Leveraging AI

131 | The most significant product ever created", per Elon Musk, Enterprise AI implementation is accelerating, and bad players are using AI in ways you need to know, and other AI news for the week ending on October 11, 2024

• Isar Meitis • Season 1 • Episode 131

Share episode

Is Tesla Really Ready to Revolutionize Transportation? The Latest from WeRobot 2024 and a Shocking AI Phishing Scam

In this packed episode, Isar Meitis breaks down Tesla’s big announcements from their recent WeRobot 2024 event. From fully autonomous taxis to robots that can walk your dog, this episode is bursting with insights on AI’s rapidly evolving role in business and everyday life. But it’s not all shiny new tech—get the details on a chilling, sophisticated phishing attempt using AI that could fool even the most tech-savvy professionals.

So, is this the dawn of a new era or more tech promises with missed deadlines? Tune in to find out—and don’t miss the crucial cybersecurity advice at the end!

In this episode, you’ll discover:

  • Tesla’s bold new announcements: autonomous “cyber cabs” and humanoid robots
  • How Tesla’s vision compares to existing self-driving services like Waymo
  • The real-world timeline for Tesla’s fully autonomous vehicles—will they meet the 2026 production goal?
  • Why investors might not be impressed (and how Uber and Lyft could benefit)
  • A groundbreaking partnership between Accenture and NVIDIA in the AI enterprise space
  • A sophisticated AI-powered phishing attack you need to know about—learn how to stay safe

Join us in the upcoming webinar on October 17 and we'll talk about: AI Planning for 2025: A Blueprint for Successful AI Implementation - https://services.multiplai.ai/ai-webinar 

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!

Hello and welcome to a Weekend News episode of the Leveraging AI Podcast, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. This Isar Metis, your host, and we've got a pact. Episode for you today with a lot of exciting news. Some of them are coming from the robotics world and the self driving world with some very exciting news, a lot of news on video generation, some big partnerships between some of the giants, some of the biggest companies in the world, as well as a really scary, sophisticated AI based phishing attempt, which we're going to get to in the end. And I promise you, you want to wait to the end to learn about this so you can keep yourself safe. So let's get started with this week's AI news. Last night on October 10th of 2024, Tesla held their WeRobot conference. Announcement event. That was a very interesting event in which they announced several different things. The first thing they announced was the cyber cab, which is a fully autonomous, basically, taxi that can take people from one place to the other. That has been the core feature of the event. And I think that everybody was expecting to see. The car itself looks like a mix between a model three and a subvert track, with a really futuristic look, it has going Lambo style doors that opens upwards. It has no controls whatsoever, meaning there's no steering wheels. There's no pedals. So there's no way for a human to control these cars. They're built to be completely autonomous and Elon Musk is claiming that full self driving capabilities. All of Tesla's are going to start being available for motor three and model Y to be able to drive completely autonomously in Texas and California as early as next year. But these cyber cabs will start production in 2026. They're also shared a new robo van, which is a really spacious full size van for 20 people that also has no controls that is supposed to be for mass transportation and can take people around, they're claiming that this will be able to take people at a cost of five to 10 cents per mile, which is really cheap. It looks very futuristic. It has no. windshield in the front or back because there's no driver in it that needs to see through the windshield and As I mentioned, it has no steering wheel, no pedals, and so on. This is currently just a concept, they haven't shared even if that's even planned for production yet, but as I mentioned, the cyber cab is supposed to start production in 2026 that needs to be taken with a grain of salt because Elon Musk has promised many things in the past on timelines that they couldn't even get close to. If you remember 2019, Elon said that by 2020, we're going to get this robot taxi that now is called cyber cab. That being said, a lot happened since and Tesla has full self driving capabilities on its cars that are driving tens of millions of miles every single day, more than any other company in the world. So they definitely have the data and the experience to build something like this right now. So maybe now the timeline is more realistic and maybe we'll start seeing cyber cabs driving around in 2026 or maybe 2027 because production is supposed to start in 2026. Now, as you probably know, Waymo, a Google company, already has self driving taxis in several different cities in the U. S. I got to drive one two weeks ago as I was speaking at AI Realize in San Francisco, and it was an amazing experience. But these are traditional cars, Jaguars, with taxis. Steering wheels and pedals and so on just without a driver to operate them versus the Tesla ones that are supposed to be built to be these cyber taxis. And in addition, they're supposed to be significantly cheaper than the way more cars, the way more cars cost over 200, 000 per car. And these cars are supposed to be in the tens of thousands of dollars to manufacture and hence significantly cheaper to buy, operate, and maintain over time. Now, what does that have to do with AI? Well, obviously AI operates this entire car and the fleet and everything else in order to make this. possible. But in addition, they have shown significant improvement on their Optimus robots. So these are humanoid robots. This is the surge. This is their newest generation of that robot. and per Elon Musk, and I'm quoting, they can do anything that you can. So he mentioned things like walking dogs, babysitting, mowing lawns and serving drinks in a party. They actually had these robots at the event serving drinks and engaging with people. So they didn't put them behind a fence for people to watch. They were actually walking around serving drinks and having conversations. So this is coming, current prototype can actually engage with people right now. The interesting parameter here is that he's claiming again, take this with whatever filters you want based on Elon's previous promises. But he's saying that the price tag of these robots once they get to full production is going to be between 20 and 30, 000 a unit. That's extremely low for a highly capable humanoid robot that can either work in a factory, replace employees across multiple industries from cleaning our streets to working hospitality, but all the way to a home usage. If you can have a robot for 20, 000 to 30, 000, doing the dishes, walking the dog, mowing your lawn, taking care of your pool, which I'm not sure is a good idea. And he actually didn't mention that. I'm just wondering if they're water resistant and a lot of other things. This may be a fair comparison to having multiple people. Somebody To clean your house, somebody to take care of your pool, somebody to mow your lawn, somebody to take care of other stuff in your house. If you hiring these people right now, and a lot of people in the U S and around the world do that right now, that's a comparable cost that in this case, you have to spend once, and then you can use it every single day to do all these things. He is claiming. That basically, and I'm quoting again, everyone in the 8 billion people on earth will want an optimist buddy. I'm not a hundred percent sure of that. There's a lot of people in the 8 billion people on earth that don't have electricity yet or fresh drinking water. But I do agree with him that 20 to 30, 000 for a device, a robot that can do many aspects of helping us in our day to day life is an attractive price point that if they can get to, it will sell more than they're selling Teslas right now. Elon says, and I'm quoting again, the most significant product ever created. So that's Elon being Elon. I'm sure there's a few things in history that we can think of that might be more exciting than a robot that can do everything, but it's definitely high on the list. And if the price point will be reasonable. I definitely see that as a huge potential success. There's obviously other companies who manufacture these kinds of robots right now, and there's going to be fierce competition. So even if 8 billion people do want these kinds of robots, not all of them are going to be manufactured by Tesla. There's a lot of other companies who are in that race. Tesla is definitely one of the front runners. The interesting thing from my perspective is despite the fact that this promises to be a huge success. New product in a huge new market for Tesla that did not excite investors in Tesla. And the stock fell down 8 percent today. The interesting thing is at the same time, Uber and Lyft stocks went up. 8%. Now, I wonder if that's because people are really disappointed with what Tesla was showing or because Uber and Lyft investors see a future where Uber and Lyft as companies will use cars like the cyber cab and maybe even the van from Tesla. So they won't have to pay drivers. They will be a lot more efficient. They can run 24 seven. And they will make a lot more money for Uber and Lyft. So I'm not sure what investors are thinking, obviously, but it's, I just found this fact to be really interesting. By the way, in my personal opinion, I think Tesla will have an amazing product with their robot. And I do think over time it will generate a lot of value to Tesla as a company and to the people are going to buy these robots. I'm obviously not making any stock buying or selling recommendations. I'm just sharing with you my personal opinion. Another interesting piece of news from Elon Musk this week that has to do with AI, XAI, which is another Elon Musk company, has held a recruiting company at OpenAI's former headquarters in San Francisco, which they recently moved into. So OpenAI left to a different location, Musk bought the building. And has run a recruiting party during OpenAI's Dev Day event on the other side of town. So that's Elon being Elon. We know he has a lot of beef with Sam Altman and OpenAI as a whole. In that event, Musk has made some interesting predictions or statements, which with Musk, they're more statements than predictions. But the first one is that XAI will achieve AGI within a couple of years. The second one is that they are aspiring for XAI to be as dominant in the AI space as SpaceX is in rockets. Now SpaceX is the world leader by a huge spread in rockets. I don't think it's realistic for them to take the same position in AI, especially that they're coming from behind. But that being said, they just last week finally They just last week had all of their 100, 000 GPUs in their mega computer online for the first time. So they've built it at record speed. In about four months they got it to start working. They were running it with 60, 000 GPUs initially. And last week they finally achieved the whole 100, 000 running at the same time. So they have right now the most capable AI training computer on the planet by a big spread. And Elon being Elon, and especially when he has the drive of his beef with OpenAI. I wouldn't underestimate them in their capability to develop extremely capable models, and they're already proven that they can progress and make significantly better models in a relatively short amount of time between the different versions of Grok so far. where does that actually put us? I'm not 100 percent sure, but it definitely seems that XAI is going to be one of the front runners. Together with the other companies that Musk himself has identified as their biggest competitors, OpenAI, Anthropic. And Google that he assumes together with X are going to be the biggest players in the future of AI. Now from X to another giant that is partnering with a third giant, Accenture and NVIDIA has announced a partnership with Google. For enterprise AI adoption. So you heard me say that many times in the show before that successful AI implementation is not just about technology. It's about having the right strategy, having the right training, having the right infrastructure in place, having a plan on how to deploy AI. When one of the best companies in the world to build. Enterprise level planning and execution is Accenture. They've been doing it for decades across multiple enterprises around the world, and they definitely have the experience in doing that. And so they've now announced a partnership with NVIDIA. To create a new business group within Accenture that will launch Accenture AI refinery platform using the full NVIDIA AI stack. And when I say the AI stack, it's software and hardware. And their goal is to build this network of refinery engineering hubs around the world. Each one will support enterprises in its region. Now, as you can expect from two companies at that scale, Accenture has identified 30, 000 professionals will be a part of this team that will put this process and this project together. And the focus is obviously helping clients build AI systems, the more specifically agentic AI systems. So the next generation of AI. Of AI agents and implementing them successfully and effectively in safely within businesses, another interesting reference point that came out of this article and out of this release. Is that Accenture already has 3 billion in bookings for AI assistance this year, meaning they're going to finish the year of probably close to 4 billion in revenue just through AI development and implementation channels. And this partnership with NVIDIA is most likely going to accelerate that. That's showing you just how much need there is right now at the enterprise level or a I solutions now they're going to use, as I mentioned, all of NVIDIA's platforms that includes a I foundry NVIDIA's a I for enterprise and NVIDIA omniverse, and it's going to be using both public and private cloud platforms and integrating both of them together to achieve the best result tailored for the needs of each company. press release, they share they already have existing hubs in Mountain View, California and in Bangalore, India. And then the next hub they're planning are in Singapore, Tokyo, Malaga, and London. So global presence from two global companies that will most likely make serious waves and help a lot of enterprises adopt AI in an efficient way. Taking that to my little universe, I definitely see that companies, even the ones that seem to have a better grasp on AI, and I'm not working with the same scale of companies that Accenture does, but I work with small and medium businesses a lot on AI implementation and training. And I see that having a plan combined with having a leadership that is bought into the idea of AI implementation generates amazing results very quickly for multiple organizations that I'm working with compared with companies that are trying to figure out themselves Not knowing exactly how to do this because they're trying it for the first time. I definitely see this approach as the right way forward, especially building these centers of excellence that will develop best practices, as well as training for employees, which plays a very big role in their plan and in the way that I do things with my clients. And I see this kind of training as one of the most significant aspects that can predict future AI implementation success. We have been talking a lot on this podcast, on the importance of AI education and literacy for people in businesses. It is literally the number one factor of success versus failure when implementing AI in the business. It's actually not the tech, it's the ability to train people and get them to the level of knowledge they need in order to use AI in specific use cases. Use cases successfully, hence generating positive ROI. The biggest question is how do you train yourself? If you're the business person or people in your team, in your company, in the most effective way. I have two pieces of very exciting news for you. Number one is that I have been teaching the AI business transformation course since April of last year. I have been teaching it two times a month, every month, since the beginning of the year, and once a month, all of last year, hundreds of business people and businesses are transforming their way they're doing business because based on the information they've learned in this course. I mostly teach this course privately, meaning organizations and companies hire me to teach just their people. And about once a quarter, we do a publicly available horse. Well, this once a quarter is happening again. So on October 28th of this month, we are opening another course to the public where anyone can join the courses for sessions online, two hours each. So four weeks, two hours every single week with me. Live as an instructor with one hour a week in addition for you to come and ask questions in between based on the homework or things you learn or things you didn't understand. It's a very detailed, comprehensive course. So we'll take you from wherever you are in your journey right now to a level where you understand. What this technology can do for your business across multiple aspects and departments, including a detailed blueprint of how to move forward and implement this from a company wide perspective. So if you are looking to dramatically impact the way you are using AI or your company or your department is using this is an amazing opportunity for you to accelerate your knowledge and start implementing AI. In everything you're doing in your business, you can find the link in the show notes. So you can, you just open your phone right now, find the link to the course, click on it, and you can sign up right now. And now back to the episode. And speaking of enterprise and the impact of AI on enterprises, Forbes shared a very interesting article this past week that is showing the status. Of AI in enterprise companies. So the first and not surprising parameters that 96% of boardrooms have generative AI on their agenda for the decisions that they're making for Q4 and for next year. Speaking of that, by the way, we are holding an AI planning for 2025 webinar. On October 17th at noon Eastern, if you haven't started planning or if you're planning and looking how to complete your plan for 2025 and AI implementation in your company, in your business, or for your personal career, Come join us. There's a link on the show notes. So you can open them right now. So you don't forget and sign up for our webinar, but I'm continuing. As I mentioned, 96 percent of boardrooms have that on their agenda. Adoption in large company has increased this year from 6 percent to 24%. So it's still very low. You're not. Very far behind. If you haven't started, there's less than a quarter of companies that actually started implementing AI. Even this year, current ROI in large enterprises is currently still very low between four and 7%, but most companies are going to continue investing in that, and they see that as an early stage improvement. process with a huge potential in the long run. I'm claiming going to, going back to what I said before, that most companies do not see huge ROIs because they don't have the right plan in place. They don't have the right training in place. They do not introduce specific use cases to specific employees in order to really rip the benefits off that. So the main challenges that this article sees. Our platform enablement. So choosing the right technology and building it around the ecosystem of the tech stack, a company already has governance, meaning managing the different models, security, ethics, and alignment with company culture, identifying appropriate use cases. As I mentioned, that's a huge deal. Like you're going to get the benefits if you start by identifying the use cases and not by buying the technology first and then looking for use cases. That's the wrong. Way to do this. And there's the right way to also identify and prioritize the use cases, training and skill development of the workforce. I mentioned that before, that's a huge issue right now where companies are investing millions and sometimes tens of millions or hundreds of millions of dollars in licenses without proper training for the workforce. And the solution should be exactly the other way around. Train the right people on the right tools as a first step, and then grow from there. The amount of licenses after these people figure out the use cases and figure out the right implementation, and you will see significantly bigger benefits much faster. And then the last aspect is responsible use and cost management. Of AI. So these are the limitations, the future trends that are identified is diversification of AI models. So a mix of closed source, open source, large models, and small models across multiple tasks within the organization. That's the direction everybody is going. I've seen that. Firsthand, as I was speaking last week at AI realize a conference in San Francisco with multiple fortune 2000 leaders, as well as multiple leaders in the AI implementation industry that share that. So that's definitely the direction it is going hybrid ecosystem, so different types of AI technologies and the rise of agents. So instead of a chat bot. Different kinds of bots to a self thinking, decision making, action taking products that are AI based and that's what agents basically mean. Now the key applications companies are implementing right now. One is everyday AI. So enhancing existing applications and existing products. Processes. The other is integrating AI into their products and services as part of their offering to their clients. And the third is AI in their innovation journey. So trying to use AI to change the trajectory of the company and of the strategy of the overall company, which will take them to a different path that is AI enabled, which will put them ahead of their competitors who are not doing the same thing. Through the interviews they've done with leaders of these enterprises, two things are very clear. One is that there's no magic formula. Each company has to find its path based on specific concepts, but there's no one size fits all as far as AI implementation. And the other is, That generative AI should be one of the top priority for any company. If they want to stay competitive in regardless what industry or size of business it is, and one of the top things, all those leaders said is the importance of preparation. The first thing that they mentioned is that training employees on their AI technical skills is a critical aspect. If you're not doing this right now, if you don't know how to do this, please reach out to me on LinkedIn or visit our website or find the link in this post to learn about our education programs that we provide. Tailored to companies or open to the public. We have all of the above. We have free training resources as well on our website. So just find the links in the show notes and find them. This is going to dramatically impact how your company can compete in the future. The other two things that these CEOs and leaders said is developing strategies. For cost and resource management is a big thing that will impact the success as well as establishing clear guidelines and oversight. And that's another thing that I'm working with my clients on very early on in order to make sure nothing bad happens from reckless AI implementation. Okay, so we talked a lot about enterprises and big players. Another big player that we don't talk about a lot and that plays a huge role in the enterprise level is Zoom. So zoom, the digital conferencing platform, held their Zoomtopia conference this past week, and they've made a huge, and they made a lot of AI announcement, which is not surprising, but they are adding more and more AI models and more and more capabilities across their ecosystem. So a co pilot that's going to be available for every Zoom meeting as a right panel that also has web access that can get access to real time information It's a feature to look up things and help you find information as you're holding the meeting in conjunction with summarizing the meeting and key things that are happening, following action items that were defined by different people in the meeting and summarizing those as well. And they're saying that this feature will be available at no additional cost to paying users of Zoom. That is expected to be available shortly. They're also adding a higher tier that's going to cost 12 a month per user in H1 of 2025 that will have what they call an AI studio for customizations of AI capabilities and company specific data sources that can be integrated into this new zoom AI ecosystem. Including integrations with known third party ops like Atlassian and Workday and Zendesk and so on. And companies who pay this will be able to personalize coaching and custom avatars. For zoom meetings and interactions that will participate and be able to assist in specific type of zoom meetings. They're also announced AI powered task detection, recommendation and completion that will be able to track again, things that people are talking about as far as what's happening in the real life and connected back to the digital world. They're adding zoom phone enhancement, so you'll be able to record and transcribe and summarize. regular phone calls versus just zoom meetings, and it's going to be connected to their entire ecosystem. And they're adding AI capabilities to their docs and data management platform that they've announced a while back. So they're not just going after conferencing. They're also trying to get market share from companies like Microsoft and Google when it comes to documentation and data management within the company. They're also adding something that I have been using Zoom for a while, but they're adding it as an easy way to do this, a way to track in person meetings and use Zoom for those as well. So you can bring your AI companion to in person meetings to summarize them and take notes and have it integrated with your entire office. ecosystem. They're also planning to add AI virtual voice agents for self service calls, so basically allowing your customers to talk to an agent through a Zoom interface that is tailored to your company's needs and quality management of those agents to track how well they're doing so they can improve over time. And they're also offering the ability to tailor solutions to specific industries, such as healthcare, education, and so on. So, big announcements coming from Zoom, going back to something that we talked about many times on this show. In order to be successful in the AI world, you need the technology, you need the data. And you need the distribution and zoom has all three. They have calls data, especially recordings from millions of calls, probably every single day that they are saving and analyzing and so on for multiple companies, they have the distribution because so many people are using zoom and they have the technology capabilities to develop these AI capabilities on top of their existing ecosystem. So I definitely see them moving in the right direction. That being said, I do not see them winning market share because of that from Microsoft or Google replacing. Docs and stuff like that. But I do see them going after new clients and providing them a lot more capabilities and charging them for that, that will allow them to provide more value and generate more money. So that's it from zoom. And now let's talk a little bit about open AI. I know you're thinking to yourself, Oh my God, it's been so long since the beginning of this episode, and we haven't spoken about open AI yet. So here we go. Let's talk a little bit about open AI. About a week and a half ago, they had their dev day early in October, and they've shared a lot of really interesting things. One of them I shared with you last week, which is the Realtime API, which allows a low latency API that allows you to generate voice responses and integrate them into your data and into your processes using that API. They also announced the capability to do prompt caching, which can dramatically reduce costs and improve latency while using the They've announced the capability to do fine tuning of images and text for GPT 4. 0. And they've also announced the ability to do model distillation, which means you can use their big models in order to train and improve the performance of smaller models. This comes obviously at the same time that they've raised 6. 6 billion dollars plus 4 billion in a line of credit, plus having some of their top talent and top leaders, including some co founders leave the company. The latest one that added to the list this week is the guy that was running the development of Sora. So those of you who don't know, Sora is the incredible. Video generation platform that they presented earlier this year. I think it was February, but they've never released that. Triggered a crazy race of video advancements that I'm going to share some of them later on in this episode, but they never actually released. Sora to the public. Nobody knows exactly why. And now the guy that was leading Sora is leaving the company. to go and do something similar at Google's DeepMind. Is that good news? No, I think none of these people living OpenAI is good news. I think it has a lot to do with leadership issues and leadership style within OpenAI. Not sure this particular departure has to do something with that, but I'm sure that everything that's going there, including their attempt to transition to a for profit organization, are contributing to this exodus out of OpenAI of some of their top talent. Now, speaking of their transition to a for profit organization, which by the way, they have to do in order to keep the money that they just raised. So The 6. 6 billion dollars they raise is contingent of them becoming a for profit company. There might be several different things that will stop this, or at least slow this process from happening. the California Attorney General and the IRS might be some of them because they need to ensure that charitable assets, because they're a non profit organization, are actually used properly. And if they're changing for a for profit organization, then all the quote, unquote, Charitable assets that were placed with them are now misused. That is problem number one. Problem number two is obviously internal disagreement. As we know, a lot of people are leaving. A lot of people are unhappy with this. They have made significant changes in their board to potentially allow that, but that doesn't mean that everybody in the board would agree with that. And that doesn't mean that previous board members will not go and place lawsuits. Saying that the company and that transition is breaching the company's mission, how it was established. So I don't think it's going to be a straight path forward. That being said, I assume since some of the biggest companies in the world, as far as investors just made a 6. 6 billion bet that this will be successful. I assume it will be successful, and there are several legal experts that believe that this is possible within the two year time frame. Now, speaking of OpenAI and Sam Altman transitioning the company to a for profit organization, Geoffrey Hinton, who's considered the godfather of AI, just received the Nobel Prize for As part of his address of his winning this award, he praised one of his former students, Ilya Saskovar, for his role at the removal of Sam Altman from OpenAI. So if you've been listening to this podcast or following what's happening in the AI world, you know that late last year, Ilya was one of the people on the board and the one that triggered the firing of Sam Altman from OpenAI. Eventually Sam Altman came back to OpenAI and Ilya left to start a competing company. Hinton has been criticizing Sam Altman specifically and OpenAI in general, stating that Sam has been prioritizing profits over safety concerns when it comes to AI development. Hinton believes that Altman's approach is really dangerous, and I'm quoting, he's saying, he used the word unfortunate. And he has continued to praise saskiverse foresight and recognizing both his potential of his new business as well as the fact that he is very clearly identifying the dangers in developing advanced AI systems. And from OpenAI to Google, Google just integrated their Imagine 3 text to image generator into Gemini. So if you remember early this year, they released those capabilities combined together. There was a huge backlash because the model was generating, what some people called woke images. So Founding Godfathers are having different races, Black Nazis, and so on, so they took it off. It was not available through Gemini for a very long time, and through the Imagine three standalone. You could not generate any images of people. So now they're bringing it back. Paid users of Gemini will be able to now generate images using the Imagine three image generator within the tool itself. It's currently only creates square format images. You can. generate images of people, but not of known people, meaning you can define what people does in your image, but you cannot make them look like specific people in the real world, at least for now. The images themselves are actually pretty good and it's More or less comparable with Mid Journey and Flux as far as generations. I still prefer, after seeing examples, and I didn't get a chance to deep dive into this, but after seeing a lot of examples, Mid Journey and Flux are still ahead, but it's a very close contender just behind them. The only annoying thing with the Imagine 3 engine, and I've seen that myself multiple times in the past, as well as this past week, is that every now and then, way more frequently than any other model, it will refuse to generate the images that you're asking it to create, saying it's not aligned with the limitation that was put on that particular platform. So every now and then, again, at a higher frequency than any other platform, it would just not generate images for you. Sometimes in situations that make absolutely no sense. Staying with Google, Liz Reed, Google's head of search, claims that Google has a big edge on other competitors when it comes to AI in general and AI in search specifically, because of its extensive data on Web content as well as beyond Web content such as maps and sports and products and Android and so on. So the amount of content they have is significantly bigger than any other company on the planet. And the fact that they have expertise in combining technology information and user experience. I agree with all of the above. You heard me say time and time again on this podcast despite the fact that Google has been struggling with AI products so far, going back to what are the keys to success, they have more data than everybody. They have the resources and the compute power and the human talent as good as everybody else, maybe more. They have incredible distribution and they cannot fail because their existence depends on AI. So Google will be a major and extremely successful player in the AI race moving forward. And from Google to philanthropic, andro just launched message batching API for processing large volumes of data. It can handle up to 10,000 queries. asynchronously within 24 hours. The benefit for that is pricing. It's going to save you about 50 percent compared to just running real time token processing, and it aims to compete with open AI's new batch processing feature that I just told you they just announced. So this capability is now available for Cloud Sonnet 3. 5, Cloud 3 Opus and Cloud Haiku, and it's currently available through Amazon Bedrock and shortly will be available through Google's Cloud Vertex AI. So two of the main hosting platforms. This concept of offline batching is actually really interesting because what it will enable companies to do is to define different methods for different processes, meaning when you need real time information and feedback, you will use One type of APIs, and when you don't need the information right away, you can save significant amount of money in getting results on these batched approaches. From a pricing perspective, the pricing is really close between Claude 3. 5 Sonnet and GPT 4. 0. With Cloud 3. 5 Sonnet at 1. 5 per million tokens in input versus 1. 25 of ChatGPT. So still a difference but not a huge difference. And on the output of tokens it's 7. 5 on Sonnet 3. 5 for a million tokens versus 5 on ChatGPT. That is a big difference. That's A 33 percent difference in price, but you get a much bigger context window on the cloud side with 200, 000 tokens versus 128, 000 tokens on the GPT for all side. Overall, great functionality. Combine that with the fact that all these platform, that all these providers are also providing us small, faster models really allows companies and individuals, if they're sophisticated enough to fine tune their processes in order to maximize the value that they're getting or the dollars that they're spending between speed and accuracy and capabilities. And all of that, while the pricing continuously goes down further and further for every usage and every token across all these platforms, until the point that we're going to get to almost free intelligence across the board. Another interesting piece of news from Anthropic this week, their co founder and president, Daniela Amodei, who is the sister of Dario Amodei, who is the CEO of the company, has been interviewed by the information In their conference this week, and she shared some very interesting facts about how Anthropic is using Anthropic. They said that their developers are using Anthropic tools to help them write code and that it's making them significantly more efficient. that much more efficient to the point that it's impacting the hiring and recruiting that they're going to do in order to grow their business. So basically those productivity boosts are offsetting future hiring needs of the company. She didn't say anything about firing people because of these capabilities. But it's still the same kind of game. If it's not generating the same amount of jobs, it could have generated. It means that AI has a very obvious impact on the job market as a whole, including computer programmers in the very near future. The other thing that you mentioned is that getting these higher efficiencies can offset some of the high cost of compute that they have in the business. So the human talent being able to work faster and better offsets some of their built in costs. Now, this is obviously one simple example in one company, but if we generalize this, it literally shows and proves that AI provides companies who knows how to use it, leverage in building higher efficiencies that can help them grow faster without investing additional resources. The other interesting thing here is that companies who develop AI capabilities for their clients can also use it in house. I've heard the same exact thing from GitHub's head of product, Chris Butler, when I sat with him on a panel at AI Realize Summit. Less than two weeks ago, where he shared that GitHub is using GitHub Copilot in order to develop the next version of GitHub that will obviously keep on accelerating itself as these tools gets better and better, because as they get better, it allows you to be more efficient to develop the next version of the thing that you're developing until we get to the point that these AI systems will be able to continuously develop and improve themselves. If we're looking for another example of how impactful and destructive AI is in the world at the biggest scale, Jason hungs, the CEO and founder of Nvidia, his personal net worth. is now more than Intel's market cap. So Jensen Hong's current net worth is 109 billion and Intel's market cap is currently 96. 4 billion. That means he can actually buy Intel, one of the most successful companies in tech history and still have some change left in his pocket. Means that AI is creating Incredible disruption across multiple industries and a company that Intel that couldn't catch up to what NVIDIA is doing has taken a deep dive, basically more than a 50 percent drop in its share price from late 2023 to today. And in order NVIDIA CEO has skyrocketed. AI's impact on value creation across multiple industries is a real thing. And if you are in any kind of business, you need to start implementing this and you need to start learning how to do this. As a business, not as individuals who can figure out little use cases, but have a real strategy and a real process into 2025 and beyond. As I mentioned in the beginning, huge announcements on the AI video generation world this week. The first and very interesting one is Meta introduced MovieGen. Now it's not something they released. It's a research product that they have been working on, but they shared its capabilities. It looks absolutely amazing. It can generate images from text prompts. You can take an image of yourself or a person and turn it into a personalized video using a text prompt and that image. It has really amazing video editing. So those of you who have seen the demos from Meta earlier this year, where they can identify aspects of a video, like the shirt that somebody is wearing in a video, like the background, like specific trees, like a basketball court, like whatever it is that you want. Meta has the capability to identify it. So right now you can not just identify it. You can use this tool to manipulate it. You can change people's clothing as the video is running. You can replace the background. You can change environments. You can change from light from day to night and so on. You Just by adding a prompt to an existing video. So as far as video editing, they are very far ahead of any other tool out there right now. And the last cool feature is that it can generate audio to the video that's on the screen. So whatever's happening on the screen, you can ask it to generate the sound that is quote unquote, the sound effect that this thing generates in their demo. They showed a quad making a jump and you can hear the engine rpm and you can hear the sound that it's doing they show the snake slithering But you can also add music to the background that will fit The kind of scene that you're trying to create all of that just by text prompting as I mentioned Amazing capabilities that sometime in the future will probably find their ways to into the meta universe of WhatsApp, Instagram, and Facebook, where you'll be able to create these images on the fly, most likely for free, like everything they have done so far that obviously puts a lot of stress on other players that are asking people to pay for these kinds of functionalities. And that's such the nature of the approach that Meta is taking to make open source and release into their ecosystem for free, all these tools that other companies are struggling to push out for paying members. Staying in this idea of open source video generation, a new platform was just announced, it's called Pyramid Flow, and it was released this past week. it's an open source platform that is available on Hugging Face That allows people to generate up to 10 second long videos. I did not get a chance to test it myself, but they are claiming that they are in par with other top generation platforms, such as runway, gen three and Luma's dream machine and clink. If that is the case, again, that's going to put even more pressure on these paid platforms to be even better and find ways to differentiate themselves, because if it's an open source model and you can use it on your own machine or host it yourself and not pay these other platforms, they have a very serious, they will have a very serious issue of generating revenue. As I mentioned, a lot of interesting news. So the next one comes from Kling, one of the cool and better tools out there. And they added lip syncing capability to their video generation. And so you can take a video of a person. Add a voice that could be a human voice recorded or generated by 11 labs or GPTs advanced voice mode, and within about 10 minutes of running in the background, it will generate a 10 second clip that perfectly syncs the lips of the people in the image to the voice that you uploaded. This is an amazing capability that will allow creators to generate full dialogues, which means complete scenes of movies using this functionality. They also announced the release of Kling 1. 5 that has a lot of other features. the most interesting one is a motion brush that allows you to choose specific areas in the image and only animate them or control them, as well as they announced that they're going to be releasing an API for developer integration. And last but not least, Runway AI, Runway has added a very needed feature to the platform that actually existed in Dream Machine, which is two keyframes. So you can add a first frame and a last frame, and the AI will generate the video in between. This could be something that makes sense. So a basketball player on the ground, and then a basketball player in the middle of a dunk, and it will render his motion in between. But this could be completely made up stuff. Like an old ancient ruined city and an advanced futuristic city as the two frames and it will generate the transition between them. There's already a lot of really cool examples. This is one of them that shows up online and that showed up in several demos that people has done and you can do that across a 10 second video. You can. Do it with a prompt where you have more control on how the transition is going to look like, but you can even use it without any problem just by dropping the first and last frame, and it will figure out what to do in between. This capability provides content developers a lot more control. On what the video looks like, including the capability to create multiple scenes and stitch them together because the end of one scene can be the beginning of a different scene and then you can render them one by one and then connect them with an external video generation tool. We spoke earlier about the importance of distribution for the success of AI platforms. So a new tool called co rover. ai just reached 1. 3 billion users. That's billion with a B. And that's a tool I haven't heard of, and I assume most of you haven't either. It is developed by a company. Company in India and the way they have reached such distribution so fast is that they've partnered with some of the largest companies in India to develop text and voice conversational AI solutions that these companies are integrating into their offerings. It supports local languages in India, 14 languages for voice and 22 languages for text. And because they have integrated this with existing large organizations in India, some government and some private sector, they were able to get a huge distribution and now have 1. 3 billion users. on an AI solution almost nobody heard of. So if you're developing any kind of solution like this, look for partnerships that can drive that kind of success for you on a very short time frame. And I promised you in the end, some scary news. Now, it was very obvious that these AI tools are also going to be used by a Bad players to do bad things. A very good example was shared this week by a Microsoft solution consultant called Sam Mitrovich. He was the victim of a very highly sophisticated phishing attempt. So he got an email. From presumably Google telling him that he needs to recover his password because things have happened But he also got a phone call that he actually didn't pick up because he thought it was scam So he got an email from Google and he got a phone call from Google Which he did not respond to because he was suspicious But a week later, he got another email and another phone call. And in this particular case, while he ignored the email, he picked up the call because it looked legit. It actually came in the caller ID from Google. Now, this was a AI voice agent that was talking to him and telling him that his account has been compromised and that Somebody got access to his account and has been doing bad things for an entire week, which connected in his brain to the email and the voice call that he missed. Now, it also used real Google data and references while having the conversation with Sam, which made it sound legitimate. Why am I sharing all of this with you? Because you need to start understanding that these tools will get more and more convincing either on their own or combined with other existing methods like a phishing method. It will be able to talk to you. It'll be able to send you emails. It will send you texts. It will be able to be. Very good at trying to hit the goals that was set to it in this particular case, get this person's gmail account credentials, and this will start happening more and more. And if we're not aware of this, we will be giving up our security and our company's security. So you have to be aware and look for the following things. One is these voices are right now too perfect. They sound still a little automated and you can still detect it. That will go away in the next few months. The next two things that are just old school awareness. One is urgency. There's always a high level of urgency in these phishing attempts. You've got to do this now or else. And the other thing is most of these companies will not do an unsolicited call to you. Claiming to be your bank or Google support and so on. You will be able to find them if you log into your account. So if you're getting anything like this, log into your relevant account, whether it's your bank account, whether it's your hosting platform, whether it's your Gmail account, log into there and see if over there, There are any messages that will pop up or that will appear in the message section that's telling you that something is wrong and then you initiate the call to them on their 1 800 number to figure out if something is actually going on. Stay safe. Keep using AI. We'll be back on Tuesday with another how to episode. Don't forget to rate and review this podcast if you find this valuable. And don't forget to come and join us on our webinar this coming week on how to create your AI plan for 2025. And until then, have an amazing weekend.

People on this episode

Head Shot

Isar Meitis

Host