Major AI News #25 - Gemini Upgrades, GPT Store Leaks, Realtime AI Phonecalls and More

So this week in AI was absolutely incredible with a range of new generative AI technologies being built on top of the new technologies coming to market. And trust me, you're going to want to pay attention because everything here is pretty much insane. So one of the first things I wanted to talk about was this. Okay, now essentially this company is called Rabbit and or the website is Rabbit Tech. And this is by far one of the best things that I've seen this week in AI and I. And trust me when I say it's absolutely incredible. So essentially what we're looking at here is their new framework, which is a large action model. So you know, we have LLMs and multimodal models, but they built something called the large action model. And essentially I'm about to play the video, but it talks about how they use this to develop their own operating system. So they have their own operating system and then they use a large action model which they've created that works, I'm pretty sure it works exclusively with their own operating system. And, and then they use that to make AI agents that actually work. Now if you haven't seen the demo, you might not know what I'm talking about completely, but essentially just watch this video and then I'll talk more about why this is completely game changing is.

A new foundation model that understands human intentions on computers. Through it, we can teach Rabbit OS how to use specific applications. In this video, I'm teaching a Rabbit how to book an Airbnb while I'm operating normally as a human. On the left screen, watch closely on the right as the large action model is learning all my inputs and im imitating my behavior in real time. So I'm trying to plan a trip to Barcelona with my wife and my daughter. The first thing I'm going to do is navigate to the anywhere option and I'm going to type Barcelona in the search field. The system's suggesting Barcelona, Spain, which is exactly where we want to go. Using the website's calendar tool, I'm going to mark our check in on the 15th and check out on the 21st. Now I'll click Add Guests and adjust the members accordingly. Now let's hit the search button and see what pops up. Since we love the beach, let's make sure to select the beachfront option. And for a more private experience, I'm going to select entire home so we have the whole place to ourselves. For the budget, I'll set a maximum at 400,000 won and a minimum of 100,000 so that all the options are within our price range, we're going to need at least two bedrooms to make sure we all have our own space. Finally, with all of our preferences set, we've got plenty of options that fit the bill. I'll just start browsing for the perfect one. Each training only takes a few minutes and does not require access to an application programming interface, also known as an API, nor do you need anything installed on your device. You only need to train each workflow once. Let's try to use Rabbit OS and instead book a room in London. My extended family is going to London. It's going to be eight of us and four kids. We're thinking of December 30th to January 5th. It's not set in stone yet, so I just want some general options. Can you look it up for me?

Sure, I can help you with that. The first option is a home in Portobello Muse house, priced at 1,348,351 won per night with a rating of 4.8.

The large action model supports mobile apps, web apps and professional desktop apps. It learns directly on the user interfaces and acts on them. We have already started the training process for the most popular apps. As you're watching this video, Rabbit OS is learning fast and adapting to hundreds of applications. The ultimate goal of Rabbit is to define the first natural language operating system that replaces apps on your device. It's time for the machines to do some serious homework.

So, yeah, hopefully you guys were able to see that demo. Hopefully they didn't use any copyrighted music, but essentially I'll give you guys a quick overview of this video. Essentially, they showed how you could essentially train Large Action model with a few quick steps. So you basically essentially say, look, this is what I want you to do. You essentially give it a demo and then it essentially learns exactly how to do that. And why this is pretty much crazy is because they have a physical device that works on their Rabbit OS system and it can pretty much do everything. So this, ladies and gentlemen, is the very first AI agent that we've seen that pretty much works. We've known that AI agents kind of work. They sort of work. Many people have tried to get GPT4 to kind of work in some regards, like 50% of tasks. There was a recent paper recently that talks about how GPT4Grounded achieves 50% of web tasks. But this is something that does work on many different things and you're able to train it easily. Which means that the ease of use here is going to be insane because it's not like something that has a high barrier to entry. You don't need to learn the software, you don't need to download any software, you don't need to learn any code. It's just very simple. Plug and play. So this right here, ladies and gentlemen, is really, really the next step. And if you don't know this company, I would recommend taking a look at this because this is of course their device. Definitely watch the keynote. You can either watch it on their channel, you can watch it on the channel that we posted the video to. I mean, I made a video, it is pretty long, but I talk about all the extensive uses for this. And before a lot of people think that this is just another gimmick, I would say watch the entire demo because you it is really, really, really good. And the demo is like 25 minutes and you might be thinking, I wish there was a five minute version. Just trust me. It's actually worth it because we are entering a new era of AI powered devices and this is something that you really don't want to miss out on because I do think that whilst many different things are gimmicks and will not be useful, I think this is going to be something that's actually pretty useful because they've actually got a real working AI agent product. Then essentially we had something from Figure. So Figure is a company that makes robots. They actually only started 18 months ago and they have made astounding progress on the ability to have robots do things that humans need them to do. So essentially they said here, figure 1 has learned how to make coffee end to end AI system trained in just 10 hours just by watching humans make coffee. Our neural networks are taking video in and trajectories are coming out. Join us to ship a fleet of AI robots. So take a look at this video because I find it to be absolutely fascinating and it is something that they touted as a real breakthrough. And I think what we're seeing here is like I said before, in 2024 will likely be the embodied AI in these robots and this robot here. In just 18 months they managed to go from conception to product, well, not really product to something that works. And you can see it's able to grab this little thing that you need to put into a coffee machine. Sorry, I don't drink coffee. But it was able to do that pretty effectively and pretty quickly. Now I do want to say as well that this is pretty fast for a robot because if you've seen robots before, one thing that people don't forget, one thing that people do actually forget is that sometimes you're seeing it in 10 times speed, which means that faster robots that we've seen from Tesla's Optimus Generation 2 and robots like this show us that these robots are getting faster and faster. And clearly these guys are learning how to make these robots more efficient and more effective. Now one thing that I did really find interesting about this as well was this part right here. So it says Figure 1 has learned how to self correct mistakes. So you can see on the left hand side, it doesn't get this, you know, little thing inside the coffee machine straight away, but when it does mess it up, it does move it around until it eventually gets it right and then it decides to close that machine. Now I know that there was some discussions on Twitter stating that this isn't impressive because there are other robots that can do this. But do remember that this is not a specialized robot designed to do this one tasks. Other robot that are, you know, narrow AI that are specialized to do one task, yes, they're going to perform vastly better at one specific task. But this is a general robot that watched humans do something and then managed to do this. So you have to think about the broad applications that this puts us in. Because if this is something that can just imagine they develop a system where you can train just based on watching humans do something that is going to be something that is really scalable and more effective than a singular AI system built for one specific task. And general purpose robots and general AI AGI is definitely what we are edging towards and it's definitely something that we do need and this puts us closer to that state. So this is something that I will leave a link to and I think this company is going to have a lot more developments in 2024 because of just how fast they've been moving. Now with more in robotics, if you haven't seen this video, I'm pretty sure you have, but yeah. So to just basically recap on this video, this is essentially the mobile aloha, which is of course the autonomous teleoperated system that manages to perform many different tasks, from as little as 30 to 50 tasks demos. This is a highly, highly effective robot at doing many general purpose tasks. And although it isn't like complete, although, you know, it does mess up sometimes, like we saw in a few videos, it is something that shows us just how effective robots are going to be in the near future. And I mean, honestly, the autonomous mode that I did see and the ability to do a wide range of tasks really did shock me. Like, I really didn't expect to see a robot be able to do this kind of stuff in such a general nature. I know previously we did see Google's Palm E robot, which did kind of shock everyone. But this definitely was a shocker because it showed us that many tasks that we thought human hands are the only things that are really strong enough, effective enough, fast enough, efficient enough to be able to do the task with the speed that we needed to be done, isn't actually true. And we can see here that this robot was able to do these tasks effectively. As I was saying, if we do manage to get these robots deployed, and if we do manage to get these robots working effectively, we will eventually get ourselves, you know, a suite of robots that are going to interact in the environment in a way that we do want because, you know, interacting with the environment in a way like this. And of course this demo that you're seeing right now is completely teleoperated, which essentially means that there is a human controlling the robot. But it does show that, you know, it's not just hands that are very good at interacting with our environment. It is these grippers that can, you know, in their claw like way, still interact in a very effective manner. So we might be seeing robots with claw like hands in the nearby future interacting with many different stuff. And I got to be honest, robotics is taking off quite a bit. And although like I said, this is teleoperated, which I will show you in a moment what exactly that means, I think that this, you know, if we do get autonomous mode to a really, really high degree of accuracy, that's going to be something that is absolutely incredible. So here we have the video of the Aloha teleoperation system. And this was something that wasn't included in the original video because this video only just came out now. But like I said, this is a behind the scenes video on how things can work. And I think that maybe in the future we might get, you know, these as jobs because, you know, a lot of robots are going to be deployed and I think we're going to see large scale use of robots. So you might actually see some people controlling a lot of robots virtually, you know, maybe from another place that actually interact with the physical environment. And I do think that that would be an interesting career robot controller. But I do think everything changes once these robots manage to somehow memorize, you know, what you're doing with the environment and manage to do it effectively. Like that kind of reinforcement, like that kind of AI explosion. So that will be fascinating to see how it develops. Then we had Luma Labs AI introducing Genie 1.0, our first step into building multimodal AI. Genie is a text to 3D model capable of creating any 3D object you can dream of in under 10 seconds with materials, quad mesh, repology, variable, polycount, and all in standard formats. And as someone that had did 3D work before, I can honestly say that this is going to be game changing in terms of your workflow. Because having something that is literally text to 3D, 3D modeling and 3D assets are one of the, you know, most time consuming things that you really do. Because not only do you need to model the model to make sure the poly counts right, you need to make sure it's all smooth, then you need to make sure that you texture it right, you need to make sure the UV maps right, you need to. It is something that is going to save hours and hours and hours of time. So I can't imagine how much, you know, how much value is being provided to those who work in the 3D space. If I was definitely someone there, that would be very effective. And I know it might just be very basic stuff because with certain projects you require a certain theme or a certain style of images. But like I said before, every week we're seeing small, small improvements on this scale. And overall, I think in the next couple of years we're definitely going to get full, amazing, you know, text to 3D cities or you know, text to 3D working rigs. So that's definitely going to be something very effective and useful for those working in the CG industry. And I can't believe I didn't see this before. So this was a video that was actually quite a long time ago. And like I said before, sometimes you do miss things because the pace of AI development is so quick that some things just do get buried under the avalanche of news. But essentially this guy managed to jailbreak Google's home device. And essentially, if you don't know what Google's home device is, it's basically Amazon Alexa, which is, you know, just Google's version. And if you don't know what that is, it's basically like a speaker that you can talk to and it can say stuff and it can control your accounts, it can play your music. So you can be like, Alexa, play this Alexa, turn off the lights. If it's connected to your home smart system. And this guy managed to jailbreak it. So it was pretty cool. And I think you guys should take a look at this because, I mean, I don't know why. And I'm pretty sure the only reason that they've not done this is due to hallucinations and maybe false facts. But AI, you know, home devices are definitely going to get a huge, huge upgrade in 2024. And if they aren't, I'd be very, very surprised because maybe, maybe the only reason that they wouldn't do it is because they're working on something far more effective. So take a look at what is possible with current technology. And this was literally in 2023, mid 2023. So this was, you know, relatively early on in the AI super cycle that we're seeing now. Hey, do I have any new messages today?

Hey, Justin, you have three new messages. There is one from your dad on WhatsApp about some unexpected snow, one from David from Twitter sharing a link to a research paper. And Jenny is messaging the tennis group on Signal about playing this afternoon.

Thanks. And did John message a few days ago about an event? When was it again?

Yes, John sent a message on Friday on Messenger about a birthday barbecue at Golden Gate park this Saturday at 11:00am okay, great.

Can you reply and say that I'll be there?

Sure. I replied to John saying that you'll be there this Saturday.

Oh, one more thing. Is Alan there? Can I speak with him?

Sure, one moment. Let me get Alan Watson for you.

Hello, Justin. How may I help you today?

Hey, Adam.

I'm just working on a demo right now, but I look forward to chatting more later. All right, good luck with the demo and let me know if I can help with anything else. Otherwise, have a great day. Yeah. In this video, I know that the latency wasn't that good, but like I said, it was someone that just jailbroke it. And previously in the video, if you have watched the rabbit one, the rabbit tech video, that one, they have 500 milliseconds, which is absolutely incredible, which is basically, if you don't know what I'm talking about when I say milliseconds, it's basically the time that it takes for the AI system to respond. And you know how when you're talking with a real person in life, they respond absolutely immediately. That's closer and closer and closer to reality. So this kind of stuff is really, really fascinating because it shows us just how much is available with the current technology. And if, you know, someone's able to do this really quickly, then it means that, you know, an entire company dedicated to doing this is going to be doing a lot like we did see previously, another thing that is here is real time. Now we did see this before, but this is real time Gen from Leonardo. And Leonardo is honestly pulling out all these stops in order to become the number one player in a AI, you know, image and AI video. Like then the on auto motion, if you don't know, it's, you know, text to video, it's really, really good. It allows you to move, you know, images and stuff like that. And this is, you know, of course, real time. Essentially it just means that you're able to generate your AI images in real time, which I guess allows you to, you know, at first I was really confused and then I realized that, you know, the problem with Mid Journey is that, you know, you put in a prompt and then five seconds later, 10 seconds later, you can see exactly what you've got. But with this, as you're entering your prompt, you're able to see exactly how certain words manage to change the direction of the image, which is going to allow you to dynamically adjust that so you can get exactly the output that you want. So this kind of thing, this much, much faster feedback that we can get from these systems is going to just give us a much, much faster workflow and just honestly be much more effective. We had this video that kind of broke the Internet and essentially it was an AI generated CCTV footage of arresting people for wearing huge boots. And this actually did take the entire Internet by storm because this actually got 25.6 million views. And this was something that, like I said before, managed to slip under the radar. But the funny thing is is that AI generated videos aren't really that good. But for some reason in the context of cctv, what we've seen is that they do look kind of real. In fact, they don't look kind of real, they look really real. As in, you know, if I was browsing through some, you know, content and I did see that, I would kind of, you know, at first glance I would kind of believe that this is real. Which goes to show that technology isn't always how you imagined it. Like when you imagined, you know, text, a video being made, you wouldn't think that, you know, the most effective use of it at the time would be fooling people with fake CCTV images. But nevertheless, it's a hilarious clip that goes to show that, you know, AI generated videos can have unexpected outcomes that everyone could enjoy. Then we had something called Mo Mask and essentially this is where we had something that's really cool because like I talked about how the CG animation industry is getting affected by AI. We had text to animation. So essentially what I'm guessing that they've done here is they've managed to, you know, have text input and then get animations output. And what's cool about this is that this is essentially a skeletal system. So you input the text and you get like a skeleton and then you can apply that skeletal rig to any character. Which means that, you know, for example, you could put a person walking, a person jumping and then you could get the animation output and then you could put that animation into any software that you need. Well, it's mainly Blender and then you could animate any 3D model. So it's really cool and like I said, it's really effective because it's, you know, for beginners who are trying to animate, this is something that's just so easy to use. And you know, like I said, generative AI is impacting everything, even in ways you'd think. In addition, we also did have Samsung say Galaxy AI is coming. Join us at Samsung Unpacked on January 17, 2024. So in around seven days we should be getting a crazy announcement. Now I think, I'm not entirely sure but I do think that this could be somewhat related to Bard because this is very, very similar to the Bard icon logo. This like Bard is these two logos right here. So I'm pretty sure that it is going to be something about Bard. But Google haven't really said anything but there are some crazy, crazy Google leaks. But before we get into that, Samsung also did have another trailer in which they talked about AI. And this is the trailer that they made. And this is the trailer that they made. Now I will play this if it is is copyrighted. I will just show you the video after. But do take a look. Phones were so different back in the day, remember? That's crazy. I'm calling you from the back of the taxi. Then came TV phones. Yep, they got bigger and smarter, Stronger even. Then they became more than phones. I mean who would have thought the flip phone would make a comeback?

So what's next?

Well, who knows. So essentially in this trailer I'm not sure if I could play the entire thing because there was some music playing in there, probably might be copyrighted. I don't want, I don't want the entire video to get taken down. But essentially they talk about how, you know, Samsung's first mobile phone was in 1988. The world's first TV phone was in 1999. And then we had immersive displays in 2011. Then we had water resistant devices in 2014 and then we had the first ever Samsung Pay in 2015. Then we had, you know, foldable phones in 2020. And then of course, get ready for a new era of mobile. And I'm pretty sure that this is going to be the AI related announcement because Samsung, like I said previously, did say that on January 17, 2024, there's going to be something that they do announce. So is it going to be similar to Rabbit's new announcement? Is it going to be better? Is it going to be worse? I have no idea. Because Rabbit did honestly genuinely shock the entire industry with their product. And I want to see, because I'm guessing that they've managed to catch everyone off guard. Is it going to be better than what Galaxy release or have Galaxy managed to, you know, figure out how to do AI Agent and embed that into a completely new phone? Then we have something interesting. Bland AI. So introducing Bland Turbo, the world's fastest conversational AI. Send or receive up to 500,000 phone calls simultaneously, responds at human level speed in anyone's voice and program it to do anything. This is pretty crazy.

Hi, this is Sophie from Bland. I saw you filled out our online form. Is now a good time?

Yeah, now works perfect, perfect.

Can you tell me a little more about your business?

Yeah, we're trying to automate our political campaigns, phone calls.

That's a really interesting use case. I'd love to get you set up with an appointment with one of our reps. Does next Thursday at three work?

Yeah, it works great.

Great. We're looking forward to chatting with you.

All right, bye now. That is absolutely incredible. Previously, like I said before, in the previous videos we talked about, as latency decreases, we're going to get increasingly more effective AI systems that we're able to converse with in real time. And finally it's here. We talked about Play HD before and Bland AI just comes out of the gate with this shockingly fast AI system. So I think that this is going to be a major, major, great piece of technology for effective use cases. But you know, in some, you know, cases where this is not used for the right cases in terms of, you know, mass spam calling people, this is going to have severe ramifications. I mean, how are we going to deal with someone who's able to call 500,000 people simultaneously just using an AI system? I mean, how are privacy regulations going to affect this? Because, you know, constantly calling people, is there going to be, you know, spam calls? Is there going to be restrictions on, you know, using these AI systems to call over a certain number of people. I mean, the entire, you know, Internet and society is now changing because we have AI agents that can browse the Internet and pretty much do anything you want. Then we have, you know, AI systems that can call you a million different times on a million different numbers and essentially, you know, call an entire city in many different voices and have conversations with people and change their opinions. So, I mean, technology is getting to a rather interesting point very, very quickly. So it's important to pay attention. And I think in the future, this is going to be a future of how many companies do reach out to people, do have support, do help people. And I wonder if people are going to start to realize that, you know, you're not talking to a real person, you're talking to an AI system. And I wonder if people are even going to care because, you know, if it sounds real and it's effective and it does get the job done, are people going to care if it is something that is such a chore? I mean, you know, there's many, many different questions that are going to be here. So I would say this is going to be something that is definitely going to be a real big player in the future. Then we had open Interpreter 2.0, the new computer update. Essentially, it's an operating system thing that you can basically just type in and it just does anything you want to on your system. So think about it like this. It's basically an AI agent that controls your entire operating system. So something like this is going to be really effective for those of you who maybe want to change certain things but don't know how to do them. I'm sure that this is going to consistently get better and better, but it was definitely something that I did want to highlight because it did get a lot of attention on Twitter and they did manage to update this and we did cover this before. But like I said, it is something that is pretty, pretty effective. And like I said, like Bill Gates said, you know, when I saw his Gates notes about AI agents. Bill Gates is usually quite right about the future and, and it's clear that 2024 and beyond will definitely be about how AI agents impact pretty much everything like they are right here. And of course, we had Adam D', Angelo, the CEO of Quora and Po, raise $75 million. And essentially they're using that funding to accelerate the growth of po, and the majority of it is going to be used to pay bot creators through recently launched creator monetization program and essentially it's basically very similar to the custom GPTs. And I would say that this is underrated because if they have 75 million DOL to pay out to creators, I would say that if you're someone that wants to earn money from AI systems and you're able to create a good bot, I would say definitely create some on PO because it is a really effective platform that doesn't have half the user base of ChatGPT. And in that it definitely means that you're going to be able to get into a niche market and you're going to be able to definitely earn a lot more money from that. And I got to be honest, I do use PO quite a bit when sometimes other services are down because they actually do allow you to use pretty much all AI systems in one, because you can use Claude, you can use GPT4, you can use other AI LLMs as well. So it's actually a really good system that I would definitely check out. And of course here we have Nabla raising another 24 million for its AI assistant for doctors that automatically writes clinical notes. So it says Paris based startup Nabla just announced that it has raised 24 million in a series B funding round led by Cathay Innovation and essentially they're valued at $180 million. So it says Nabla has been working on an AI copilot for doctors and other medical staff. And the best way to describe it is that it's a silent work partner that sits in the corner of the room, takes notes and writes medical reports for you. So this, like I said before, if you remember previously in the certain videos where we discussed how AI will impact healthcare, we did discuss how NABLA and other technologies are going to be impacting healthcare in assisting doctors in, you know, note taking, research, diagnosis, and right now we're currently seeing the first stages of that. So essentially what NABLA does is NABLA uses text to speak technology to turn the conversation into a written transcript. It works both in both and it works works both in person consultations and telehealth appointments. And after the patient has left, the doctor hits the stop button. NABLA then uses a LLM model refined with medical data and health related conversations to identify the important data points in the consultation, medical vitals, drug names and pathologies. Then they generate a thorough medical report in a minute or two with a summary of the consultation prescriptions and follow up appointment letters. And these reports can be customized to the doctor's needs with personalized formats for your notes. So yeah, this is going to be really effective and this is going to be able to save a lot of time. And I really do hope that we are seeing a lot more health stuff. There was also something crazy in health related to AI. So this guy, Brian Rommel, made human thermal GPT and what he was able to do with this was absolutely incredible. So he was able to get this AI vision system to diagnose someone with a rare condition. I'm not exactly sure who it was, but he says the air output from this human thermal GPT was the right lymph gland in the neck is showing an abnormal higher heat range than would normally be expected. It is recommended that you follow the recommended steps below to manually test the area and report the results to your doctor. And then you can see here he tweeted that Justin Dr. Diagnosed this person with a very, very early detected strep throat and said, I don't know how you could have known this yet, as there is no redness or swelling, a swab verified it and they are now on an antibiotic round. They never felt sick at all. Human therpo GPT success. And one thing that has been in discussion for quite some time now, and that is more of a discussion on Twitter and even Greg Brockman, which I will show you in another tweet, is that these AI systems that we're using do have really specialized knowledge. And if we do get them to the point where they're really, really effective at being able to identify patterns and to be able to identify, you know, what's going on in your body, they're going to be more effective than a doctor, because Greg Brockman, I think it was Greg Brockman that recently talked about how he was only able to get a diagnosis for a certain illness for his wife because he was able to get one doctor in conjunction with another doctor to effectively communicate. And when they were effectively communicating, they managed to see that A combined with B produced C. And, you know, many doctors specialize in certain fields, and when they specialize in certain fields, they don't really know sometimes how it affects another area in a certain specialty. And it was something that he was able to use as an example to show why if we managed to get an AI system that's able to combine everything, draw patterns from a lot of data that doctors simply wouldn't be able to do, we're going to be able to get personalized healthcare that's much more effective than what we currently do have. So you can see here that this is where he said a Reason we need beneficial AGI. After five years of pain across many systems in her body, a broken foot from stepping off a curb, debilitating migraines, fatigue, joint pain and instability. My wife was recently diagnosed with a genetic genetic disorder called Hypermobile EHRs Danlos Syndrome. And because the medical system is designed for individual specialities, while heads affect every system in her body orthopedics, cardiology, neurology, gastroenterology, dermatology, etc we spent five years seeing more doctors and specialists than in her whole life prior. Most doctored was only focused on what was relevant to their own speciality. We were lucky that her allergist put together the pieces after observing and hearing her full set of systems and issues. As human medicine has progressed it seems like we increase doctors depth at the expense of breadth. We need better tools to be able to deliver depth and breadth simultaneously to patients. This is one promise of AGI if built right, reliable, individualized, affordable healthcare in your pocket. Like a panel of today's top doctors across every specialty working together in concert to keep you healthy and without you needing to fax forms between them. There's still a long way to go on this technology and on learning how to deploy beneficially along with the appropriate professional human oversight in high stakes areas like medicine. But the promise is getting increasingly clear, thoughtfully approached by technology developers, healthcare providers, governments and society. There's much hope for better care for every member of all of our families, including our non human furry ones ones. And it couldn't have been said better. If we do manage to get this right, the implications are going to be great for everyone. Then essentially we had another AI avatar demo from hey Jen. If you haven't seen hey Gen before, it's something that we did talk about but it is now gaining a bit more traction in the sense that people are starting to realize how realistic this stuff does look and it is kind of shocking. It is also kind of cool to see how far technology has come, but take a look quickly because it is a little bit concerning. This is Justine's AI avatar coming to you live. It's so good to meet you. Do I look and sound realistic enough or can you tell that I'm not the real Justine? Let me know what you think. And I do think that if this technology gets any better to the point where we're able to use just a picture of someone. And I know deep fakes are a thing, but you can only really get that from simply being able to have multiple images of a person. I think once you're able to do it from one image and you're able to do it from like simply a second, like you are with audio, then this is going to become a real issue. And then last but not least, I can't believe I nearly missed this, but we had a GPT Store leak. So apparently this is what the GPT Store is going to look like. And I do think that this is real based on the user interface, based on some of the gbts that are there, and based on how it's been organized. Now, this hasn't been confirmed. I mean, in a couple of days we're going to really know if this is true or not. But right now this is speculation, but there has been some, you know, floating around ideas that this could be real. So I would say take a look at this, see what character, see what categories they're going to have, because you can see that they've got writing, research and analysis, program programming, education, productivity, lifestyle. You can see what GPTs are already popular, which is AI PDF. I don't know why GPT5 is trending. They've got consensus, formerly researched GPT, Grimoire Scholar, AI. So it will be fascinating to see how these GPTs do work. I think I might make one, but I'm not entirely sure which route I want to go down. But I definitely do use the custom CPTs for a lot of personal applications and I do think that when it does release, it's going to be a huge, huge deal. So that will be fascinating to see how exactly you earn money and how exactly that all does work. One of the last things I do want to show you just how crazy AI generated images are getting is the fact that Mid Journey and Stable Diffusion are able to generate images like this now. And previously we knew that things looked realistic and I can't believe I nearly forgot this. Of course, Google Bard does have some leaks, so essentially they're going to be doing a paid tier. So it says try Bard Advanced for three months on us. And I'm pretty sure that's how we're going to be able to access Gemini Ultra through the paid subscription, just like how we access GPT4 through the ChatGPT plus subscription, which is 20amonth, so that should be coming soon. And then of course we do have additional Bard leaks, which is Motoko, which will reportedly allow users to create custom bots similar to GPT Store, which would be fascinating. Then power ups, gallery, tasks, sharing and more. And honestly, I have no idea what a Motoko is or what these power ups are going to be because Google is very secretive with their star stuff. But whatever it is, we know that we're going to find out soon because with the release of Gemini Ultra going to be coming in Q1, which is what they did say early 2024, we're really going to figure out exactly what this is soon. And in addition, something that I did want to add, and I really hope this is true, because if it isn't Apple, I honestly don't even know what to say anymore. But a new version of Siri with generative AI is rumored again for WWDC in 2024, which is essentially Apple's big conference where they announce a lot of stuff. And I really do hope they do this because if they really don't, they are going to be behind the eight ball. And I know that Apple does like to delay things, but I think AI is one of those things where delaying it just isn't going to work. But I guess we'll have to see because other companies are moving very, very fast with their applications and AI is really different to a lot of other industries. If you enjoyed this Week in AI, don't forget to subscribe to the Channel. Subscribe to the weekly newsletter and don't forget to subscribe to the Channel. Sam.