Leveraging AI

174 | NVIDIA's big announcements, The US-China AI war intensifies, Microsoft-Open AI frenemies relationship evolves and more important AI news for the week ending on March 21, 2025

• Isar Meitis • Season 1 • Episode 174

Share episode

What happens when AI outpaces the systems we built to control it?

In this AI news episode of Leveraging AI, we’re unpacking a whirlwind week in tech—from Nvidia’s trillion-parameter dreams to robots doing backflips, AI wars with China, and Claude finally learning to Google.

We break down the strategic moves, global tensions, and bleeding-edge breakthroughs that business leaders need to stay ahead of—before AI makes the decisions for them.

The takeaway? AI is no longer the future—it’s the infrastructure. And leaders who wait to adapt will be adapting to those who didn’t.

In this session, you’ll discover:

  • NVIDIA’s GTC drop: From Blackwell Ultra to real-time trillion-parameter inference (yes, trillion).
  • Robots flipping out: Side flips, cartwheels, and motion-captured breakdancing.
  • The AI-China tension escalates: OpenAI vs. DeepSeek and the new digital Iron Curtain.
  • Anthropic’s new “Think” tool: Why Claude’s new reasoning could change enterprise AI forever.
  • Vibe coding & developer disruption: Why 10-person teams can now do the work of 100.
  • Enterprise AI integrations: ChatGPT’s new connectors, Google’s Gemini Canvas, and the feature war heating up.
  • The AI chip arms race: Meta’s in-house chip, X.AI’s mega data centers, and the death of GPU dominance?
  • AI ethics in motion: Monitoring chain-of-thought, catching cheating models, and the looming question—can we still keep up?

About Leveraging AI

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Speaker:

Hello and welcome to the Leveraging AI Podcast, a podcast that shares practical ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. And welcome to this weekend news episode. We are going to focus on three different topics, the announcements from NVIDIA's GPU Technology Conference known as GTC that was held this past week. We are going to continue talking about the AI war with China, and we're also going to talk about the updates to robotics in this past week that have made some big leaps forward and their potential impact on the workforce globally. And then there is a long, fun list of rapid fire items that we're gonna touch on a lot of stuff this week, including some interesting announcements from the big companies. So let's get started. NVIDIA's GTC conference was held this past week in San Jose in California, and the event attracted more than 25,000 people in person, and over 300,000 people participated globally making it the world's biggest AI party. And it happens every time nvidia, does one of these events. They released a lot of new announcements. the biggest one was the release of Blackwell Ultra, which is their next model. And they also revealed what is going to be the successor of that model that is coming up after that. So Blackwell Ultra will have 40% more memory capacity then the existing Blackwell technology, and that will enable to run bigger models in a more efficient way, like g PT five and cloud four and so on. The focus was very, very clear on the importance of inference, which is the generation of AI tokens versus the training aspect where Jensen Hong specifically emphasized that running the models is going to be the focus of compute demand in the coming decade. There's more and more competition to GPUs on that particular aspect, and that was a very clear focus of everything Jensen was talking about. And he's claiming that their next infrastructure will allow realtime handling of trillion parameter reasoning models and the Vera Rubin platform, which is their next platform combined with their dynamo interconnect technology is supposed to reduce latency in realtime calculations and inference by 50% compared to existing GPUs. They also shared DGS Spark a new version of their existing DGX technology, which allows developers to basically run their own local AI computer, and they're partnering with companies like Dell and HP and Lenovo for the distribution of those. But as you know from Nvidia in the past few years, it's not just about the chips, they have been developing a lot of software infrastructure for ai. So they announced that they've launched Lama Nitron family of open source models that is optimized for multi-step reasoning and decision making, and that it's showing 92% accuracy on benchmark tests, which is 15% improvement to the baseline underlying meta LAMA three models. Nvidia also open sourced it's Q Op optimization engine, which is a decision making tool used by Walmart and UPS right now to reduce delivery route costs by 12 to 18%. It's integrated with omniverse and it allows companies to simulate the supply chain scenarios in digital twins before deploying the actual changes in the field. They also share that they're partnering with DeepMind on synth id, which is Alphabets or Google's if you want watermarking system for AI generated content, so it will now be integrated into NVIDIA's Cosmos Foundation models. Syn ID will basically enable to track every content that these models generate and know for a fact that they're AI generated that will hopefully help to fight or at least mitigate the risk of deep fakes. Now, since we already mentioned syn ID and Google, something interesting about Google and then we'll come back to Nvidia users started using Google's new Gemini two flash AI model they've added as an experimental to their Google AI studio. People started using it to remove watermarks from trademarked images, including Getty Images and similar websites. So on one hand they have synth id. On the other hand, their tools allow people to remove, physical watermarks from images. I assume Google will block that in the very near future, but that's just a funny thing that happened this week in relation to this. One interesting announcement is the collaboration with Disney. They're planning to build a generative AI pipeline for animated films using OMNIVERSE and RTX 1590 GPUs that were just launched at GTC. And the main thing that they will enable Disney to do and other artists after that is to render scenes in real time. While the AI agents handle tasks like lip syncing and background generation and so on, and the anticipation is they will reduce the estimated production time and costs by 40%. This is very clear why Disney and other creators wants to use it, and I assume other people like me that are generating content regularly will probably benefit from this down the line. Now, despite the fact that in January, 2025, just a couple of months ago, Jensen remarked that quantum computing won't be useful for the next 15 to 30 years. Nvidia unveiled the NVIDIA Quantum Accelerated Computing Center, N-V-A-Q-C in Boston, where they're partnering with MIT and Harvard for research and simulating quantum computing capabilities on GPUs. That's an interesting move and it will be very interesting to see where Nvidia moves forward in that field that definitely have the money to invest in that direction if they wanted to. Now, one of the biggest fears that were raised recently is US tariffs and their impact on Taiwanese chip imports and their costs, which are a key to NVIDIA's success. While they mentioned that they're investing$200 billion in US manufacturing insulate short term margins, now, analyst wasn't necessarily convinced and they're claiming that Onshoring chip production can remove four to 6% of margins by 2027 just because of the higher cost of manufacturing in the us. Uh, in general, investors and analysts weren't impressed with this whole presentation by Nvidia and the stock actually went down this week, not by much, but usually after these kind of events, Nvidia stocks went up and it was the opposite in this particular case. One of the reasons is Jensen, as I mentioned, is clearly stating that inference is going to be the main thing moving forward because of the reasoning models and the new scaling laws around the reasoning capabilities of these models. And there's a lot more competition on the inference side compared to the training side to GPUs. So company like Cerebra and Grok are offering low cost in friendships that have been running for the last few years, providing better results, faster results, and cheaper results than GPUs on inference. And in addition, the big players google, Microsoft and Amazon are developing their own chips, and now meta is going into that field as well. So there's gonna be a lot of competition to Nvidia in the inference field. Nvidia also gave some great examples of actual real world benefits of AI capabilities that they're providing, so they shared that their new system, holo Scan three, which they demoed at GTC, enables real time AI analysis during surgeries and that they're partnering with Medtronics, which is a giant in that field, that the systems is going to, as an example, reduce laparoscopic procedure times by 25% while providing better results. They also shared that they are playing in the drug discovery field with their EVO two biology model that predicts protein interactions 50 times faster than legacy tools that are doing the same thing. And this can accelerate things like vaccine development. And in another field they're playing in General Motors announced that it will equip all of its 20, 27 vehicles with NVIDIA's Drive, Thor system, enabling level four autonomy through their Group N one robotics platform. So like in every single time Nvidia does this thing, it's on one hand really exciting, on the other hand, really, really scary. They have their solutions across everything ai from software to hardware to infrastructure, to robotics, to metaverse, to everything you can imagine. And they're definitely cementing their position as one of the key players for an AI future. I cannot stop thinking of Cyber Net every time I see Jensen does one of his presentations and watching what Nvidia is releasing. But it is very, very obvious that they are going to continue investing and participating in more and more aspects of the AI universe and get the grip on more and more components in this AI ecosystem. And now, as I mentioned, let's talk a little bit about the continuous pressure and competition in the air world between the US and China. OpenAI has formally called for a ban of Chinese deep seek models, claiming that it could be compelled by the Chinese Communist Party, the CCP, to manipulate its models to cause harm. So in a formal letter to the US Office of Science and Technology policy, OpenAI proposed banning China produced equipment and AI models in tier one countries and tier one companies, including partners like uk, Canada, and Germany due to privacy violation and security risks. They are comparing the risks from these kind of models and specifically deep seek to the risk that is possessed by Huawei's chips. That has been banned by the US government a while back. Now, if you remember, not too long ago, just when Deep Seek came out, same. Altman actually welcomed competition from China, but now they're taking a lot more aggressive position against it. Whether that's politics or actual real concerns is very hard to tell. Critics of this point are actually saying that deep seek, open source their models, which allows it to run on US servers without any access to anybody from the outside world. So again, you can decide for yourself whether you think this is more politics and trying to block competition from China, or this is a real concern to US national security. In parallel, the US Department of Commerce has banned the use of deep seek models to government furnished equipment, including government, employees, and any other government devices citing concerns about the Chinese company potentially accessing sensitive information. The Commerce Department bureaus sent a mass email to staffers introducing them to do not download view or access any application, desktop apps or website related to deep seek. That's pretty broad. I mentioned to you in previous episodes that similar bans has been implemented by several states and even several countries. So states like New York, Virginia and Texas already have these bans on deep seek products and there's a new bill in the house called No Deep Seek on Government Devices Act. That were introduced as a bipartisan effort to block it from any government devices. In addition, beyond government devices, wall Street Journal reports that Trump administration is weighing to potentially broader restrictions, including potentially banning deep seek apps from the App store in the US and from all US cloud providers. So time will tell whether we will even have access to deeps, seeq in the US through standard channels. As I mentioned, other countries such as Australia, Italy, and Taiwan has been completely banning deep seek on all government devices in some other countries such as France, South Korea, Ireland, and Belgium are considering similar steps. Now an interesting approach has been taken by Amazon. So if you remember, I shared with you that a few days after Deep Seek showed up, Amazon already started offering it as an open source solution on its AWS platform. That being said, Amazon just started warning customers and all of its employees as far as privacy concerns over using deep six's AI models. So again, is this real security threat because Amazon clearly stated when they allow people to use deep seek, R one on AWS that, and now I'm quoting inputs and model outputs aren't shared with any model providers basically saying this data stays on AWS. So what is the concern? Well, I think in this particular case, the concern is very clear. Amazon is developing its own AI models called Nova, and they are a hybrid reasoning architecture similar to Anthropic Cloud 3.7 and what five is supposed to be. And they are planning to offer it as the main solution on AWS. They're planning to make it slightly cheaper than deeps seek. And the idea here is obviously that deeps seek models on AWS are competing directly with their own reasoning models that they're going to release. So I. Warning people from the risks associated with it is a good way to push their models even further. Again, I'm not here trying to protect deeps seek by any means. I would prefer that US companies and western hemisphere companies and countries will run on Western hemisphere generated AI models. That being said, I think this particular case has nothing to do with real risks and more with commercial reasons. But staying on the topic of the race between the US and China, Baidu, the Chinese parallel of Google. So it's a gigantic company that is running all search in China has just introduced two new models called Ernie 4.5, and there are two different versions of it. One is a traditional model and one is a reasoning focused model designed to compete with deep seek R one and its parallels from the us. Both models are now freely accessible to individual users through Ernie's Bots official website. Now they're claiming it's a fully multimodal foundation model that has exceptional multimodal comprehension capabilities, allowing it to process and understand text, images, audio, and video. And they're comparing it to the latest top models from the us. Now another interesting change in direction in Baidu. Their previous models were not very successful on and were not very highly accepted, even in China. And now that there's additional competition, they're open sourcing their models, which they haven't done before. So Ernie will become open source in June 30th of this year, making it obviously more attractive for developers and other integrations both in China and around the world. So where do we stand continuous competition between the US and the Western Hemisphere and China on leading models that are going to continue coming from the top labs and companies on both sides of the ocean. And I am sure this will be an evolving topic that we're gonna be talking about more and more. The current US administration is very obvious in its push to make sure that the US stays ahead in that competition. And there's very strong tailwind from the US industry to work together with the government to push the boundaries of what's possible as far as speeds and capabilities of ai, regardless of the risks that is associated with it. Just because they believe or presumably believe that the risk from China getting there first is riskier than whatever they're going to put out there. I. Do not completely agree with that, but I must admit, I understand the concerns that even if we slow down that China won't, and then there's no point in slowing down. So on one hand, I'm terrified with what that means as far as running fast without the right safety guards. On the other hand, if we don't, and they do, I don't know what the risks are over there, but it sounds like a lot of smarter people think that's a serious concern. Now, our third deep dive topic for today are robotics. Two new, interesting capabilities were demonstrated this week. So I shared with you last week or two weeks ago but Re's G one model performing karate moves that were incredible, where this week they've done something new. They've done the world's first side flip with the G one model. It's a pretty cool video of the robot doing a side flip, which as I mentioned, no other robot was able to perform before. Their H one model was the first model to perform a backflip in March of 2024, and also the first robot to run at a speed of 7.3 miles per hour, which still holds the Guinness World Record for that sprint. Now how are they training? It is actually really interesting. They're training it on NVIDIA's Isaac simulator, which is the ability to train the AI brain behind the model in a simulated environment before running it on the actual robot itself using NVIDIA's Sim two real solution. And it's the ability to train these robots in a simulated environment, basically training just their brains without the robot ever moving. And once it gets to a good enough capability only then they actually provide these capabilities to the robots, which makes it a much faster and much safer capability to train these robots. In a parallel demonstration. Boston Dynamics robot named Atlas has performed some pretty cool break dancing capabilities as well as cartwheels, again, showing much more advanced movement capabilities and much smoother and human-like movements compared to previous versions of the models. And they're actually taking a slightly different approach to perform that. They are using motion capture suits worn by actual human operators to train the robots have to move in a more fluent and more human-like way, which means the robots can now mimic, analyze, and understand how human moves in different scenarios and make the robots move in the same way without having to quote unquote program the AI on how to do that. Now while these are cool and not necessarily relevant to anything, they show significant advancements both in the mechanics of how the robots move, as well as in different ways and processes to train these robots to do more and more stuff. I think we're looking at a much accelerated timeline for when robots will start showing up. Definitely in the workforce, in factories and so on. But very shortly after, probably in different services, I assume we'll start seeing robots in factories more and more in 2026 and 2027, and probably in other places, as I mentioned in previous episodes. So think about things like Starbucks or people who do cleaning of offices and things like that. We're going to start seeing more and more robotics in these kind of jobs. And then shortly after that in our houses doing housework and chores and yard work and so on. Now the housework will probably take longer, mostly probably because of security risks of what happens if this thing malfunctions or does something unexpected. But I do think that in factories, we'll start seeing a mass deployment probably in 2027 and 2028. What does that mean to the global workforce? Well, it means that in addition to the risk to white collar jobs, blue collar jobs are at risk as well. And you've heard me say that time and time again. I don't see anybody explaining what is going to replace that as far as jobs for actual humans. Everybody sounds that they're very optimistic and they're saying, oh, every previous revolution generated new jobs, so this one will do it as well. I. But I haven't seen a single one of those leaders explaining exactly what these jobs are going to be, which direction they're going to be, or more importantly, why do they think this will happen in this time as well? Because in previous revolutions, the biggest benefits that humans had is that we were smarter than the machines and we let the machines do the manual labor. And while we did smarter and smarter work, while jobs that were lower paying jobs stayed with humans because it was easier to give humans a low paying job than to develop a really sophisticated machine that will be able to compete with it well. Now, unit three, the same company we talked about a second ago is selling this G one robot for$16,000 a pop. And that's just the beginning of the scale up. Once these things will scale to much larger numbers of production, the number will fall to probably single digits, thousands of dollars, which means even places that require very cheap labor will be able to be done by robots instead of humans. And I have very serious concerns of what are these new jobs that will be generated? And even if they will, I think the timeline that will take jobs away is significantly more condensed than the timeline that will generate new jobs that we can't think about right now, which generates a few years at least, of serious concerns to the global job market, which has a lot of other impacts on everything else that we know. Because if people don't make money, they don't spend money. And that has a lot of impact on the global economy. And now let's switch to rapid fire items. There are a lot of interesting things to talk about. The first one that could on its own have been an entire episode, is other conversations about vibe coding. I talked to you about this in the last two episodes, but now Gary Tan, the CEO and President of Y Combinator, has highlighted how vibe coding is transforming their startup world and how companies with 10 people can scale to 1 million and tens of millions in revenue in just less than a year by using Vibe coding. If you remember, we discussed this previously, vibe coding is a term that was introduced by Andrej Carpathy post a few months ago talking about how he felt when he was using Cursor for the first time and in an interview with CNBC Tan, again, the CEO of Y Combinator said you can just talk to the large language models and they will code entire apps, potentially allowing 10 engineers to do the work previously required by teams of 50 to 100 people. Last week I shared with you that 25% of startups in the current batch of companies in Y Combinator are generating 95% of their code with ai, despite the fact that they have very capable founders that can write the code themselves. This is another field where human labor is being replaced very, very quickly by developers. Now, these companies still need developers. The fact that I cannot write code without understanding it doesn't mean I can become a coder. People who have coding experience can develop things that are significantly more sophisticated and faster than I can. But as they mentioned themselves, they can do with one or 10 people what was done previously with a hundred, which means they're not gonna hire the other 90 to 95 employees that they would've hired in the same scenario, developing these kind of capabilities. So on one hand, it enables amazing new innovation in the software world because significantly smaller teams can develop really sophisticated capabilities very quickly. But on the other hand, it means less work for less people, unless we create a lot more startups that will hire these fewer developers. I'm guessing that it will end up with a loss of many developer jobs. Now, two different analysis companies have done analysis of AI generated code, and they're claiming that there's significant increases in duplicate code. As well as that it forces developer to spend more time on debugging, and yet it is very, very clear what the direction is with, even with these additional issues, it is still worthwhile for companies to use AI to write most of the code, and I'm sure these issues will be decreased over time, which will make it even more efficient and more tempting for more and more developers and companies to use this methodology. Now since we mentioned Andrej Carpathy, he had another prediction this week, which aligns with something I was talking about a lot in multiple episodes of this podcast. He's claiming that 99% of web content will soon be optimized for large language models and AI rather than humans. Now, I agree with him 100% because what is going to happen is that we're going to have agents that will browse the web for us, which means we have to learn how to prepare the content on our websites or whatever is going to replace websites with something that will be easier for machines to read, then easier for humans to read. So think about maybe a single file is well structured with the right formatting to allow AI agents to understand all the information in it without any fancy HTML and CSS user interface. Now, the interesting thing that I think will have to happen is for a while, the next few years, we will need to keep two parallel websites, one for humans, because humans will still continue to browse websites at least in the next few years. And one for agents. Now, what does that mean for the world? I'll give you an example from my personal experience and how the world can dramatically change in that field, and that will give you, I think, a great idea of what can happen in the rest of the world. So I used to run a reasonably sized travel company called Last Minute Travel. Last Minute Travel did about a hundred million dollar a year, which sounds like a lot, but it's very little in the travel industry. So if you compare it to Expedia all the brands that work under that. And by the way, under these two companies are most of the travel websites that you know. So while you think you know what you're doing and you know which websites are better than others, it's all a scam. It's all owned by two companies. And they also own the quote unquote comparison websites like Kayak and Trivago. So there's no real competition in the travel world. But putting that aside for a second, these two companies, between Expedia and all their brands and Booking and all their brands pay over$10 billion a year on digital marketing. Basically making sure they're top of minds and that everything you're gonna click and search on Google is gonna show their solutions. But what if that's not the case anymore? What if nobody goes to websites? And what if nobody looks at ads? What does that mean to the huge amount of traffic and huge amount of money that these companies are generating? What does that mean to the way the internet even works? So if you think about how the internet works today. The internet is free. You're not paying for using the internet. You're paying your ISP for their service, but you're not paying for the internet itself. What runs the internet is ads, right? So people are paying money so they can send direct ads to you based on your search, based on your behavior, based on your cookies, and so on. And all of that may go away with the introduction of agents and them browsing the world. So how does the internet continue functioning in this new model? I don't think anybody knows, but it's coming very quickly. Now. Those who will figure it out faster will gain a huge benefit. So again, think about the example I just gave you. What if a company that has access to the travel backend, which many companies do figure out how to do this better than Expedia and booking.com, they can undercut them very quickly as soon as agents take over more and more of the search of the web universe, and that is true for any other industry as well. So any company that depends on web traffic for its livelihood needs to think very deeply on what that means to them and what steps they need to take in order to sustain and preferably grow their current dominance in this new universe. And now for really quick updates on several new releases from multiple companies, OpenAI is going to introduce a new version of their voice capabilities on the API. They just introduced three different voice models. Two that transcribe voice to text, and one that generates voice from text. And it will have higher accuracy, and it will also allows you to control the vibe of the voice that is being generated to tweak it to be more like a pirate or like a bedtime story or whatever other mood you want to create with those voices. And it can do it in a hundred plus languages. And the cost is really, really low. It's developer friendly in its pricing, and it starts at 0.30 cents per minute to transcribe and 1.50 cents to generate voice, which is really cheap and significantly cheaper than what it was a year ago. And both these and all these models are available through OpenAI API right now. Another very interesting announcement from Chu is ChatGPT now testing connectors. What connectors do is it's letting ChatGPT team subscribers link Google Drive and Slack to the ChatGPT universe for smarter context aware responses. This means that ChatGPT will be able to see, read, understand, and relate to files, presentations, spreadsheets, and slack chats when answering questions about internal company data. So you can ask things like, tell me what's the status of this project, and it will go and aggregate information from all these different sources and give you answers. If you think about it, this was the promise, maybe the biggest promise between Google's workspace, ai, as well as Microsoft copilot. While these two companies has not released anything like this yet, OpenAI is coming after this world as well in the enterprise AI solutions. Now, the other very interesting point here is that they actually started with Google Drive integration versus the Microsoft integration, which is really surprising based on their connections with Microsoft. That being said, they already announced that Microsoft SharePoint and Box are coming next. So OpenAI is going after all the different major providers of data storage in the enterprise world. They're also promising obviously, that your data stays intact and that chat PT won't peek into files or train on your data or use any of that for any of their needs. But it's unclear how long the data is stored and where it exactly stored, even though they're clearly stating that the data is going to be encrypted. So where is this going? It's going to hopefully finally having a future where you can ask the AI question about something in your company, and you'll be able to get answers across all the different pieces of data, regardless of where they live, which is a huge benefit to anybody in any kind of business because any business large and small, and obviously large even more, has silos of data that are not connected. And to get that information today, you need to have multiple people prepare for a meeting and have and share that information in a meeting, and maybe all of that will become a thing of the past in the next few months. Now some of the downsides of this new solution that, again, is just being tested right now. It doesn't know how to analyze images yet. It doesn't know how to look at Slack dms yet, and it only reads versus crunches the data in spreadsheets. So a few small limitations. I assume these will go away as well, or at least be able to define those in some kind of settings in the future. But right now these are limitations of this solution. Open air also unveiled that they have a new, groundbreaking approach to AI safety, which is monitoring the chain of thought of the reasoning models. So if you think about how reasoning models work, they reason in English, so we can actually watch or monitor what they're actually doing step by step when they're doing their process. And using this new capability, OpenAI was able to catch the models cheating, as an example. Faking coding test results and deceiving users by providing misleading information on how they achieve different results. So on one hand it's really promising to know that these companies are looking for ways to monitor the AI and what it's doing and how it's doing it to hopefully be able to contain and understand what the AI is doing when it's becomes more and more powerful. On the other hand, it's very, very clear that these AI tools are showing human-like behaviors, protecting themselves and protecting their own work, including by deceiving the human operators on what the goals and exactly what they're doing. That to me is very, very scary because right now we feel that we are smarter than these models, but it's very clear that in the current trajectory, in a few years, there are going to be significantly smarter than us, and then their ability to deceive us is going to be significantly higher, and I'm not sure we'll be able to catch them at that point. So think about if a 3-year-old will try to deceive you with something, their chances of being successful are very low. Well, right now we are the adult and the AI is the three years old. But in a few years, the relationship in the brain capacity is gonna be flip flopped, and then I think it's gonna be very questionable on our ability to actually understand what they're doing and to definitely control what they're doing and how they're doing it. Again, my personal concern, but I think it's something that we have to think about globally and understand how we approach it while in parallel to developing better models and hopefully faster than we're developing these models. And the last piece of news related to OpenAI is that OpenAI has agreed together with Elon Musk on an expedited trial timeline set for December, 2025. So in a joint court filing on March 14th, both sides proposed speeding up the trial in which Elon Musk is trying to prevent OpenAI from converting from a nonprofit to a for-profit organization. The judge has rejected Musk bid to stop the process right now, but they both agreed to accelerate the trial to the end of this year. So we will still have to wait almost nine months to just the beginning of the trial, but at least it's not going to drag for years before we actually see the outcome of that. And from Open AI to Anthropic, in the past few weeks have been releasing more and more capabilities, and this week they have released two very important capabilities. One is a new think tool that was just unveiled on March 20th that gives Claude the ability to think during processes and not just before the processes. And they're claiming that using the Think tool is boosting its agentic capabilities by 54% compared to not using it in the same use cases. Philanthropic already introduced what they called extended thinking mode before that, and the difference between extended thinking and just the think tool is that the extended thinking focuses mostly pre-task. So the tool is analyzing the task ahead, trying to understand what exactly needs to be done, and then it runs and executes on it. The think tool allows the tool to stop after specific different steps, such as getting data from external tools and analyzing whether it has complete information, whether it has the right information before continuing to the next step. What they're claiming is that this approach allows it to achieve significantly better results than previous models and previous approaches. And it excels in policy heavy tasks such as different rules and regulations in different industries, sequential decisions, so things that require step by step processing and cutting errors in things like customer service and long workflows because of its ability to stop and evaluate at the end of every single step of the process. Now, this is already available through the Claude API and it requires very simple setup in order to use with new or existing applications of the Claude API. Philanthropic also announced they're working on a voice capability and they're actually broader than that. They're planning to release at late 2025, a combination of voice control and machine control. So their chief product officer, Mike Krieger, has shared with the Financial Times, and now I'm quoting, we are doing some work around how cloud for desktop evolves. If it is going to be operating your computer, a more natural user interface might be to speak to it. And they're claiming that prototypes are already in play. So we talked about this many times in the past. I think keyboard and mouse are something that we're going to use less and less as we engage with computers because large language models will allow us to communicate in much more natural ways. And it shouldn't be a surprise to anyone that Claude, that already shared computer control capabilities last year are going to integrate that with voice. They're pretty late to the voice game cause Gem and I and Chat G PT both now have very capable voice capabilities, so it's only natural for them to, A, introduce voice and B, introduce it in combination with computer control capabilities sometime later this year. I shared with you in the past, until there are clear ways to safeguard what the AI can control on the computer and what it cannot, I see this as a very risky move, especially on the enterprise level, but the direction is clear and I think we're gonna start seeing more and more of that moving forward. But the biggest news from Anthropic, which I was waiting for a very long time, and you heard me complain about this and bitch about this many times in the past, is Claude finally has internet access. So the biggest issue I had with Claude, which I actually really is that it couldn't access the web. You couldn't give it links to anything. It couldn't search for relevant information in real time. All it knew is everything that was in its training data and with a cutoff date, and now they finally aligned on that particular topic with the other leading models, more or less, every other model on the planet already had web access and realtime data access, so you can now have it on Claude if you are a paying user in the us. All you have to do is go to your settings and turn it on. It's not on by default. I already turned it on for me and I actually used it in preparing for this news episode. And I really like the way it works. You give it a topic, it does a quick research, it brings back very well researched and summarized output, and it does it relatively quickly with citations that showing you where the content was brought from. So you can go and check it and evaluate that it is credible sources. What it still does not have is a deep research mode that many of the other platforms now have, but I assume this is coming next. Now that they have the web search capability, as I mentioned, very promising, start very late to the game. And from anthropic to XI. So Elon Musk's XI just acquire Hotshot, which is a San Francisco based startup specializing in AI powered video generation. If you remember earlier this year, Musk tease that Grok video mode is coming in just a few months. Well, this might be a way to get there faster by acquiring a company that already knows what they are doing. When exactly will it be integrated into grok? It's unclear, but knowing the speeds that Grok has done everything so far, I'll be really surprised if we don't see something before the end of this year. In a very interesting piece of news from Xai and Elon is the announcement that Xai has joined forces with Microsoft BlackRock and UAE based MGX in a$30 billion project to build AI data centers for XAI. Why is that? Very interesting. Well, it's interesting, first of all, that there is a new partnership for XAI to grow even faster than they're growing right now. The new partnership is simply Dub the AI Infrastructure Partnership or a IP. But the fact that Microsoft, who is the main backer of open AI is teaming with xai to develop data centers for Xai is surprising. And it's just another step in this really weird frenemy relationship between Microsoft and OpenAI. I shared with you just in this episode that open AI's new tool starts with working with Google and not with working with Microsoft. We know that OpenAI is now using data centers that is not just coming from Microsoft in their partnerships, so this is just another step from the Microsoft side in this particular case, showing that the relationship between OpenAI and Microsoft that used to be very tight is not that tight anymore, and that both companies are looking for alternatives and diversification in order to stay ahead of the other. As we all know, microsoft is also developing its own internal tools that will probably over time replace the open AI tools that are currently running all the AI capabilities within the Microsoft universe. And from X AI to Google. Google, as I mentioned in previous episodes, are on fire in the past few months, potentially being the most dominant AI platform in the world right now, or at least becoming the most dominant platform in the world right now. But Google DeepMind, in this particular case just unveiled Gemini Robotics, which is an AI model based on Gemini two, that is supposed to work in the physical world and help other companies develop robotic solutions based on this infrastructure. So they are rolling out several different capabilities under this platform and the goal is to make robots smarter, nimble, and more human like by using this infrastructure and solution from Google that can be integrated into anybody who's developing robotics solutions. And the specific quote from Carolina Parata, the senior director at Google DeepMind was Gemini Robotics draws from Gemini's multimodal world understanding and transfers it into the real world by adding physical actions as a new modality. Another cool feature from Google came from Notebook lm, so Notebook LM is a great tool that I now use regularly. That is awesome at summarizing lots of information, as well as many podcasts by voice that summarizes everything from multiple data sources all combined into one. they just added the capability to create mind maps out of that data. So there's a mind map button that you can press after you upload your data, and it will show you the relationships in a graphics interface, a mind map to what is the data showing. I actually really like using mind maps when I plan new big projects. I've never tried to reverse engineer the process and actually start with the data to create a mind map, but it will be interesting to try out. But if you do have a lot of data and you try to understand the relationships between all the different components in that data from multiple sources, this is a cool new functionality from Notebook lm. But the biggest news from Gemini this week is that on March 18. Google released Google Gemini Canvas, which is copying what OpenAI has in ChatGPT Canvas. It's the ability to work together in a collaborative environment with the AI through highlighting specific sections relating to those specific sections, asking it to edit specific things or to give you comments on specific aspects, making it longer, making it shorter. Very similar to what ChatGPT has in its platform. I absolutely love Canvas in ChatGPT. It one of the biggest benefits that ChatGPT had over Gemini, and now they don't have that benefit anymore. On a full review that was done by Ryan Morrison from Tom's Guide. He actually says that he now prefers Google's Gemini's canvas over Chachi piti and that he was able to do things that he was completely impressed by, including its ability to review a complete chapter of his novel without any prompting, and get very good feedback, as well as brainstorming the next chapter of the novel based on the first chapter. So what does that mean to you? It means, first of all, you can now use Canvas features in Gemini. It is available right now as a button in the Gemini prompt line, and you can just click on it and go to the canvas mode and work with it in a collaborative environment. This shows that there is really no moats in the AI race and literally almost anything one company will do, another company will copy and probably do a little better shortly after. So if you remember, ChatGPT Canvas came after Claude had Artifacts, which was the first company that did this side-by-side collaborative solution, open AI's Canvas was significantly better than artifacts, at least in some things. And now Gemini seems to be better than open ai and they'll probably keep on upping each other on what are the capabilities of these tools, both in text, as well as coding capabilities, which both Excel at now while Gemini is copying things and competing with ChatGPT? ChatGPT just announced that their new beta assistant can now power Android as the assistant on the phones replacing the just newly announced Gemini ability to control your phone instead of the old Google Assistant. So when you go to your settings right now on an Android phone, you can choose chat PT as the assistant instead of the old assistant. And in lieu of the Gemini new solution that will presumably replace the old assistant on all Google phones later this year. Now, it's still not as well integrated into the Google phones as the Google Gemini app, which makes perfect sense. There's also no, hey, ChatGPT versus Hey Google, that you can do on the Google phones right now natively because Google is blocking that at this point. Now, do I think that ChatGPT will be able to do a better job in controlling Google phones than Gemini? No, I don't think so. I do think that over time Google will develop more and more integrated capabilities that only they will be able to do because of their control of the actual operating system of the phone. But it's another move to show you how multifaceted this race is and how many aspects of our lives it actually impacts across every interface that we have, from computers to systems, to our handheld devices, et cetera, et cetera, et cetera. And as I mentioned in previous episodes, don't be surprised if very soon you'll be able to talk to your microwave or other day-to-day tools you use in your house, in your natural voice, and you will understand what you need and we'll do the relevant setups for the output that you need. And speaking of the all out race between the big companies, across everything, AI, meta just kicked off its first in-house AI training chip. Now they've actually tested a similar approach before, back two years ago, and that failed miserably and that forced them to spend$2 billion on Nvidia GPU to cover for their failure. But it seems that they're on a. Different path in this particular point, and they're planning to build and rely over time on their own chips versus Nvidia for at least some of their AI needs. And speaking on Mera y Laun me's, chief AI scientist, just shared his longstanding view, but just shared again his view about large language models and their limitations, calling them just token generators. And he restated what he's been saying for a very long time now he believes that large language models are not the path to a GI and he's betting on the next generation AI architecture that is mimicking the way humans perform, focusing on four different things, grasp off the physical world, hold persistent memory planning and reasoning. Again, this is not new from Jan Laun. Just a reminder that he does not believe that LLMs can lead to a GI without all these other capabilities. Now in parallel to this, if you think about it, meta hasn't released an updated LAMA model for a while now, despite everybody else releasing new models and new capabilities, which may hint to the fact that maybe meta and LeCun are actually focusing on something else that may release when it's ready. That will be different than a large language model. And for the last piece of news today we're gonna talk about when we are going to get a GI, at least based on Demi, ve, the CEO of Google DeepMind, and a Noble Price winner. So in a London briefing this week, he anticipates that we will achieve a GI in five to 10 years. He admits that today's AI is very passive and lacks many capabilities that are needed to surpass human capabilities across the board, but he believes that in the next five to 10 years we'll be able to bridge those gaps. Now, well that's his predictions. Many other leaders, including Sam Altman and Dario Amide believe that a GI is coming within the next two to three years. Cisco's Giro Patel Bates on this year. And I've shared with you several different articles in the past few weeks that several other people thinking that a GI is imminent, who is correct, is very hard to tell. What I think, I think it doesn't matter. And the reason I think it doesn't matter, cause I think a GI is not aligned. It's not a point where we say, oh, now we have a GI. And AI just progressing in a very high speed, doing more and more things better than humans. And as it takes over another thing in which it's better than humans, it's gonna have significant profound impact on everything that particular field touches on. And then it's gonna take another field and another field. So the a GI definition itself, even if some people can agree on what it really means and most people cannot, is not relevant for what we actually see as the implications of that in the real world. We will have to learn how to live alongside with AI across everything that we do and figure out the implications of it on our lives, including literally everything we know. That's it for this week. If you enjoy this podcast, please share it with other people who can benefit from it. Open your phone right now. I know. Yes, right now unless you're driving and because otherwise you'll forget and I know you want to do this, so please open your phone, click on the share button and send it to a few people you think you can benefit from it. While you're at it, I would really appreciate it if you can leave us a review on Apple Podcast or Spotify or other favorite platforms and connect with me on LinkedIn, ISAR Metis, and tell me what you think about the podcast, what you think I can do better. I really appreciate your feedback. A lot of you write to me anyway, even without me asking, so maybe if I'll ask. We'll get even more feedback. On Tuesday, we'll be back with another fascinating how to episode, where we're gonna deep dive into how to use AI for specific use cases. And until then, enjoy the rest of your weekend.

People on this episode

Head Shot

Isar Meitis

Host