AI Shocks Again: AI Robots with Human Brain, AGI, OpenAI, DeepMind & More (July Monthly News)

The AI scene is electrified with groundbreaking advancements this month, keeping us all at the edge of our seats. A mind blowing AI robot with human like intelligence has the world in shock. Google DeepMind's jest AI learns at an astonishing 13x faster pace. OpenAI search GPT and Critic GPT the force behind Chat GPT's prowess are disrupting industries. Strawberry, their most powerful AI yet takes center stage. GPT for all 3.0 is the AI sensation, causing a frenzy. While OpenAI's AI health coach promises personalized wellness solutions. Llama 3.1 emerges as a contender and Nemo AI boasts 128k context capacity running locally and free. Microsoft's new AI search could redefine how we navigate information. While OpenAI's latest model has the tech world buzzing with anticipation. There's a lot to cover, so stay tuned until the end to catch all the details. Alright, let's get started. Alright guys, let's talk about the insane pace of AI and robotics development. Lately, it's been moving fast, real fast. But what's happening in China right now is on a whole other level. They've created an AI robot that's unlike anything we've seen before. And trust me, when I first heard about this, my mind was blown. Let's cut to the chase. Cut to the chase. This isn't your average robot with a silicon brain. Nah. These mad scientists have gone and given it a brain made from actual living cell. You're not mishearing me. We're talking about a robot with a partially biological brain. They've crossed a line that many thought was uncrossable, redefining the very essence of artificial intelligence. Now, before you start picturing some kind of cyborg monstrosity, let me paint you a clearer picture. They're calling this thing a brain on a chip. Sounds pretty futuristic, right? Well, it kinda is. Instead of relying on traditional computer circuits, these researchers have grown teeny tiny brain cells in a lab and plopped them onto a microchip. With surgical precision, they've transplanted the essence of human cognition into the cold heart of a machine, kickstarting a whole new era of smart machines. But here's where it gets really trippy. Our brains, right? They're these complex networks of neurons, all firing and connecting. Somehow, these absolute geniuses have managed to recreate that on this tiny chip. Now let's rewind a bit and talk about how they actually pulled this off. It all starts with stem cells. The superheroes of the cell world. These little guys can transform into pretty much Any cell in your body. It's like cellular shape shifting. So these scientists took these stem cells and essentially said, today you're going to be brain cells. And somehow they actually managed to do it. They coaxed these cells into growing into this intricate network of brain cells, mimicking what you'd find in a human brain. Then, in a move that probably made every sci fi writer jealous, they stuck this cellular brain onto a special microchip. And just like that, robot brain. Now you might be thinking, this is cool and all, but why should we care? Well, let me tell you, this is a game changer. Most AI we deal with today is based on computer programs, right? It's all coded and pre programmed. But this is a whole different ball game. This robot can actually learn and adapt like a living creature. It's not just following a set of instructions. It's experiencing things and changing based on what it learns. It's closer to how our brains work than any AI we've seen before. It's beautiful. You have to guess, what's it look like? It's thinking. The implications of this are huge. Just take a second to consider the possibilities. Think about how this could revolutionize our understanding of the brain itself. Artificial intelligence. By studying these lab grown brain networks, we might uncover new ways to tackle neurological diseases like Alzheimer's or Parkinson's. It's not just cool science and it's potentially life changing stuff. But let's not get ahead of ourselves. This tech, as amazing as it is, also opens up a whole can of worms when it comes to ethics. We're entering uncharted territory here. At what point does a robot with a living brain become, well, alive? It's not just a philosophical question anymore. It's something we might actually have to grapple with in the near future. Now I can almost hear some of you freaking out. Skynet, anyone? Are we on the brink of a robot uprising? Relax, we're not quite there yet. This is still early days and there's a lot of work to be done before we see this tech out in the wild. But it does mean we need to start having serious conversations about the implications of this technology. The thing is, this isn't just about creating smarter machines. It's about blurring the lines between artificial and biological intelligence. We're talking about a future where the boundaries between AI and biotechnology start to disappear. It's exciting, it's a little scary, and it's definitely going to change the world as we know it. But here's the thing. With all this potential comes a huge Responsibility? Are you afraid that I'm going to turn into some kind of criminal? Quit worrying about me, okay? We're not just talking about technological advancement here. We're talking about potentially redefining what it means to be alive, to be conscious. These are big heavy concepts that we as a society need to wrestle with. But that's the nature of progress isn't pushes us out of our comfort zones, forces us to confront new realities. And while it can be daunting, it's also incredibly exciting. Now interestingly, Jeff Bezos and Softbank have invested a staggering $300 million in a startup called Skilled AI. And now it seems these folks aren't going for the whole lab grown human cells brain in a robot thing. Instead they're aiming to build a full on artificial brain from the ground up. They say they're pioneering what they call a robot brain, which is about crafting a general purpose AI brain that can be installed in any robot, making it capable of performing complex and varied tasks without needing specific programming for each task. Skilled AI, led by some brilliant minds from Carnegie Mellon envisions a future where robots can handle adverse physical conditions or perform dexterous manipulations in industrial and household environments. They're pushing the envelope towards developing artificial General intelligence, or AGI, which aims to achieve human like intelligence in robots. Now there's actually another breakthrough in robotics. Maybe not as crazy as human brains in robots, but still pretty wild. An Indian company's just jumped into the robot game and they're not messing around. So Zen Technologies and their partner AI Turing Technologies have cooked up this four legged robot. They're calling it Prahasta, which is apparently some badass warrior from Hindu mythology. And let me tell you, this thing's no joke, a robot dog. But instead of fetching your slippers, it's carrying an assault rifle. Yeah, this isn't Boston Dynamics Parkour bot. This thing's built for defense. Now Prahasta's got some seriously cool tech under the hood. It uses something called lidar, which is like radar but with lasers to scan its surroundings. And it's got AI that helps it understand the terrain and plan missions. These guys released a video showing off what Prahasta can do, and it's pretty impressive. This robot's climbing stairs like it's nothing, with a gun strapped to its back. It's hopping over obstacles and gaps like some kind of mechanical gazelle. And get this, they say it can even pick up a fully grown adult. I'm talking about a robot dog that can fireman carry a 180 pound person, which could be potentially life saving in dangerous situations. If this robot gets knocked over, it can get right back up on its own. They're calling it Auto Resurrection, which sounds pretty dramatic, but when you see it in action, it's hard not to be impressed. Now they've also unveiled barbaric URCWs. It's basically a remote controlled weapons station that they're claiming is the lightest in the world. And it's not just for land. And they're saying it can be used on ships too. The folks at Xen are pretty hyped about all this. Their big boss, Ashok Atluri, is talking about how these robots are going to change the game for defense operations. He thinks this is going to make people sit up and take notice of how important robotics could be in combat and recon missions. It's worth noting that Zen's not new to this game. They've been in the defense tech business for a while, making things like anti drone systems and combat simulators. In fact, they've shipped out over a thousand simulators and training systems worldwide. So what does all this mean? Well, we're looking at a future where robots aren't just in factories or delivering packages. We're talking about robots that could be on the front lines. But of course, just like with the brainchip robots, this raises some big questions. How much autonomy should we give to armed robots? What, what are the ethical implications of using AI in defense? These are heavy topics we're going to have to wrestle with as this tech keeps advancing. All right, super exciting stuff coming out of Google's DeepMind. So they've come up with a new way to train AI models that's not only faster, but also way more efficient. This method is called jest, which stands for joint Example Selection. And trust me, it's pretty revolutionary. So let's talk about it. Alright, so if you know anything about AI, you know that training these models usually takes a ton of time and a lot of computing power. It's like trying to teach a child every subject in school one by one. And it's super energy intensive. For example, systems like ChatGPT guzzle a lot of power and water to keep those massive data centers cool. There are even comparisons to the energy demands of Bitcoin mining. Yeah, it's that intense. But here's where Jess comes into play. The researchers at DeepMind have found a way to make this whole process worse. Way more efficient. How efficient? Like up to 13 times fewer iterations and 10 times less computation. That's a game changer for both speed and energy consumption. So how does JEST work? Instead of picking individual pieces of data to train on, which is what's usually done, Jest looks at groups or batches of data and picks the best ones that work well together. Think of trying to learn multiple languages instead of learning English, then German, then Norwegian separately. You might find it more effective to study them in a way where what you learn in one language helps you understand the others. That's essentially what JEST does, but with data. Here's a bit more on the technical side, but I'll keep it simple. The method JEST uses is called multimodal contrastive learning. This means it looks at different types of data, like images and text together, and it identifies dependencies between them. This approach significantly speeds up the learning process because the AI can learn from these well matched data batches much more efficiently. What makes JEST even cooler is that it starts with something called a pre trained reference model. This model helps steer the AI towards the high quality, well curated data, making the training even more efficient. Now, the study showed that using JEST they could achieve the same performance as traditional methods with up to 13 times fewer iterations. Also, it requires 10 times less computational power. So in terms of energy consumption, that's a massive reduction, which is great news for the environment. Now the just research comes not a moment too soon, as the tech industry and world governments are beginning discussions on artificial intelligence's extreme power demands. To give you an idea, AI workloads consumed about 4.3 gigawatts in 2023, which is almost equivalent to the annual power consumption of the entire nation of Cyprus. And this trend is only going up. A single ChatGPT request currently costs 10 times more in power than a Google search. It's estimated that AI might take up a quarter of the United States power grid by 2030. Therefore, any method like JEST that can drastically cut down on the power required for training AI models is incredibly significant, potentially easing the environmental impact and the financial costs associated with AI advancements. Furthermore, the success of the JEST method hinges on the quality of the training data. The bootstrapping technique utilized by JEST requires a highly curated initial data set to function effectively. This means that while JEST shows great promise for large scale, well funded AI projects, it might pose challenges for hobbyists or smaller developers who lack the resources to compile high grade training data. Essentially, this method emphasizes the old adage of garbage in, garbage out, underlining the necessity for top tier data to achieve optimal results. If Implemented widely, however, Jest could revolutionize how major players in the AI industry approach model training, possibly reducing costs and power consumption while accelerating development timelines. Another neat thing they found is something they call data quality bootstrapping. It sounds fancy, but it's pretty straightforward. It means using a small, highly curated set of data to guide the training on a much larger, messier set of data. So even if you start with a tiny, really good data set, Jess can help you scale up to train on a vast amount of data efficiently without losing quality. This is a big deal because high quality data sets are usually hard and expensive to create. Now, this method isn't just limited to one type of data or task. It works across various benchmarks, which means it's versatile. For example, they tested it on a data set called WebLI, which is a large scale collection of images and text from the web. And the results showed remarkable improvements in both learning speed and resource efficiency. So if Jest proves effective on a larger scale, it could drastically reduce the power needed to train AI models. This means we could develop more powerful AI tools with the same resources we use now, or use fewer resources to create new models. Essentially, we're looking at faster advancements in technology, less strain on our energy resources and a smaller environmental footprint. It's a win, win, win. Alright. Now you know the big AI companies in China have been making some major claims about their large language models at this AI conference happened recently. It's like a competition to see who can boast the most impressive advances, you know what I mean? So there's this company called SenseTime. They're kind of big shots in the AI game over there. They just unveiled the latest versions of their sense Nova LLMs. And get this, they're claiming the new 5.5 model is like 30% better than the previous one from just a few months ago. Wild, right? The CEO Shu Li, he was on stage hyping it up big time, saying the key to success in Chinese AI is basically building these high level logic systems using synthetic data for different industries. Xu claimed that this 5.5 model supposedly outperformed GPT4O, the latest and greatest from OpenAI, in five out of eight key metrics, according to some benchmarking data from a platform called Open Compass. I mean, take that with a grain of salt for sure, but still, that's a pretty big flex. Flexing on OpenAI like that is a gutsy move. Now, despite making these huge claims about their model's performance performance, SenseTime's stock price actually tanked like 16% on the same day in Hong Kong. Crazy reversal, right? You'd think investors would be stoked about supposedly having a model that can take down GPT4, but I guess the market had other ideas. And it's not just SenseTime making noise at this conference. Oh no. Alibaba's cloud division, the folks behind the Tongyi Chanwen LLMs, they were bragging about getting like millions of new downloads and a surge in customers for their AI platform in just the last couple of months. The cto, Joe Jingren, he was really pushing this whole open source angle hard, saying their models are now fully open source and closing the gap with closed source ones from the big Western players. Smart move, if you ask me. Makes it more accessible and gets more eyes on their tech. But here's the wildest part of it all. There's this AI startup CEO, Yan Junji, from a company called Minimax, and he straight up predicted that in the future there will only be like five companies making these huge LLMs globally. Now that's a hot take if I've ever heard one, right? I mean, could you imagine that kind of consolidation happening? Just five players controlling the entire large language model space worldwide? Insane to think about. Of course, Yan didn't actually say how many of those five would be Chinese companies, so we'll have to wait and see on that one. But the implication is clear. He thinks there's a real chance for Chinese firms to be among that tiny elite group someday. Whether that's realistic or not, I'll leave that up to you to decide. Oh, and back to SenseTime for a sec. They also showed off a couple of other new models in at this big Shanghai conference. The first is called 5O and it's this multimodal model that can combine different data types, like text, images, video, all that jazz into one seamless experience. And they did a live demo with the 5o that was low key, mind blowing, not gonna lie. So there's this SenseTime employee on stage, right? And he takes a video of himself just standing there, seemingly nothing special. But then the 5O model instantly identifies that he's attending this World AI Conference. That thing, just from analyzing his staff badge and T shirt in the video. It's wild, but it gets even crazier. The model could then answer follow up questions about what the conference is for and all the deets. Like it achieved true understanding from just that little video clip. I'm talking next level multimodal AI capabilities on display here, folks. So yeah, between the bold performance claims, the open source pushes the predictions of an LLM oligopoly and the awesome tech demos. It's been an absolute whirlwind of big moves and even bigger ambitions from the Chinese AI giants at this conference, and that's not even getting into what companies like Baidu and Tencent were unveiling. It's definitely an exciting time in the AI space, that's for sure, and it'll be fascinating to see how all these moonshot plays actually pan out in the real world. Can the Chinese models really take down the Western juggernauts? Will we see a small group of ultra dominant LLM providers emerge globally? So many intriguing possibilities. One thing's for certain though, these Chinese tech companies are clearly not messing around. They've got their sights set on AI supremacy and they're going all in to make it happen, whether that's through cutting edge R and D, savvy business strategies or sheer bravado and ambition. OpenAI has announced a new AI search engine called SearchGPT. Obviously it's powered by AI and it's setting up to be a serious competitor to the likes of Google and Bing. Lets unpack what this means and why it's such a game changer. So OpenAI is rolling out SearchGPT as a prototype to a select group of users and publishers. It's designed to combine the capabilities of OpenAI's AI models with real time web searches. What's different here is that SearchGPT doesn't just spit out a list of links, it actually organizes the information and presents it in a more user friendly way. Let me give you an example from OpenAI's blog posts. Say you search for music festivals in Boone, North Carolina in August 2024. SearchGPT won't just throw a bunch of links at you, it'll give you a summary of the festivals it found, complete with little descriptions and links to the sources. They say their new search engine aims to make finding information online faster and more efficient. Everything is presented in a clear conversational format, which makes the whole experience a lot more intuitive. OpenAI is planning to eventually integrate these search features to directly into ChatGPT, which could be a major upgrade for users who already rely on ChatGPT for various tasks. Now OpenAI is also addressing some of the major concerns that have come up with AI in search. We've seen other AI search tools like Google's AI Overviews and Perplexity, face backlash from publishers. These tools have been accused of taking content without proper attribution, which can hurt traffic and ad revenue for the original sites. But OpenAI is taking a different approach by partnering with publishers and giving them control over how their content is used. They've even got big names like the Wall Street Journal and the Atlantic on board, which speaks to the level of trust they're building. One of the standout features of SearchGPT is how it handles attributions. When you search for something, you get answers with clear inline citations and links to the original sources. This not only helps users know where the information is coming from, but also directs traffic back to the publishers. This move is directed at fostering a more symbiotic relationship between AI tools and content creators. In the early stages, only about 10,000 test users will have access to SearchGPT. OpenAI is using direct content feeds and working with third party partners to build the search results. This approach is all about getting feedback and fine tuning the system before for a wider rollout. The idea is to make web searches not just faster, but also more accurate and reliable. Now OpenAI's entry into the search market comes at a time when Google is facing significant challenges. The U.S. department of justice has an antitrust lawsuit against Google accusing it of monopolizing the Internet search industry and this opens the door for competitors like SearchGPT to make a significant impact. Alphabet shares fell 3.6% in after hours trading following OpenAI's announcement of SearchGPT. Google, which currently dominates the search market with about 90% of search engine traffic, is facing a direct challenge from OpenAI's latest innovation. This development highlights the growing competition in the AI search space, particularly between Microsoft backed OpenAI and Google. OpenAI's unveiling of SearchGPT has been seen as a significant threat threat to Google's search dominance since the rise of ChatGPT in 2022. In response, Google has integrated AI features into its own search engine to fend off competition from rivals like ChatGPT and Perplexity, which is backed by Jeff Bezos. Despite these efforts, speculation about Alphabet's potential vulnerability to AI competition had not materialized, as evidenced by Alphabet's recent double digit advertising revenue growth for its second quarter. SearchGPT, however, represents a more direct and formidable challenge to Google than previous innovations. OpenAI described it as a prototype designed to combine the capabilities of their AI models with real time web information, providing fast and timely answers with clear and relevant sources. Another really interesting aspect of SearchGPT is the visual answers feature. While full details are are still a bit unclear, it seems like this will involve AI generated videos and images to enhance the search experience. This could be incredibly useful for visual learners or for topics that lend themselves well to graphical representations. OpenAI has been steadily moving towards tighter real time web integration for ChatGPT. GPT 3.5 was an early step in that direction with more current information. Then came the Browse with Bing feature. A basic implementation, but a clear signal of their intent. Now SearchGPT takes a major leap by making AI a central part of how we search the web. It's evident that OpenAI doesn't just want to compete with existing search engines. They want to fundamentally redefine how we interact with online information. By making searches more conversational and seamlessly blending up to date web data, they aim to create a smoother, more efficient user experience. This prototype phase is critical for working out any kinks and ensuring the system functions well for a wide audience before a broader rollout. They are obviously taking the time to gather feedback, make adjustments and deliver a polished product. It's clear now that SearchGPT isn't just about finding information. It represents a reshaping of the entire search experience to be more engaging, visually compelling and tightly interwoven with the ever evolving web. So what's next for SearchGPT? OpenAI plans to keep iterating and improving the experience, especially in areas like like local information and E commerce. They'll be seeking feedback from both users and publishers to make the service even better. If you're interested in trying it out, you can sign up for the waitlist on their website. This SearchGPT release could really shake up the online search industry. With Google facing major antitrust lawsuits and new AI powered search tools like SearchGPT emerging, we're seeing big changes in how people will find and access information on the Internet. This is an area that's heating up so. So it's worth paying attention as things are about to get competitive and disruptive. Alright. Now turning to a more critical matter. OpenAI is struggling with major financial problems. According to internal projections, the company is expected to lose a staggering $5 billion this year. That's right, $5 billion. To put that number into perspective, that's roughly the GDP of a small country. So what's going on? Well, it turns out that developing and maintaining AI models like ChatGPT is extremely expensive. The company needs to pay for massive amounts of computing power to process all the data and generate responses. And we're not talking about just any computers. We're talking about supercomputers that can handle enormous amounts of information. But that's not all. OpenAI has also been hiring a Lot of new staff to help develop and improve its AI models. In fact, the company has doubled its workforce in the past year alone. And all those new employees come with a hefty price tag. Now, you might be wondering how OpenAI plans to make up for all these losses. Well, the company does generate some revenue from ChatGPT, but it's not nearly enough to cover its expenses. In fact, estimates suggest that ChatGPT only brings in around $100 million to $200 million per year. That's a tiny fraction of what OpenAI needs to break even. So who's footing the bill? Well, it turns out that Microsoft is providing subsidies to help cover its losses. But even With Microsoft's support, OpenAI is still struggling to find a way to make its business model work. The problem is that their primary focus is on research, not generating revenue. The company wants to push the boundaries of what's possible with AI, but that's a costly and time consuming process. And while ChatGPT has been a huge success, it's still not clear how OpenAI plans to monetize it in the long term. This raises some serious questions about the that fact. Future of OpenAI. Can the company find a way to make its business model work? Or will it continue to rely on Microsoft subsidies? And what happens if Microsoft decides to pull the plug? The truth is, OpenAI's struggles are a reminder that developing AI is incredibly challenging and expensive. It's not just a matter of throwing a bunch of money at the problem and hoping for the best. It takes time, effort and resources to create AI models that are truly useful, useful and effective. But despite these challenges, OpenAI is still pushing the boundaries of what's possible with AI. And that's something to be excited about. With its talented team and focus on research, OpenAI has the potential to make some truly groundbreaking discoveries. OpenAI has recently launched a fascinating new model called Kritik GPT. And it's creating quite a buzz in the AI community. Critic. GPT is essentially an AI designed to critique other AI models, specifically targeting errors in code produced by ChatGPT. And if you wonder why OpenAI felt the need to create such a tool, the answer lies in the challenges posed by the increasing sophistication and complexity of AI systems like ChatGPT. So ChatGPT, powered by the GPT 4 series of models, is already pretty advanced and is continually learning and improving through a process known as reinforcement learning from human feedback, or RLHF. This means that human trainers review ChatGPT's responses and provide feedback which the model then uses to refine its future outputs. However, as these AI models get better and more nuanced, spotting their mistakes becomes a lot harder for human reviewers. This is where Critic GPT will be very useful, even crucial. So this model, also based on the GPT4 architecture, was created to help identify and highlight inaccuracies in ChatGPT's responses, especially when it comes to coding tasks. The main idea is that Critic GPT acts like a second layer of review, catching errors that might slip past human reviewers. And it's not just theoretical. The results have been impressive. According to OpenAI's research, human reviewers equipped with Critic GPT outperform those without it 60% of the time when assessing ChatGPT's code output. This means that the model can significantly enhance the accuracy of AI generated code by spotting mistakes more effectively. Training Critic GPT involved a process similar to what was used for ChatGPT itself, but with a twist. OpenAI's researchers had AI trainers manually insert errors into code generated by ChatGPT. Then they provided feedback on these inserted mistakes. This helped Critic GPT learn to identify and critique errors more accurately in tests. Critic GPT's critiques were preferred over ChatGPT in 63% of cases when dealing with naturally occurring bugs. One reason for this is that Critic GPT tends to produce fewer small, unhelpful complaints, often called nitpicks, and is less prone to hallucinate problems that aren't really there. Another interesting finding from the research is that agreement among annotators or the people reviewing the critiques was much higher for questions involving specific predefined bugs compared to more subjective attributes like overall quality or nitpicking. This suggests that identifying clear objective errors is easier and more consistent than evaluating more subjective aspects of code quality. OpenAI's research paper discusses two types of evaluation data, human inserted bugs and human detected bugs. Human inserted bugs are those manually added by the trainers, while human detected bugs are naturally occurring errors that were caught by humans during regular usage. This dual approach provides a comprehensive understanding of critic GPT's performance across different scenarios. Ah, interestingly, agreement among annotators improve improved significantly when they had a reference bug description to work with. That just highlights the importance of having a clear context for evaluation, which helps in making more consistent judgments. Now, critic GPT's performance is not just limited to spotting errors, it also enhances the quality of critiques human reviewers often kept or modified. The AI generated comments indicating a synergistic relationship between human expertise and AI assistance. This Synergy is crucial because while Critic GPT is powerful, it's not infallible. It helps humans write more comprehensive critiques than they would alone, while also producing fewer hallucinated bugs than if the model worked alone. The ultimate goal of Critic GPT is to integrate it into the RLHF labeling pipeline, providing AI trainers with explicit AI assistance. This is a significant step towards evaluating outputs from advanced AI systems, which can be challenging for humans to rate without better tools. By augmenting human capabilities, Critic GPT helps ensure that the data used to train AI models is more accurate and reliable. Data used to train AI models is more accurate and reliable, leading to better performance of these models in real world applications. Now, OpenAI also implemented a method called Force Sampling Beam Search to balance the trade off between finding real problems and avoiding hallucinations. This method allows Critic GPT to generate longer and more comprehensive criteria critiques by using additional test time search against the critique reward model. Essentially, FSBS helps Critic GPT be more thorough in its critiques without going overboard on imaginary issues. FSBS is actually a fascinating technique. During fsbs, Critic GPT forces the generation of specific highlighted sections of code using constrained sampling to ensure these highlights are accurate. The model then scores these highlighted sections based on a combination of critique length and the reward model score. This balance ensures that the critiques are not just comprehensive but also precise, reducing the likelihood of hallucinations and nitpicks. The FSBS method involves generating multiple samples for each input and selecting the best scoring critiques. This approach enhances critic GPT's ability to identify and articulate significant issues in code, making its feedback more valuable for human reviewers. In practice, Critic GPT has shown that it can help human reviewers write more comprehensive critiques while reducing the number of nitpicks and hallucinated problems. For instance, in the experiments, human reviewers assisted by Critic GPT wrote substantially more comprehensive critiques than those working alone. This was true for both human inserted bugs and naturally occurring bugs. Moreover, critic GPT's performance isn't just limited to code. The researchers also tested its ability to critique general assistant tasks. They found that Critic GPT could successfully identify issues in tasks rated as flawless by a first human reviewer, which were later found to have substantial problems. However, it's important to note that while Critic GPT enhances human capabilities, it can't completely replace human expertise. There are still tasks and responses that are so complex that even experts with AI assistance may struggle to evaluate them correctly. But by working together, human and AI teams can achieve much more than either could alone. So by using AI to help fix AI, OpenAI is addressing one of the fundamental challenges in AI development the difficulty of evaluating and improving increasingly sophisticated models. Critic GPT not only helps catch more errors, but also improves the quality of human reviews, making the entire RLHF process more effective. There's still much work to be done, but Critic GPT is a clear example of how innovative approaches can help tackle some of the most pressing challenges in AI development now. It's no secret that OpenAI is deeply invested in pushing the boundaries of AI, constantly refining its systems, models and overall vision on a global scale. However, a recent development has caught many by surprise. OpenAI has made the decision to completely sever its ties with China, going as far as blocking access to its API within the country. And this week, OpenAI made a big decision to block access to its site from mainland China and Hong Kong. This means developers and companies in those regions can't use some of the most advanced AI technologies anymore. This move by OpenAI isn't too surprising because of the ongoing geopolitical tensions and competition in technology. However, it's a significant moment in the AI world that could intensify the tech cold war. This decision will have major impacts on the future of AI both in China and around the world, and setting the stage for even fiercer competition among leading AI powers. OpenAI's decision comes in response to increasing government demands and the rivalry for AI dominance. This choice helps protect the company's intellectual property while navigating the complicated geopolitical landscape. It highlights the growing digital divide between China and Western countries, which is becoming a defining feature of this tech war era. By cutting ties with China, OpenAI is contributing to to a broader trend of tech decoupling, where the US and Chinese tech ecosystems are becoming more separate, according to experts. For Chinese AI companies, OpenAI's blockade presents both challenges and opportunities. On the downside, not having access to OpenAI's advanced models like GPT4 could slow the adoption and integration of cutting edge AI technologies. This is especially tough for startups and smaller companies that don't have the resources to develop similar models on their own. However, the move could also spark innovation in China. Without access to OpenAI's technology, Chinese companies might push harder to develop their own. This could lead to a new boom in AI research, making the Chinese tech scene more energetic and self sufficient. Big Chinese companies like Alibaba, Baidu and Tencent are in a good position to take advantage of this situation. They have the money, talent and infrastructure to to boost their AI research and development. This could lead to these giants making even more efforts to innovate in AI and build their own alternatives to OpenAI's models. Moreover, the Chinese government has been heavily investing in its tech industry with large amounts of money and supportive regulations. This could lead to a rush of new AI research, increasing competition among Chinese companies and helping China keep up with other countries. OpenAI's move will also affect the global AI landscape. It's likely to lead to a more fragmented AI world where different countries and regions align with either the US or China based on their access to AI technologies. For example, countries in Southeast Asia and Africa, which have strong economic ties with China, might favor Chinese AI solutions. On the other hand, Europe and North America might rely more on American based AI technology. This split could have significant implications for international cooperation, data sharing, and the development of global AI standards. By controlling who can use its technology, OpenAI is exercising digital sovereignty. This is part of a broader effort to ensure that AI technologies are developed and used in ways that meet ethical standards and security requirements. I still think that international tech collaboration is vital, but companies viewing China as a crucial, crucial market now face complex geopolitical challenges. Apple, for example, is reportedly seeking local partners to provide services compliant with China's strict AI regulations, showing how firms must navigate these delicate waters. In the end, the future of AI depends not only on technological advancements, but also on the geopolitical strategies and policies that shape its development and use. OpenAI is apparently working on some new and exciting technology. According to an inside source and some internal documents obtained by Reuters, OpenAI has a project codenamed Strawberry that's aiming to take their AI models to the next level. So what is Strawberry? Well, from what I can gather, it's a novel approach to how OpenAI's AI models work, and it's focused on enhancing their reasoning capabilities. You see, one of the biggest challenges in AI right now is getting these models to not just spit out information, but to actually understand and reason about the world like humans do. Current AI models, as impressive as they are, still struggle with things that seem like common sense to us, like recognizing logical fallacies or playing simple games like Tic Tac Toe. They often hallucinate or make up information when faced with these kinds of problems. You know what I mean? Like they might be able to recite facts and figures, but putting that knowledge into practice in a reasoned, logical way is, is a whole different ballgame. But OpenAI is hoping that Strawberry will be a game changer in this regard. The Details are being kept tightly under wraps, even within the company itself. But from what I can tell, Strawberry involves a specialized way of processing the AI models after they've been trained on massive data sets. Like after the models have learned from all this data, Strawberry is supposed to help them take that knowledge and use it in a more reasoned, human, human like way. The goal is to enable these models to plan ahead, understand how the physical world works, and reliably work through complex multi step problems. Essentially, they want their AI to be able to reason and think more like humans do, which is no easy feat. We're talking about replicating the kind of intuitive common sense reasoning that comes so naturally to us, but is incredibly difficult to encode into an AI system. Now this is a big deal because improved reasoning is seen as the key to unlocking the true potential of AI. With better reasoning capabilities, AI could potentially make major scientific discoveries, plan and build new software applications, and who knows what else. The possibilities are kind of mind boggling, to be honest. Imagine an AI that can not only process and analyze data, but also use that data to formulate hypotheses, test them out and draw logical conclusions. That's the kind of thing we're talking about here. According to the sources and documents, OpenAI is specifically aiming for Strawberry to enable their AI models to perform what they call deep research. This means the models would be able to autonomously browse the Internet, gather information and use that information to conduct research and solve problems all on their own. Like, they wouldn't need human guidance or intervention. They'd be able to essentially teach themselves and explore new areas of knowledge. Kind of like how a human researcher would approach a new topic. They're even planning to test Strawberry's capabilities by having it do the work of software and machine learning engineers. And if Strawberry is successful, we might see AI systems that can not only crunch data and spit out code, but actually design and implement entire software applications from start to finish without any human involvement. I should also mention that OpenAI isn't the only company working on improving reasoning in AI. Tech giants like Google, Meta and Microsoft are also experimenting with different techniques to tackle this challenge. But Strawberry seems to be OpenAI's big bet in this area. Like their main focus when it comes to advancing AI reasoning. Now this AI has some similarities to a method developed at Stanford called Self Taught Reasoner or star. This method, it's supposed to enable AI models to like, bootstrap themselves into higher intelligence levels by creating their own training data. Basically the idea is that the model can learn and improve itself without needing humans to constantly feed it new data. One of the creators at Stanford said that if things keep going in that direction, we as humans have some serious things to think about. Like if AI systems can continually enhance their own reasoning abilities, who knows how advanced they could become? It's both exciting and a little bit scary to think about, if I'm being honest. Of course, the Strawberry project is still shrouded in mystery, and it's uncommon clear exactly how close it is to being released or implemented in a real world setting. Now, OpenAI has been privately signaling to developers and other outside parties that they're on the cusp of releasing technology with significantly more advanced reasoning capabilities, which is likely referring to Strawberry. So we, we might be seeing something big from them pretty soon, who knows? They actually showed a demo of some research project with new human like reasoning skills and at an internal meeting recently, but it's unclear if that was Strawberry or something else entirely. Regardless, it seems like they're making some serious progress in this area. Oh, and get this. One of the sources said that they saw earlier demos of what staffers claimed was the project formerly known as Q Star, which was apparently already seen as a breakthrough within OpenAI. Another source even said that OpenAI has tested AI internally that scored over 90% on a math data set, which is like a benchmark of championship math problems. Crazy stuff, right? Now I know what you're probably thinking. If Strawberry is so groundbreaking, why haven't we heard more about it? Well, that's a great question. And the truth is, OpenAI is playing their cards pretty close to the chest on this one. They're notoriously secretive about their research and development, which makes sense when you're working on technology that could potentially be world changing. But at the same time, they've gotta be careful not to overhype things too much. You know, we've seen the companies make grandiose claims about AI breakthroughs before, only for the reality to fall a little bit short. So it's understandable that OpenAI would want to keep things under wraps until they're absolutely sure they've got something game changing on their hands. That said, the fact that they're being so tight lipped about Strawberry is kind of telling in itself, don't you think? Like if it was just another incremental improvement, they'd probably be a bit more open about it. But the secrecy surrounding this project suggests that they really believe they're onto something big, something that could fundamentally change the way we think about and interact with AI. And if anyone can deliver on the promise of truly reasoned human like AI, it's probably OpenAI. And if Strawberry is even half as revolutionary as it's being made out to be, well, it could usher in a whole new era of AI capabilities that we can scarcely imagine right now. Of course, as always, there are also potential downsides and risks to consider, as is always the case with powerful new technologies. As that Stanford professor said, if AI reasoning abilities continue to rapidly advance, we as humans might have some serious existential questions to grapple with. What does it mean for our place in the world if machines can outthink us in virtually every domain? But hey, those are conversations for another day, right? For now, let's just bask in the excitement and potential of what Strawberry could represent, alright? Now on another note, there's a breakthrough in AI hardware that's particularly exciting for ChatGPT. A startup named Etched, founded by two Harvard dropouts, has introduced a new AI chip called Sohu. They claim this chip can run models like ChatGPT up to 20 times faster than Nvidia's Channel Top H100 chip and 10 times faster than the more powerful B200. If true, this could drastically improve AI performance and capabilities. Nvidia's GPUs like the H100 and B200 have dominated the AI hardware industry, making them a leader with a market cap surpassing $3 trillion. However, Etch, based in California, aims to disrupt this with their specialized Transformer ASIC Application Specific Specific Integrated Circuits chip designed specifically for transformer AI models. This means Sohu excels in running models like ChatGPT, OpenAI's Sora and Google's Gemini. But it cannot handle other AI models like convolutional neural networks used for image recognition. Etched has already raised $120 million and secured pre orders worth tens of millions of dollars. They've also partnered with TSMC to manufacture the 4 nanometer chip, which will help scale up production in their first year. While these advancements are still theoretical, the potential of Sohu could lead to exciting new AI products that were previously impossible due to GPU limitations. These new developments show just how fast AI is moving forward. OpenAI is trying to make AI think more like us, which could change how we use technology every day. Meanwhile, Etched is shaking things up with a new chip that could make AI run well, way faster than ever before, challenging big names like Nvidia. It's an exciting time because we might soon see AI that's both incredibly smart and super efficient. As we look to the future it's thrilling to imagine all the new possibilities this technology will bring into our lives. The AI world has been on fire these last few days. OpenAI introduced five level road to AGI Elon Musk dropped major updates on Grok 2 and Neuralink's brain chips, Samsung's upgrading Bixby and Amazon's rolling out new AI tools. Plus there's big news from OpenAI, AMD and even a virtual beauty pageant win, so stick around for all the details. Alright, so OpenAI has set a five level system to track its progress toward human level artificial intelligence and it's almost at level two right now. The company is on the first level which is conversational AI. They shared this system with with employees and plan to share it with investors soon. The first level is where we are now with conversational AI like chatbots. Second level, called Reasoners, involves AI that can solve basic problems and is comparable to a human with a doctorate but without tools. The third level, agents, refers to AI that can perform tasks for several days on behalf of its user. The fourth level, Innovators, describes AI that can help help develop new inventions. Recently OpenAI demonstrated a research project using GPT4 showing it has human like reasoning skills. The system for defining these levels was created by OpenAI leaders and can be updated based on feedback from employees, investors and the board. Interestingly, in May, OpenAI disbanded its super alignment team which focused on AI's existential risks. Their work is now integrated into a other research projects within the company. Alright, Elon Musk just dropped a major update. Grok 2, the next version of his AI tool might be ready next month. This announcement came in a reply on x Musk mentioned Grok 2 is going through fine tuning and bug fixes, probably ready to release next month. Now Grok 2 will be available to all X users who subscribe to the Premium plus service at $16 per month or $22 via the app. Unlike traditional chatbots like ChatGPT, Grok is designed to be witty and politically incorrect with real time knowledge of current events. It's fully integrated into the X platform. Musk also talked about Xai's ambitious plans. The chatbot is trained on 24,000 H1 hundreds and Xai is working on a 100K H100 system to create the fastest training cluster in the world. Instead of partnering with Oracle, Musk's team is handling this internally to stay ahead in the AI race. Though there's no exact release date yet. Keep an eye on Musk's social media for the official announcement. Now. He also just dropped some exciting news about neuralink's brain chip technology in a recent livestream. He aims to give people superpowers with these chips, which have tiny wires and electrodes that monitor and stimulate brain activity. Right now, the focus is on helping people with neurological conditions. The first patient, Noland Arbaugh, received his chip in January and reported significant improvements. Musk has ambitious plans for the future, envisioning enhancements even for healthy individuals. He mentioned integrating Tesla's humanoid robot parts to provide cybernetic superpowers. The next neuralink patient is set to receive their implant soon. The new device will have more threads and fewer electrodes per thread, potentially doubling its power. Neuralink is also working on safety measures like skull sculpting to reduce risks. Musk hopes to test the chip on several patients this year and thousands of in the future, depending on progress and approvals. And no, they haven't randomly implanted chips in people's brains yet. Okay now Samsung is set to launch an upgraded version of its voice assistant Bixby this year. TM Roh, Samsung's mobile chief, revealed this in an interview with cnbc, highlighting that the new Bixby will use Samsung's own AI models. This upgrade is part of Samsung's broader strategy to integrate more AI features into its devices. Launched in 20202017 with the Galaxy S8, Bixby offers functions like live translations and restaurant recommendations. The new version will incorporate generative AI technology, enhancing its capabilities. Despite this upgrade, Samsung will continue allowing multiple voice assistants on its devices, including Google's AI assistant. This announcement comes as Apple also plans to enhance Siri with AI, including integration with OpenAI's ChatGPT. Samsung's movement aims to push more AI features across its product range, encouraging users to upgrade to premium devices. Recently, Samsung unveiled new foldable phones, the Galaxy Z Fold 6 and Z Flip 6, alongside new AI tools to boost user convenience and satisfaction. TMO emphasized the importance of AI in making products more appealing, suggesting that enhanced AI features will drive sales of new Samsung products. So expect a smarter, more capable Bixby soon and stay tuned for more updates. Alright now. Microsoft and Apple are stepping back from OpenAI's board amid growing antitrust scrutiny. Microsoft, which invested $13 billion in OpenAI, immediately gave up its observer seat. Despite this upgrade, Samsung will continue allowing multiple voice assistants on its devices, including Google's AI assistant. Apple was expected to join, but now won't either. Instead, this announcement comes as Apple also plans to enhance Siri with AI, including integration with OpenAI's ChatGPT. Samsung's move aims to push more AI features across its product range, encouraging users to upgrade to premium devices. Recently, Samsung unveiled new foldable phones, The Galaxy Z Fold 6 and Z Flip 6, alongside new AI tools to boost user convenience and satisfaction. TMRO emphasized the importance of AI in making products more appealing, suggesting that enhanced AI features will be drive sales of new Samsung products. So expect a smarter, more capable Bixby soon, and stay tuned for more updates. Alright now. Microsoft and Apple are stepping back from OpenAI's board amid growing antitrust scrutiny. Microsoft, which invested $13 billion in OpenAI, immediately gave up its observer seat. Apple was expected to join, but now won't either. Instead, open.