so this is something that I'm surprised people haven't built before. But Anthropic are making this into a very, very efficient tool. And I'm guessing that this is going to become commonplace in the world of AI because whilst yes, a lot of people have spoken about how prompt engineering is the future for, you know, standard use cases, I think this is going to be really, really useful. And I think that this is something that is quite important, especially for individuals trying to get to grips with how they can use prompt engineering at a lot better way. And of course this is prompt engineering. If you're wondering, is my prompt engineering career going away? I think this is maybe why Anthropic was hiring for prompt engineering because I remember a while back I was actually looking at AI careers for some people and on Anthropic's thing they were actually hiring I think it was around 2 to 5 to 300k a year for a AI prompt engineer. So maybe they were using that to build the system and maybe they're going to continue to use that person to make this system even better. But I still think a prompt engineer is needed because, you know, if you're working within an organization, you're still going to need to develop custom prompts to your entire organization and have the best prompts in order to get the most out of the future AI systems. But, but I'm wondering one thing, I'm wondering is if this will change because as models get better in terms of their reasoning abilities, you know, as OpenAI manages to make breakthroughs, I'm wondering if those prom techniques will still be as effective. So that's of course going to be something that, you know, it might change as the future evolves. Then we actually had big news here. Like this is actual big news. And you can see here's an early preview of 11 Labs Music. All of the songs in this thread were generated from a single text prompt with no edits title is started to sing pop popty rock, top country, top chart song. And this is pretty crazy because what we have here is 11 labs doing music. And someone did recently state that, you know, oh, there's going to be a new AI music tool that completely blows everything out of the water. I mean, I got to say this is rather impressive and I know that eleven labs has a lot more users news. And you can see here's an early preview of 11 Labs music. All the songs in this thread were generated from a single text prompt with no edits title is started to sing pop poppity rock top country, top chart song. And this is pretty crazy because what we have here is 11 labs doing music. And someone did recently state that, you know, oh, there's going to be a new AI music tool that completely bros everything out of the water. I mean, I got to say this is rather impressive and I know that 11 Labs has a lot more users than Suno AI and Yu Dio. So I'm wondering if this is going to bring it to the mainstream because recently there was actually a AI song that recently did just go viral. And I'm wondering how this is going to impact the music industry because I don't have too much impact in in the music industry in terms of like knowledge and stuff like that. But what I do know is that it can impact it in surprising ways. And a recent AI generated track going viral. Maybe, maybe it's going to Change how some platforms are. I mean, are we going to see million different AI generated songs uploaded to these platforms? Are they going to be swamped? Are they going to be different rules and regulations? But enough about the industry. This is 11 labs doing their thing, and I'm going to show you guys 60 seconds of three different songs, because they're three different genres. So whatever your musical taste is, then you can, you know, judge how good you think it is.

We were having fun programming young Dreaming that one day we'd make it work. Lines of code we'd write all night Hoping that one day we'd get it right. Can we teach the machine the same?

Sing.

Can we teach the machine to dream?

And then.

It started to sing. No, this can't be real. It's just a machine. But the melody Oo, it's all too real. And those words, they make me feel.

That literally sounded way too realistic. But here is the jazz version, which is just. Just listen.

We were having fun, fun programming young Dreaming of one day becoming someone. Lines of code we'd write all night Hoping that one day we'd get it right. Can we teach the machine to sing? Can we teach the machine to dream? And then, ooh, it started to sing. No, this can't be real, it's just a machine. But the melody, it's all too real. And those words.

Now here is Broke my heart. And this is the last one I'm gonna play you. There is one more, but I just want to continue with the video. So here's the last one I'm gonna play before I give more of my opinions on this insane software.

If I never said before Tell me what you're here for. What's wrong with being me? Why do you have to see Love gone empty? Love gone empty. Broke my heart to fill you up. Broke my heart to fill you up. It's like breathing in and out. Or inhale the love and let you down. Broke my heart to fill you up in. Like breathing and out. Cause I can't no more Go through the back door. Cause you're too much to bear. Cause I don't even care.

And I gotta be honest, this is really, really uncanny. Like, really, really uncanny in a different way. Because, you know, hearing an AI voice is sometimes okay. Because usually an AI voice doesn't ever sound human, truly. Because, for example, you know, I can, you know, stop. And then I can talk like this. Like, a little bit slower. Like, I can slow down my words, or I can, you know, speed them up and sound a little bit More enthusiastic like this. And then I can speed and speed and speed. But an AI system just won't do that. Now with this though, with music, this is uncanny to me because it just sounds so realistic. Like I really do believe that there is a human singing that. But knowing now that that is an AI is just really weird. And the thing is I've listened to a few music tracks before and some of them, once I started to enjoy them I just turned them off because I like this is just weird. Like I don't want to become some like person who's merged with AI already. But maybe that's just me being like an old man afraid of technology, even though I'm not that old. But the point here is that I think we're starting to move into that strange area where AI is just truly, truly excelling in a lot of different areas. And I'm sure that a lot of people don't won't even realize that this is AI. So I'll probably be making a full video on this because this is just truly incredible, like genuinely like mind blowing stuff. But as someone uses who uses 11 labs pretty frequently, I'm wondering how good this will be. Considering the fact that you know some of their voice stuff, it's great but sometimes it can glitch out. So I'm wondering how the music will work on that because you know, people already using 11 Labs quite a lot. So we now have three main music competitors. We've got Yuio, we got Suno AI and we've got 11 Labs Music. So with that being said, that that's just something that's pretty interesting. Now here's something I really do want to talk about because I spoke about this before a while back and I kind of knew that this was going to get done. But I think people might be freaking out a little bit in the wrong way and I'm going to explain why. So it says another reason why Open models are important. Leaked OpenAI pitch for media companies. So it says additionally members of the program receive priority placement and richer brand expression in chat conversations and their content based benefits from more prominent link treatments. Finally, through PPP, OpenAI also offers licensed financial terms to publishers. So some people are basically stating here that this might be bad because what they're going to be doing now is I guess you could say is like when we had search engines and now it's just, you know how Google search is pretty much unusable due to ads. We're now moving to an area where that is going to be the same. So with this, basically, if you don't know, what we currently have is a system where, you know, you talk to ChatGPT, you ask it a question, it doesn't really have a priority on the type of articles it chooses. It doesn't, you know, favor one or another. It doesn't really favor one piece of information from any piece of source. But right now I think what's going on is OpenAI and some other companies have realized that there's actually an opportunity there because priority placement and richer brand expression are things that you can do. For example, let's say I ask an AI, you know, how do I fix my car? Or let's say I say, how do I get a better night's sleep? It might suggest certain brands. Because what I've realized is that when you talk to chat TVT and you ask it certain things, sometimes it will suggest certain brands. And those brands, like being, you know, placed in a generative AI system is a huge, huge, huge deal for a company. You're going to get a lot of customers from that, okay? Because a lot of people are going to be switching over to these systems in the future. And if your brand is there where, where they're going to be, you know, suggesting your content more, they're going to be favoring your content more. And your content benefits from more prominent link treatments, meaning that, you know, if they ask you about, you know, the weather or something, it's going to link to your website. They ask you about, you know, information on a particular topic and then it says, oh, you can visit this website here. That is going to be a really, really interesting thing. So I'm wondering if that is going to change the dynamic of how we interact with these AI systems because some people are saying that it's the enchitification of generative AI, because maybe this might actually be a way for, I guess you could say, open source models to make money as well. Because we know that open source struggles to make money. But if you have something where, you know, you can work out a few deals with a few brands to get your, you know, links and your, you know, prominent placements, you know, within the chat system, that could be a viable method. So I think it's going to be interesting to see how this works in the future. But I don't think this is something that's not going to happen at all. Because, you know, at the end of the day, like I said before, these companies, they're here to make money, so how can they make more money? Striking more deals making the platform, but, you know, hopefully it's something that's so subtle that we honestly don't notice.