In short
Podcast Notes: Leveraging AI - Episode 89
Episode Overview Title: The (AI) World Changed This Week! Everything You Need To Know About the Announcements from OpenAI, and Google Host: Isar Meitis Date: [Insert Date] Description: This episode dives into the latest announcements from OpenAI and Google regarding their AI advancements, discussing their implications for business and the AI landscape.
---
Key Themes and Discussions
- AI Developments from OpenAI
- New Model Announcement:
- OpenAI announced a new model, GPT-4.0, which is described as "Omnichannel."
- This model features:
- Improved speech recognition and response time.
- Ability to engage in real-time conversations and understand interruptions.
- Enhanced multimodal capabilities, allowing the model to interact visually with the environment through video feeds.
- Desktop Application:
- A Mac-native application for ChatGPT was released, emphasizing usability and integration with daily tasks.
- Monetization Shift:
- Formerly paid features are now available for free, albeit with limited usage. This move is perceived as a data collection strategy to enhance AI's understanding of real-world interactions.
- Integration with Google Drive:
- Users can now easily connect to their Google Drive directly from ChatGPT, allowing for seamless file uploads and management.
- Google’s AI Innovations
- AI-Powered Search:
- Google has integrated AI into its search functions, providing direct answers rather than a list of links, improving efficiency for users.
- Android Integration:
- The AI capabilities are now embedded within Android systems, allowing for smoother interaction across applications.
- Project Astra:
- Google announced its project to integrate AI into wearables, allowing devices to interact with the real world through cameras and sensors.
- Implications for Business and Employment
- Business Efficiency:
- The advancements signal a future where AI can handle tasks traditionally done by knowledge workers, raising concerns about potential job displacement.
- Unemployment Concerns:
- Discussions highlighted fears of significant unemployment in high-paying jobs due to AI capabilities, with no clear strategies on how to address this potential crisis.
- Future of Work:
- The need for re-evaluating educational pathways for the upcoming workforce was emphasized, as traditional roles may become obsolete.
- Ethical Considerations
- Data Privacy and Security:
- The integration of AI into more personal and business applications raises significant privacy concerns. The conversation highlighted the urgency for regulatory measures to ensure safe AI deployment.
- Need for Regulatory Frameworks:
- Sam Altman, CEO of OpenAI, is advocating for an international agency to oversee AI development and ensure safety.
---
Additional Insights
- Real-World Applications:
- The episode detailed several use cases for the new AI capabilities, emphasizing their potential to transform daily tasks in business and personal settings.
- Future Predictions:
- The host speculated on the pace of AI adoption and its transformative impact on various industries, suggesting that businesses need to adapt quickly to keep up.
---
Action Items
- Join Weekly AI Hangouts:
- Interested listeners can engage in discussions about AI applications in business by joining the Multiply AI Hangouts hosted every Friday.
- Feedback Encouraged:
- The host expressed a desire for listener feedback on the episode's format, encouraging comments for future shows.
---
Conclusion This episode of Leveraging AI emphasized the rapidly changing landscape of AI, driven by significant announcements from OpenAI and Google. With potential implications for business efficiency, employment, and ethical considerations, the discussions underscored the urgency for professionals to embrace and adapt to these changes.
Listeners are encouraged to engage with the content and stay informed about the evolving role of AI in their industries.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Transcript
Automatic transcript. May contain errors.0:00Hello and welcome to a weekend news edition of Leveraging AI, the podcast that shares practical ethical ways to improve efficiency, grow your business, and advance your career. This is Isar Maitis, your host, and we're going to do something a little different today. Some of you may know that I'm running the Multiply AI Friday Hangouts every single Friday at 1 p.m. Eastern. There's a really cool group of people that are now getting together every single week, and we're just discussing AI at businesses. People bring stuff from their work, things they have questions about. Other people help them solve it.
0:36I bring my own capabilities and knowledge. We just discuss progress in AI and different questions. And it's just a really cool and loose and great environment to share AI capabilities. This particular weekend was a little different because everything that happened with the news from both Chachi PT and Google on Monday and Tuesday of this week. So what I'm going to do in this episode, I'm going to give you a glimpse into the conversation we had this Friday on these news, because a lot of it is news related. These are the biggest news of the week. So it's going to be in a little different format than you used to.
1:10But I think it's going to provide a lot more depth into some of my thoughts on some of these topics with some participation from the audience. We had a lot of tangents that I removed by editing them out so you don't have to listen to them. but the core that stayed is directly related to the news that was shared by OpenAI and Google this week and some of the ideas and questions that people had from the audience of us doing the Hangouts. At the end of the recording from this week, I will add some big news that happened in a very short format so you don't miss some of the unique and interesting things that happened.
1:46I hope you enjoy this format. I would actually really like to know your opinion about this once this goes live. So if you like it, let me know and I can do more of this. As I mentioned, we're doing these Hangouts every single Friday. And if you want to join these Hangouts and be a part of the cool gang that actually participates in the conversation, if you have something to share, or if you just want to learn how people are implementing AI in their businesses, just send me a direct message on LinkedIn. I will gladly send you a link to join this. This is nothing secretive or it's not paid. We're just getting together to learn AI together.
2:17So as I mentioned, ping me on LinkedIn. And now let's dive to the recording and this week's news.
2:28I'll give you a quick recap of what happened this week, and then we can talk a little bit about the implications of what's coming. There were two big announcements, right? So Google announced that they're doing Google I.O. event on Tuesday, which shortly after they did that, OpenAI said they're going to do a big announcement on Monday. They didn't tell anybody what that's going to be. And there were a lot of rumors. The biggest rumor was that they're coming up with a search engine of their own because they bought a subdomain or they bought the certificate for the subdomain search.openai.com. So then they're like, oh, so they're doing a search thing.
3:08So everybody thought that's going to be the big announcement. Some people thought maybe it's going to be the announcement of GPT 4.5, GPT 5. There were a lot of rumors across every channel you're on in the AI space. And then on like Friday, before the Monday, Sam Altman tweeted that not chat GPT-5, not a search engine, but something we think you'll find really magical. And that was it. And then on Monday, they've announced a lot of things. But the main thing they announced is a new model that they called GPT-40 that stands for Omnichannel. So if you see GPT-40, it's not 40. It's 4.0 and omni-channel.
3:53And it's a model that was really trained completely multi-modal. And they started releasing it in like chunks and steps. So the main thing is it understands speech regularly, like better than it did before. It can speak back with almost no delay. So you can have a real conversation like you're having with other people or with multiple people with it in the loop. So think about inviting it to company meetings and have it take notes and participate and give ideas and so on, literally as a person. Like it talks like a regular person. You can interrupt it in the middle and it understands that you're starting to speak and it's going to listen.
4:31The ability to have an actual voice conversation with no interaction is incredible. That is already released, I think, to everyone. They think that they demoed that they did not release yet, at least not to me. So if somebody has it, please don't tell me because I'm going to be really pissed. But no, tell me if you did get it, is the ability to actually engage with the video. So you can open the video and show it the world around you and it understands what it's seeing and it can engage with it. And the use cases for that are basically endless because now it understands the world. Anything from what's going on in traffic to things you need to fix in your house or your car to how to install things to code on your computer that you can literally show it.
5:20I help your kids solve math problems in their notebook, et cetera, et cetera, et cetera. Like literally arts and crafts things, whatever you can think of, it can look at together with you in real time and have a communication with you about it. And this is a complete game changer from everything that we've seen. They've also announced a desktop app for Mac. So if you're lucky like me and you're a Mac user, there's a Mac native platform right now, which is interesting for two reasons. One is they have a very serious partnership with Microsoft that is paying the bills for them. and they chose to go with a Mac application first.
6:05So that's an interesting number one. Interesting number two is the usability of that. You can share the screen with the desktop app so it can work with you on anything that you're working. So it's the same vision concept that you can do with your phone, but just with anything that you have on your computer. So you can ask it to help you with - Do you have that? Do you have that already on your Mac? Yeah, so I installed it this morning. I didn't get a chance to play with it yet, but I have it installed on my computer. That doesn't introduce any security concerns? No, none whatsoever. I'll say something.
6:41My next sentence was, and they, for the good of their hearts, opened all these capabilities for free. So all the stuff that was paid, including creating GPTs, including using the new model, including creating images, data analysis, like all the stuff that used to be in the paid version is now available for free, only with a very limited bandwidth. And when I say very limited, if you're actually using it like a business user, it's not relevant to you, so keep on paying. You get like whatever, six messages every three hours or something ridiculous like that. So it's like a try before you buy more than anything.
7:22But to me, the real reason that I think that they did it is data. if everybody's going to start sharing their screens on their desktop apps and everybody start opening their cameras on their phones, this is the most incredible access to the real universe any AI company has other than Tesla. And because Tesla has been recording videos from whatever, nine cameras in each car for three years. But Tesla's world interface is limited to what you can see from the road. So they know how the universe around the world behaves more than any other company on the planet by a very big spread. I think the number they shared when they, the only number they ever shared is I think they get 3 billion frames every day.
8:14So that's Tesla. But again, that's limited to what you can see from a car. Chachupiti, once you start using your camera and you start sharing your screen, will have access to everything. And so I think the play that they're telling, yes, we want everybody to have access to free ChatGPT is awesome. I'm not saying they don't mean that, but I think the real play is they will start collecting stupid amounts of data on the real universe, how people are using it, how the model is working, what data can they collect, how to analyze it. That's the play. and you get to hear it first. I'm going to talk about it in my podcast, but the podcast only goes live tomorrow.
8:57So you get to hear what I think first. I think it's a brilliant data play, the fact that they open it all for free. As I mentioned, the use cases are literally endless. Now, the really crazy thing is that they didn't mention at all something that they just released. I got it on my phone yesterday. No, actually even two days ago. I'm like, okay, that's interesting. So I have an Android phone. And in the ChatGPT app, which I use all the time, there's now a button that looks like a little library. I'm like, oh, I wonder what this is doing. So you click on that. It's connecting because I'm already on Android and it has access to all my Google stuff already on the phone.
9:37It opens your Google Drive. So you can upload any file from Google Drive straight into ChatGPT. So as of today, they started rolling it out to ChatGPT on the web. Now, not everybody has it yet. I have two accounts. I have one that is a Teams account. And over there, I have it. And I have my regular business account that is just me. And over there, I don't have it yet. And I've seen a lot of people and chattering about whether you have it or not. But the cool thing is you can connect to Google Drive and you can connect to Microsoft OneDrive. So depending on which universe you're in. So in the little in your chat line where it has the paperclip thing to upload a file, if you click on that it will open a menu that has three different options if you have the feature already option number one is to connect google drive option number two is to connect one drive and option number three is what we had before which is the ability to upload stuff from your computer but just that seamless connection to where we have all our files anyway is amazing to me the obvious next step which i was really hoping it's doing when i saw that little folder icon, it's not there yet, but I'm sure it's coming, is just talk to the folder.
10:51Like I want to create a project folder about my project and I want to have my Excel. I want to have my proposal. I want to have all the analysis. I want to have all my product file. I want to have all of that in a Google folder and just be able to talk to the folder. There is zero doubt in my mind that's the next step. And so it's not there yet, but this is rolling out and it's again, mind-blowing. Now, to add on top of that, you can now upload an Excel file into ChatGPT and use ChatGPT within an Excel file. So it looks like, I don't know if you've seen Copilot in Excel, where you have the actual Excel file and Copilot window on the side.
11:35It's the same thing, just within ChatGPT. So you can use it to analyze data. it creates amazing graphs that you can actually now manipulate the graphs with a chat on the side. I want to change this to blue. I want this to be in the... It's like everything that Microsoft is doing on the co-pilot side, they just built into ChatGPT itself. And you can just upload files straight from your folders and start manipulating them there. It's absolutely mind-blowing. Like my head spinning with ideas and things that I have to try this weekend. And I can't because I have three kids' birthday parties, Israeli Dependence Day party, and there's so many things happening.
12:15But I'm like, I'm not going to be sleeping a lot in the next three days. I'll tell you this. Lots and lots of new stuff to play with. So this, all this craziness. So everything is free, but limited, free, new model that does amazing things as far as engaging with the real universe, including the capability to actually have human-like conversations with it on anything, video access from your phones to this, which I haven't seen anybody that has yet, desktop, a desktop application that I already have that can see your screen and share and learn and help you with anything you're doing, and the ability to connect to Google Drive.
12:53All of that was announced on Monday.
12:58Tuesday was Google I.O., right? So this is when Google released all their stuff. And same kind of thing, right? It's like the amount of stuff they announced is insane. So they're changing everything to be AI focused. So search, I don't know how many of you signed up to have the AI-based search in Google. I've done that as soon as it was possible. So I've had it since, I don't know, probably late last year when they announced Gemini, maybe. No, I think it was even before they announced Gemini. So I've been using it for a while, but it's getting better. So now I think it's built in for everybody.
13:35You don't have to sign in for anything. And you get like the AI result first before you get all the search results, which basically means you don't have to click on anything because instead of giving you a thousand different options to go research on yourself, it gives you the answer. and you can follow up with follow-up questions and so on. So very similar to what Perplexity is doing on their platform. And it's working well. I've been using it for a while and it's very helpful. The other thing that they've announced is they're going to embed this into Android phones to be able to chat with everything, going back to the question, oh, what about security?
14:14Do everything on Android. So they're adding the AI layer, the Gemini layer, not as an app, but as an OS feature, meaning it can work with regardless of what you have. You can open Gemini and it can interact with what's on the screen at that moment. So the same concept that we've seen from ChatGPT doing to desktop and the same thing that Microsoft is going to do with everything Windows, Android is going to have built into Android. again, security, different question. From my perspective, Google knows everything about me anyway, because I'm in the Google universe for business. I'm in the Google universe in my personal life.
14:56I have an Android phone. I've got a Google home thing. They know which underwear I wear every day. I think it's at that level. So I really don't care at that point because they know everything that I do. And I use Google login for probably 50 % of the stuff that I'm logging into. so they know everything about me anyway so if there's anybody i'm going to give access to it's them because i at that point it doesn't really matter anymore so that's another very interesting thing they announced they have a similar thing so they announced project astra which is the ability to open your camera and walk around and show it stuff and have a conversation with it and it will respond then remember and in the demo i highly recommend you go and watch some of the demos both from ChatGPT and from Google.
15:39They're both troubling and exciting at the same time, like a lot of stuff in the AI world. Both of them demoed stuff both from a phone as well as from some kind of a wearable. So that's the next thing. So we all, I assume most of Meta has a partnership with Ray-Ban. How many of you knew that? Raise your hand. Some of you. Okay, so there's Ray-Bans today that look like regular Ray-Bans, but they have a camera and a microphone and a headset. built into them. So you can basically share the world with Meta. I have very little doubt this is where the world is going, which again has so many issues and questions that I don't even know how to start addressing.
16:21Like the most simple one is privacy. Google have the same ability now to see the world and engage with it in a seamless way. So once we're going to wear glasses or Meta is now developing another news from this week meta is now developing earbuds that will have a camera in them and so even people don't wear glasses and you don't want to wear glasses all the time unless you have to and you don't want to wear sunglasses ray-bans because it could be at night or at home or it's going to be weird but half the people now have earbuds in their ears all the time so if you have that and that has a little camera on the on it and it can see the world as well then you see where this is going.
17:04And so zeroed out in my mind that this, where this is going, Google were very clear that this is going to be integrated into everything. They've already started creating a lot of really cool use cases where I think this is where it starts getting real interesting. So the whole concept of agents that understand what you're trying to do and can actually give themselves tasks and solve things on their own is we're starting to see the first sprouts of that. And it's going to get really wild later this year. But if, let's take for a second Google and Microsoft as an example, because these are very well-defined ecosystems that we engage with all the time.
17:43And think about if it has access to your emails and to your Google Drive and to all your Excel files and all your reports and all your Google Meet communications, both chat and recorded calls, it basically knows everything about your business. Everything. And then you can just ask a simple question. Say, hey, I'm getting ready for this meeting with so-and-so. Collect information for me. And it will just go and collect information for you. And that will be from all the sources because it knows everything. The other thing that they started building in Google that they shared that, again, just follows the same concept, which I think it's just humanizing it so we feel comfortable, you can create team members that are AI.
18:33And then you give them a role. And you basically say, hey, I'm going to create Tina. And Tina is going to be my project management assistant on this project. And you give her access to specific things. And you give her access to the Slack of that project. and you give it access to, you invite it to all the meetings and so on. And then it can take actions on anything you need in the project. Summarizing things, creating statuses, creating Excel files, sending invoices, like literally anything you need. This is available later this year from Google. And again, I don't see Microsoft doing anything different this year.
19:13So the way we do business, everything we do in business, is going to have an AI entity infused into it, which on one hand is incredible from an efficiency perspective. Most people need somebody to know everything that they're doing and be able to pull stuff together very quickly for them when they need it. And the direction is very clear and the speed is not something that any business is ready to deal with. because even just the announcements again this week are mind-blowing but this is this week we like they keep on developing and deploying stuff all the time gpt5 is not out yet there's the apple worldwide developer conference coming up next month i'm sure they're not going to stay silent And so we're walking into a universe where AI is going to be part of literally everything we do.
20:19And I think the spread between the people who understand it and can capitalize on it and can run fast with it to those who are still, yeah, I don't even know. I haven't tested it yet. I know it's going to grow so fast that it's going to be very hard to close. Like in three, four years, everybody's going to close the gap. But the opportunity to right now do things that are literally magic, magic. Just think about on anywhere you are on Google. I'm just a Google user, but it's going to be the same thing on Microsoft. Anywhere you are in the Google universe. You can ask any questions about anything and you will put together an amazing summary for you.
21:05They showed something like not a business, but a personal thing of it of like, oh, let's plan a trip. And she basically says, okay, we're a family of four. We're going to this and that destination. My son likes sushi and my son like art and my husband likes sushi, something like this. You can find my hotel reservations and my flight reservations in my Gmail, please plan a two and a half. No, please plan the trip for us. And then it knows it's two and a half days because it has the flights. And so it literally builds an agenda for two and a half days based on sushi and art. And then it says on the second day, we want to do this.
21:42So everything switches around and it's these kinds of things. But now you take this to the business world where it knows everything about everything. And you can ask any question and create any report and generate any email and draft any agenda and follow up. Literally, the work of business will be done by this machine way more effectively than we can ever compete with. And then the question we have to ask ourselves, two questions, is one, how do we implement it in our businesses in a way that doesn't break it? And B, what do we do with the people? Because that it's a question I was asking about like 2025 and it's becoming a 2024 question as I saw everything that's going to get deployed now.
22:37So very exciting, very scary all at the same time on steroids. Any questions? that's after this anyone talking about this yet what do we do with the people i was just thinking yesterday i can't remember what i was doing doing some ai thing and oh i watched your podcast because you that woman was talking about that product and it was things that i had already thought about like creating a gpt for hr creating gpt for program management all the documents and train things and just for our software and these kinds of things. And I was just thinking in my head, like admin roles can already be gone, essentially, with just what we already have with AI.
23:26And is anyone talking about it? Is anyone thinking? I know everybody's scared. We're all afraid of it, but who's actually talking about what we're going to do with the people? I don't think anybody has a good answer. I don't think anybody important enough is raising big enough flags. If you want to touch on news and to show you how much this is not the main focus, Jack Schumer just released a paper about where they want to take this from a legislation perspective. And it's like a high level umbrella kind of conversation right now. It's still not anything. But the safety aspect of this is number three or four on the list.
24:12and none of it talks about this. It talks about safety concerns of this can hurt our democracy because deepfakes and it can impact our elections, which is the core of how this country runs, right? Okay, and then he talks about chemical weapons and biological weapons and stuff, but he doesn't talk about what do we do if next year there's a 30 % unemployment of people who are making between$150 ,000 to$500 ,000 a year. What do we do? Nobody has a clue. Haven't we seen this before, though, with jobs being sent out over the Internet for like transcription? And there's already been so much of this with the Fiverr and being able to take jobs that would normally have a full administrative person local and just being able to be sent out over the Internet to anyone.
25:06That's going to hit those kinds of, what would you call those, offshoring of jobs pretty heavily. So, go ahead. One thing I was thinking about was when you said that, Jonathan, was just we saw in 2020 what happened when everybody stopped being able to work and the impact that had everywhere. but that was something that most economies have tried to bounce back from. This isn't something like that, where it's not like, hey, I'm just going to go away after a few years, and companies are going to start opening, and people are going to start going back to work. Yeah, I want to say two things in addition to what Danielle said, and it is one is speed.
25:55So if you think about, let's go further back in time, There were other revolutions, right? Industrial revolution, and then electricity revolution, and then the internet revolution, and then mobile. Each and every one of those took a lot of jobs away. The industrial revolution before this, we're doing everything by hand. To work a farm, you needed 60 people, and now you need one guy with a tractor, and he can do the same work. but it took 200 years. Now let's take it to, okay, that was 200 years ago, but let's look at the internet revolution, right? The last really big one that we had. I always give that example and some of you might've heard it.
26:39The first time I connected to the internet, I remember that day vividly because it was the most underwhelming experience of my life. My roommate at the time brought this device to our house and I was living in Tel Aviv and he's I brought this thing. It's going to be awesome. What is it? It's a modem. Okay. But what is it? Okay. Allow us to connect our computer to the internet. Like what's the internet? I don't know, but it's like a new exciting thing. And it's, we got to get on the front of this. I'm like, okay, we made dinner. I had dinner and then connected the thing. And we went, those of you who are old enough, remember that.
27:15And after whatever, two minutes of this, we got like a little prompt, like a DOS triangle thingy on the top left corner of the screen. There was no user interface for it yet. And we're looking at this and like, okay, so now it's, I don't know. And that was it. That was my first time connecting to the internet. And that was, I think, 93. The first time I got an email address, my first Hotmail email address, I got in 99 when I went backpacking in Latin America. That was seven years later. So in seven years, I went from connecting to the internet for the first time to having an email address. Forget about this, that I can see what satellites are doing and all over the...
Read the full transcript
28:06So it took a very long time, which meaning building new jobs that are a necessity of every new thing was running in parallel to losing jobs. And in some cases, actually faster. In this case, there are two major differences. Major difference number one is speed is insane. Like nobody, not a single organization in the world can deal with what's coming. And two is how wide it is. This is not a specific role, a specific niche, a specific industry, a specific title, a specific level of education. This is every knowledge work, literally every knowledge work will be able to be replaced by some kind of an AI machine within the next two to three years.
28:59Now, add to that the fact that robots are coming shortly after, these humanoid robots that you don't, and people are like, why do we need humanoid robots? Why not build a robot for a specific task that is more well-built than 10 fingers and four limbs? Like why? And the reason is all the machines we have, all the stuff we have, all the universe we live in, we built for us. So if you build a humanoid robot, it can replace any job that a human is doing because it doesn't have to be customized to how to do this thing in a factory. You can build the same robot and it will do anything that we've built interfaces for.
29:41Now, over time, that will change. The internet, as we know it, I'll go back to search. We are not going to search the way we're searching right now. Some of us already don't, like I don't. But what does that mean? Do we need websites anymore? probably but not sure and if we need websites they need to be optimized for these robots and not for me not for humans i'm not going to go to the website you go to the website do the research come back tell me what i need to know what do you think about page search i've been trying to figure that out like what's going to happen to paid search google owns that now is that gonna dissipate or not be necessary or something else?
30:27To me, that's the biggest financial question of our generation. Literally. Google makes, I don't remember the exact number and just can probably find it in two seconds or one of you. I think they make around close to$50 billion a quarter from search. Close to$50 billion every single quarter. That's what makes Google doesn't exist. forget about Google, Alphabet, like all the other companies, all the cool stuff that they're doing makes 1 % of the revenue, probably costs them 20 % of the capital they're investing. And without paid search, there's no Google. So they're making a really big bet, the biggest bet any company ever made in the history of companies, but they don't have a choice.
31:27They were trying to delay this as long as they could, and they did. They sat on the sidelines and yeah, gave us some stuff, but they understood, I think very clearly, that perplexity is out there with a tool that competes with them. Microsoft is going to put something that competes with them. ChatGPT is going to put something that competes with them. And then a few other startups that are doing similar things. And all that needs to happen for Google to lose 25 % of its value, which I don't even know what that number is, but it's a really big number, is for them to lose 5 % of their current dominance in world search.
32:06That's it. If they lose 5%, five points of their whatever, 82 % of global search, from a share value perspective, they're fucked. Big time. And so I think they understand they don't have a choice.
32:31You said something, Danielle? Oh, I said they're not the only one that's fucked. the market's fucked at Google.
32:41We're walking into a territory that nobody knows how it looks like. There's also the question about news media. If your agents can just go out there and bring in the news for you, who's going to go to a news site? And then the entire advertising for all news sites collapses as well. Yeah, again, there's so many dominoes that are starting to fall, and then what are the dominoes that are going to knock down? Again, I don't think anybody knows. To me, the biggest concern from an economical perspective is unemployment. We've never had, ever, high rates of unemployment on the top jobs in the market.
33:25When companies need to lay people off, they lay off one executive, two VPs, and 12 ,000 other people. It's going to be the other way around. The people who drive the economies of the world, it's not just the US, the people who drive the economies of the world because they have the biggest spending power are going to be out of a job. And now you're asking, oh, so what job do we create for them? And nobody has an answer yet because these hasn't revealed themselves yet. Now, do I think it will come? Based on every other time it happened in history, it will come. I just think it will take a much longer amount of time for it to reveal itself.
34:07And then we have a very long period in which we'll have to figure it out. And that's not going to be pretty. Again, that's my personal thoughts. I don't have a crystal ball. I really hope that I'm wrong. By the way, the other question that we need to ask is education. I have three kids. One of them is a junior in high school. She'll be a senior next year. what does she need to go to for college what did what do i tell her to go study have you heard anybody asking that question what colleges need to teach what high schools need to teach what middle schools need to teach did you see that duolingo stock took a massive a big drop after the announcement of of 4-0 because you can just talk to 4-0 and it'll teach you a language so who needs duolingo yeah the amount of businesses that are going to go out of business because the big AI companies will just do everything, is it's going to be probably the biggest erase of capital in history.
35:11And it's already starting to happen because so many companies are doing something that this thing now can do better, faster, cheaper, blah, blah, blah. And this is why it's still expensive. like the amount of compute that it requires to build and the amount of computer requires to run is still really expensive but that's going to go down like google's announcement i don't know how many pricing of these to to run inference with these models so inference is how much it costs you to generate stuff so not the inputs you give it but what comes out of the models it's called inference the cost of inference vary open source models around the small open source model around 12, 15 cents for a million tokens.
35:53So 700 ,000 words. The average open source are around 25 cents. The closed sourced ones are in dollars already. And so if you're looking at ChatGPT4, like I think $10 for inference. And if you're looking at Anthropic Cloud3 Opus, so their biggest, largest model, it's$70 compared to the small open source models at$0.12. Now, again,$70 for a million tokens, so paying$70 to generate 700 ,000 words is not a lot of money. It's still cheaper than any employee on the planet. So even if that was the cost, it's way worth it than hiring somebody. But the price is going to go down now. So when Google did their announcement on Tuesday, they're releasing their most capable model, Gemini 1.5 Pro.
36:53Inference is going to be$7 if you're running really large prompts,$3.5 if you're running really small prompts. and they have a smaller, faster model they call Gemini 1.5 Flash and that's going to be like 35 cents. And Chachupiti did the same thing, right? Prices are going to go down dramatically. A, because the companies who build the models themselves will figure out better algorithms to be more efficient. B, because the tech is, like the hardware is getting better. How many of you, Grok, not the model by Elon Musk, but Grok, the hardware company. So there's a hardware company that is private, and if there was a way to invest in them, I would.
37:38They created a completely new architecture of chip that they call LPU, language processing unit, instead of GPU. And it runs between 10 and 1 ,000 times more efficient than the latest GPU by NVIDIA on inference. So NVIDIA still rules from a hardware perspective when it comes to training models, but they are significantly more efficient when it comes to running models. How many of you have heard of Liquid AI? Liquid AI is another private startup that if there was a way to put money into them, I would. It is a startup that was founded by several brilliant people, PhDs from MIT. and they've developed AI platform that is not transformer-based.
38:31So the entire architecture of every AI thing that we know right now on the text and language side is built on an architecture called Transformers that Google researched and invented and shared with the world in a transformer paper back from 2017. And literally everything we know is based on that. They've developed something else. It's not transformer-based. It's a completely new concept of AI. And it scales significantly more efficient. And it runs with significantly less compute. And it has significantly larger context windows. And so there's many different ways, both on the efficiency on the computing side, as well as the algorithms, as well as the architecture, like everything is going to become significantly cheaper.
39:16And then the question is, how good is the product? Is it well integrated? Does it help me solve my needs in an efficient and effective way? This is going to be the difference between all the different models and not the cost because the cost is going to be negligible. So this is the end of the recording from our AI Hangouts. After that, there was a lot of Q &A and discussions about people and what they're doing in their companies, which is less relevant for the podcast. And now I want to give you some additional news, big news that happened this week. Two of them are directly related to real world implications of AI.
39:52So Google DeepMind, research group that is leading the development of AI within Google, released AlphaFold3. It's a groundbreaking AI system that can predict the 3D structure of proteins of live matter. And it can do it in a level of accuracy that is as accurate as x-rays and cryo-electron microscopes. So this is a complete revolution of how scientists and researchers can look into how life is built, the actual Legos, the building blocks of life. Now, the amazing thing, it can get this level of accuracy of proteins that build live matter in minutes, a task that used to take months up to this capability.
40:38This obviously has direct and profound implications on drug discovery, disease research, and just understanding the fundamentals mechanism of how life is built. And this is already released and available for researchers around the world. Those of you who know a little bit about Demis Sassabes, who is the person who is leading the charge in DeepMind, he has very deep belief that this can change life on Earth. for the better. And most of his research goes down to that path on how we solve big problems for humanity and for our planet. On a completely different aspects of practical implementation of AI, YMO, the autonomous rail-hading service that is owned by Alphabet, which is the Google mothership, has achieved a very interesting milestone this week.
41:32They just surpassed 50 ,000 paid trips per week in the cities they operate like Phoenix, San Francisco, and LA. So these are driverless cars that are picking up paid passengers in three major metropolitans in the US. And they are getting more than 50 ,000 trips every week. This tells you where the world is going. Like the concept of autonomous driving is not a concept anymore. It's a reality. and my assumption is we're going to see more and more of these and not too far in the future. We'll see very few, if any, human-driven cars on the main roads. It's going to be significantly safer. It's going to be significantly less pollution and traffic and so on because we're just going to have less cars.
42:22There's going to be algorithms. We're going to make sure these cars are available where they need to be available at the time they need to be available at, which means less cars, less traffic, less parking issues, and so on. A completely game changer. And I don't think that's going to happen in a year or two, but within about a decade, I think the pendulum is going to shift more to that side of the swing. Another big content agreement was signed this week. This one is a deal between OpenAI and Reddit. It will give OpenAI access to everything in Reddit through their API, which means users of ChatGPT will be able to get real-time access to content that comes from conversation within Reddit, which is maybe the most active conversation, the most active chat room platform in the world, at least in the Western hemisphere.
43:11I'm not exactly sure what's happening in China. And so this is another step in the direction of getting external, proprietary, highly detailed and focused content from multiple platforms into these models. Reddit already signed a similar deal with Google, and this is just showing you how powerful the content that they have, that they now have these licensing deals of Google and ChatGee. And the last piece of news comes from Sam Altman, obviously the CEO of OpenAI. He is now advocating and fighting for the establishment of an international agency to oversee regulation of artificial intelligence and to ensure its future safety.
43:50If the slightly doom and gloom conversation we had at the AI Hangouts that you just finished listening to, you understand how much something like this is needed. Will he be successful? How much will it cover? Will Chinese company join in? What happens with open source groups? All of these are still open-ended questions, but I'm really happy that people that are at the tip of the spear of the AI efforts are understanding the implications and are at least trying to take action in order to make it safer for all of us. Before I say goodbye, I wanna remind you that you can help us promote this podcast and you can help other people with AI literacy by simply sharing this podcast.
44:33So pull up your phone right now, click the share button on your app and just share it with several people either directly on your chat platforms or post it on your favorite social media. And also if you haven't done this so far, please rank and rate this podcast on your favorite platform, whether it's Spotify or Apple Podcasts or whatever other platform you're on. I would really appreciate it. And it helps us get to more people and help them get educated about AI. So maybe together we can change the trajectory to be more positive impact than negative. So on this positive note, I will say goodbye for now.
45:09I will most likely record a much more detailed use case related episode related to all the news that OpenAI and ChatGPT released this week. And we're going to release that on our Tuesday coming show. And after that, we'll go back to our regular hosting of guests and so on. This was just a huge week with really big implications. And I want you to really understand what could be the impact of the business beyond the high level that was discussed on the AI Hangouts. And until then, have an amazing day.
From the publisher
Is the AI revolution moving too fast for you to keep up?
With groundbreaking announcements from OpenAI and Google this week, it's crucial to stay ahead of the curve. From multimodal capabilities to seamless integration with everyday tools, AI is transforming business operations and beyond.
In this special weekend edition of Leveraging AI, host Isar Meitis shares insights from the latest Multiply AI Friday Hangout, highlighting the key news and their profound implications for business leaders.
About Leveraging AI
- The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
- YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
- Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/
- Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events
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!



