Everyday AI Podcast – An AI and ChatGPT Podcast - EP 210: OpenAI Sora, Gemini Ultra 1.5, NVIDIA Chat with RTX, Andrej Karpathy leaves OpenAI and more. AI News That Matters - Feb. 19th, 2024

Episode Date: February 19, 2024

The last 7 days of AI news have been crazy!  OpenAI announced its amazing text-to-video Sora, Gemini released Ultra 1.5, NVIDIA's Chat with RTX, Andrej Karpathy leaves OpenAI, and more! Here's this week's AI news that matters.Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode pageJoin the discussion: Ask Jordan questions on AIRead the newsletter on this episode: Read it hereRelated Episodes: Ep 204: Google Gemini Advanced – 7 things you need to knowEp 181: New York Times vs. OpenAI – The huge AI implications no one is talking aboutTomorrow' Show: OpenAI's Sora: The larger impact that no one's talking about.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTimestamps:02:40 OpenAI text-to-video model Sora07:50 Gemini 1.5 update: Developer and enterprise access.15:19 NVIDIA's Chat with RTX19:18 OpenAI announces 2 new AI agents.21:12 OpenAI and Sam Altman's ambitious plans.25:21 Tech companies form unofficial accord to avoid penalties.29:14 Discussion on AI models and election safety.31:18 AI companies shift data acquisition to formal agreements.33:56 Potential impact of AI on publishing industry.Topics Covered in This Episode:1. OpenAI's Sora2. Google Gemini 1.53. NVIDIA Chat with RTX4. Andrei Kapathy's departure from OpenAI5. Reddit's AI content dealKeywords:Large language models, Star Wars, token memory, ChatGPT, Google, Gemini Ultra 1.5, NVIDIA, Chat with RTX, Andrei Kapathy, OpenAI, generative AI, Everyday AI, OpenAI SOAR model, AI legislation, social media regulations, disinformation/misinformation, 2024 US elections, Reddit, AI content deal, AI companies, content providers, SORA, text-to-video model, Entre Kuparthy, Tesla, AI agents, Sam Altman, GPU chips, deep fakes.Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Even if you were trying to keep up with the AI news this week, good luck.

When it comes to the generative AI space, the last seven days have been absolutely nutty. There's been so much happening, so much news that actually matters to all of us. So that's what we're going to be talking about today on Everyday AI. Welcome. My name's Jordan Wilson. and I am the host and everyday AI, it's for you. Well, it's for all of us to help us all keep up with what's going on in the world of generative AI and how we can use it to actually grow our companies, grow our careers.

All right. So we do this once a week. On Mondays, we go over the AI news that matters. So you don't have to spend three hours each day trying to keep up. You can just spend about 20 to 30 minutes on Monday morning and we'll sift through all the nonsense and we'll tell you what actually matters to growing your company and to growing your career. And there was a lot this week. There's a lot. All right. So we're going to get into it. But before we do, as a reminder, if you haven't already, make sure to go to your everyday AI.com

and sign up for the free daily newsletter because, yes, we're going to be going over the news that matters over the last week. But man, we still have a lot of other daily news that you should be looking at as well. Soft bank looking to raise $100 billion, a new mistral model, Apple building an AI coding assistant to compete with Microsoft's GitHub, 11 labs with some new announcements, so much late breaking news. But if you haven't already, make sure to sign it for that free daily newsletter. So you can keep up. And I tell people, we have what I think is one of the greatest and largest resources of free generative AI information on the entire internet. We've had more than 210 episodes now. And you can go back read the newsletters for each and every one. You can go back

and listen to the podcast. You can go back on our website and watch the live stream and learn from more than 120 different experts that we've interviewed. So make sure you go do that. All right. And if this is helpful, please consider reposting this and sharing it to keep your network in the loop. All right, enough. Let's now talk about the AI news that matters for this week. So first and foremost, even if you've been sleeping under a rock, you somehow probably heard wind of open to a AI's new SORA text to video model. All right. So let's go over high level what this is and what it means.

All right. So Open AI has announced their new text to video model called SORA. So a couple highlights that I think separate this from other text to video models out there. Some of the big players you've probably heard of are runway and PICA Labs. Also, the big players have been announcing their own models as well. Google has announced the in the air. And then you also have meta that has announced EMU video. But at least the publicly available models right now kind of leading the space are runways,

Gen 2 and PICA Labs 1.0. So here's kind of some things that you need to know about OpenAI's new model, their SORA. The biggest differentiator is the quality. y'all like when I first saw this I posted a comparison video and I'll leave that in the show notes today as well. But if I'm being honest, I could barely believe it. I could barely believe it. I'm like, is this real?

Right? Which to tell you the truth, I never really do that with Gen. AI. You know, I've used hundreds. I should probably keep track somewhere because it's probably getting near a thousand. But I've used hundreds, personally use hundreds of generative AI tools somewhere like, okay, yeah, that's nice, that's useful. This is one of the few, again, it's not open to the public yet, but this is one of the first

that I've looked at. I've looked at the results and I've said, there's no way. You know, there's no way. But it's, you know, apparently true, right? So originally I was even hesitant to share about it, you know, on LinkedIn or to even tell people about it because I'm like, all right, there's no way this could be a ruse, right? And I waited because eventually Sam Altman was the CEO of OpenAI and obviously their popular product chat. GPT, Sam Altman was sharing new videos in real time based on what people were kind of talking about on Twitter or requesting on Twitter.

So you saw Mr. Bees asked for, I think it was a monkey playing chess in a park. And within minutes, there was a video posted. So it was so good. the quality was so good, I didn't believe it at first, which again, for the hundreds of generative AI products that I've seen and I've used and we cover them, you know, we cover, you know, a handful of new products every single day in our newsletter. I couldn't believe it. The quality was that good.

It's in another category. It's in its own category right now, right? Like I wouldn't, you know, again, we do have to wait until the model becomes publicly available. but it's in its own category right now. So another big thing that I think separates this new model from others is it can go up to a minute long for video generation. So put it in a short text prompt and you get a video that is up to a minute long versus normally, you know, these other platforms are normally around four to five seconds.

And, you know, some you can kind of extend them up to maybe 16 seconds. So a couple other things that are important to keep in mind. It's not publicly available yet. And then also, it's right now, it's only available to red teamers, which are people who check it for safety in a select group of artists and designers. But like I said, so far, the results so far out exceed everything. And it's not even close. It's not even close. Yeah, I definitely agree with Dr. Harvey Castro here in the comment saying the SORA infrastructure hints that we are closer to AGI.

Yes. You know, there's some interesting concepts in SORA. And, you know, we'll make sure to share more in the newsletter because this is one of those things that you just kind of have to see to believe. It is very hard to describe on a podcast or even on a live stream showing you screenshots. But it is extremely, extremely impressive. So impressive, we are having a dedicated show tomorrow. So if you do want to know more about the new. text to video model from OpenAI, the SORA model. Make sure to tune in tomorrow because we're going to be talking about the larger impact that I think no one's talking about. Because I think this model from Open AI hints at something much larger than text to video. All right. So make sure to join us tomorrow for that one. All right.

Our next piece of AI news that matters is Gemini Ultra 1.5 is released. Well, it's kind of released. All right, so we'll get to that. But this is the new model from Google. So right now, this is why I said it's kind of released. Right now, it's only developer and enterprise customers have access to a limited preview of Gemini 1.5 Pro via the AI Studio and Vertex AI. So if you are a kind of a Gemini advanced customer right now, you do not have access to Gemini 1.5. unless you are a developer or enterprise customer accessing it through their AI studio or Vertex AI.

So a couple of things that are important to keep in mind here, some of the improvements of the model. Let's talk about that. First and foremost, it's the context window. Okay. So this is from Google. Again, you know, it's not really out there in the wild. So not a lot of people are able to test this yet. but a 1 million token context window for enterprise customers.

And then your normal commercial customers are going to get 128K tokens. Okay. So what that means, a million token context window is huge. So the way large language models work. And this is the most basic way that I can explain it. And hopefully it makes sense, you know, if you're driving to work in your car or walking your dog right now without showing you, a visual. But if you think of, you know, back in the day, the old Star Wars credits, right?

And you kind of saw how they came onto the screen, you know, kind of at an angle. You can see a pretty big chunk of the kind of the opening sequence, the words, you know, in a galaxy far, far away, right? And eventually, you know, you'll have a couple paragraphs on the screen, but eventually that top paragraph will disappear as new text comes on. Okay. That's essentially how large language models handle memory, right? It can't remember literally everything that you put in, right? So as you have a conversation back and forth with a large language model, it'll start to forget things, right? And that's what you get into these memory windows. So everything is a token. So I like to tell people this. Right now, chat GPT is probably one of the

more popular large language models out there. It has a 32K token memory, which is about 26,000 words. So after 26,000 words back and forth, it starts to lose its memory. So you can see how one million tokens might be great because one of the, I think, one of the largest downsides of working with a large language model is that memory. And so many people, I would say 95 to 99% of people who are using large language models don't use them correctly. and one of the biggest things is they don't understand the memory context.

So this is a pretty big announcement here from Google pushing to $1 million for enterprise, for enterprise customers and developers. Also, this is from Google's own reporting, as they said, that there is a 87% increase on their own benchmarks. So essentially an 87% increase from Gemini 1.5, Gemini Ultra 1.5, versus Gemini Ultra 1.0. So 87% increased performance on these different benchmarks. However, you still shouldn't go ask what Gemini is.

Still is not very aware, right? So even this morning, I have Gemini advance right now. It is two months free, and then it's $20 a month after that. But you should keep in mind, most workspace accounts, right? So as an example, my business, we have our, you know, what used to be called G Suite, but we have our Google Workspace. We can't access, you know, Gemini, advanced. You have to have a Google One subscription.

And for whatever reason, Google makes it very hard, right? For people, small businesses that are using Google Workspace to pay money and to use their AI products. I don't know why. So, you know, I think so many people, myself included, when we're using Gemini, we have to use our personal Gmail accounts. So we can't really test it in a real world scenario where we're working, you know, with our connecting. I think one of the best prospects of Gemini is connecting with your Gmail, right, with your work Gmail and your work calendar, in your work Google Drive, and your docs and your spreadsheets, right?

But you can't right now for whatever reason because so many smaller workspace accounts do not have access to Google's AI products. I don't understand it. Google, can someone reach out to me and tell me what's going on here? But anyways, even as of this morning, Google is not aware. You know, I'm asking Google, are you Gemini Ultra 1.5 or Ultra 1.0? And it doesn't give me a straight answer. It says, I'm not either Gemini Ultra 1.5 or 1.0.

That's another thing. Hey, new large language model companies, when you release updates, you need to be more clear about what it is, who gets access, and your model itself and its system prompts should know specifically because otherwise there is confusion and people aren't going to adapt to your products if you are confusing the majority of customers. All right. Now, some more large language models. Hey, I like this here from Woosie.

Woozy, thanks for the comment. So joining us live says, I feel like at one million context window, it's going to start being me that forgets what I'm working on in these long chats and not the model. Absolutely. A one million token context window is so much, right? It's more than, yeah,

it's more than I think most human. beings can remember. So that's a great point. All right. So we have more large language model news. So, Nvidia has publicly released Chat with RTX. So they did announce this last month, but just, I believe it was Tuesday of last week. So just under a week ago, they did now publicly, Nvidia publicly unveil its new, I'll say software called Chat with RTS. All right. So let's talk a little bit about what this is, what it does, and why you might want to use it. Okay, so right now, chat with RTX from Nvidia is a free local model that you can download on GitHub.

So if you don't know what a local model is or a small language model, don't worry, literally just a whole episode on this last week. So we'll make sure to link to that in the show notes. However, you do need a PC right now to take advantage of chat with RTX. You cannot access it on, at least out of the box. You can't access it on Apple computers on a Mac. So it also does require a PC with an Nvidia, Gforce, RtX30, or 40 series GPU with at least 8 gigabytes of video RAM. So there are some technical minimum requirements for your PC.

So for the most part, if you have an old PC from a couple of years ago, probably not going to be able to use chat with RtX. You will need a much newer PC with specifically one of those two Nvidia GPU chips, the GIFR-TX-30 or 40 series chips, and as well as 8 gigs of video RAM. So the chatbot can use a couple different models, including Mistral or Lama, those open white large language models, and you can search through local files to provide answers. That's the big thing right now. So this is one of the first kind of popular models from at least, you know, your household names. You know, when I'm talking about that, I'm saying, you know, Microsoft, OpenAI, Google, etc.

This is one of the first models that brings, you know, rag or retrieval augmented generation front and center to this new chat with RTX software. What that means is you can essentially interject your data and have it live. and stay within the model, right? There's kind of workarounds that you can do this in chat GPT, also in BART, and in Microsoft Copilot, where you can just upload some of your documents in certain chats, but depending on which model we're talking about, they handle that differently. So a little bit differently with chat with RTX, it does kind of lead with a retrieval augmented generation or RAG approach first.

So what that means is you can interact with the AI model using local files as a permanent data set and it can support various file formats like like.tex.p.d.f. doc, XML files, as well as you can incorporate, which this is what I'm excited about. You can incorporate information from YouTube videos and playlist. So it should be pretty exciting. And I'm sure I'll be covering this more. Actually, a lot more in March. I will be at the, the Nvidia GTC conference. Invidia is actually bringing me out there to partner with them to broadcast from, San Jose live that week. So if you're going to be in San Jose in March at the GTC conference,

make sure to reach out to me and let me know. We'll meet up. We'll say what's up. We'll go talk. We'll go talk dorky stuff at the conference. All right. Another piece of AI news. Man, this has been such, such an action-packed week.

It's so much I have to take a sip of water. Oh, man. Adobe just introduced an entirely new way to create. bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the All In One Creative AI Studio. Powered by Adobe's Creative Agent, Firefly AI Assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the Assistant.

The Assistant orchestrates multi-step workflows, drawing on 60-plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, and Illustrator, Premier, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director.

Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. Here we go. Another big piece of AI news that matters. And here's why it matters. Andre Capathy is leaving Open AI. So if you don't know, Andre is one of the most, well, like I would say one of the most well respected in prominent figures in the field of

artificial intelligence. And he just announced his departure from Open AI for the second time. So he did have this kind of announcement on Twitter, this tweet that he put out that I'm sharing on my screen. He said, hi, everyone. Yes, I left Open AI yesterday. First of all, nothing happened. And it's not a result of any particular event, issue or drama. But please keep the conspiracies coming as they are highly entertaining.

And then Andre went on to say, actually being at Open AI over the last year has been really great. The team is really strong. The people are wonderful. And the roadmap is very exciting. And I think we have a lot to look forward to. My immediate plan is to work on personal projects and see what happens. Those of you who follow me for a while may have a sense of what that might look like. Cheers.

So a little bit more about Andre. So he was a founding member of OpenAI. And he initially left in 2017 to join Tesla leading their AI teams there. And then he just came back to Open AI just a little more than a year ago. So founding member, then he left to lead kind of Tesla's AI efforts for about. about five years, and then he just came back to OpenAAI for about a year. So a couple other things, and I'm going to tell you here why this matters to everyone. All right here.

So Caparthe had been hinting at his involvement via Twitter. He had this kind of as his bio that he was working on projects related to AI assistance, mentioning his work on, and I quote, building kind of a Jarvis at OpenAI, end quote. And then he's kind of reflecting on his vision for creating helpful and conversational AI systems. So the timing of this, I think, is particularly interesting. So it was just about a week prior to this that the first kind of initial or official reporting came out about open AI's assistance or their agents that they were working on. So a new reporting, again, this was about two weeks ago, about a week and a half ago, just been. before Andre's announcement came out and said that OpenAI was officially working on two different

types of AI agents. So one of the agents can take over and control your device, right? You give it access to. So whether that's, you know, your desktop or your phone, we don't know yet. But one of their agents or one of their agent functionalities is taking over and controlling a physical device. That's number one. And the second kind of agent or the second agent function is being able to complete task for

you across websites and apps. Okay. So that sounds kind of like a Jarvis type model, right? And the Jarvis, if you're unaware, that kind of stems from the Marvel series and Ironman and Tony Stark where he had his Jarvis assistant that could do anything, right? He could talk to it and Jarvis can perform actions on his behalf. So this is what Andre had been hinting at in his Twitter bio. And again, the news of this preceded Andre's announcement by just one week.

So I don't like to speculate a whole lot on everyday AI because I like to always be a fact-based, right, a fact-based media company that's bringing you the news. We bring receipts. But it is interesting and you do have to think, are these two things related? Maybe Andre was not fully on board with, you know, this kind of pseudo-announcement. Maybe he had a different direction for AI agents. I'm not sure. Or maybe everything's just great, like he said.

But the timing of these things was, I'd say it had to have played into this decision. Yeah, and I agree with Svetlana, really intrigued what he's up to. Yeah, so you have to keep your eye on what Andre may be working on. You have to keep your eye on open AI and what's happening with their agents. Also related to this, it has to be related to this, is Sam Altman's out there raising seven. trying to raise $7 trillion for GPU chips. Yes, that is a trillion with a T. As in Mr. T, I pity the fool who doesn't understand what can be done with $7 trillion

to build essentially an Nvidia competitor to create these GPU chips that all these generative AI companies need to power their AI. But also, combine that. Combine that $7 trillion. building your own chips, which right now, they're very limited in availability, and then combine agents, right? And then let's talk about the future of work and talk about future versions of GPT5 and talk about, oh, when you have SORA, right?

Now you can kind of see, yeah, something pretty big is going to be happening around the corner. So you've got to keep your eyes on whatever Andre may be working on. You've got to keep your eyes on what Open AI is doing with agents with this $7 trillion. Wow. Yes. Can agree. That's a huge, huge arms race. Yeah, something, something's definitely going on. Well, here's something that's going on that's, I'd say positive. All right, another piece of AI news that very much so matters. Ready? So big tech is banding together on AI to combat election related misinformation, disinformation. All right. So this, this story kind of just kind of broke. over the weekend. So we're going to dive into it a little bit here. And hopefully this will make

sense on why it matters, but I'll still explain to you. All right. So a group of about 20 tech companies have come together to fight AI misinformation and elections. So some of those companies include, this is not an exhaustive list, but your big names of who's who in AI and social media. So we're talking about Microsoft, Meta, Google, Amazon, OpenAI, Anthropic, Snap, TikTok, X, IBM, and others, all coming together. And kind of their goal of this is to focus on combating AI misinformation and disinformation as well as deepfakes. Because according to this report, that we kind of flash on screen here from CNBC, that we've seen a 900% increase. year over year in deepfakes. And this is obviously raising serious concern about election-related misinformation and

disinformation. All right. And if you maybe aren't from the U.S., let me kind of bring you into where we are right now. All right. In the next coming weeks, we are going to start to see a lot of primaries here in the U.S. election cycle. So what that means is over the next, you know, a couple of months, the two major points, parties, Democrats and Republicans, go through a series of primary elections. And essentially,

these elections determine who each, you know, one candidate from each party that will be representing their respective party in the November 2024 general elections. All right. So for the most part, you know, it's everyone kind of knows who the candidate is, but this is, you know, still a process that we go through here in the U.S., the primary election and the general election. So you have already started to see the amount of deepfakes and AI that have already been used. You know, probably one of the most prominent ones was when there was a deep fake version of President Joe Biden that was on a robocall. and it was telling voters in the New Hampshire primary, which already happened a few weeks back, not to go out and vote, right? So that was obviously deep fake. It wasn't authorized. You know,

and there's a lot of investigations still ongoing. And since the FTC has banned the use of AI robocalls, so companies that do, that can obviously face, you know, imprisonment as well as some pretty large fines. So kind of two separate issues, but now we have the big tech companies coming together. And they're essentially of, you know, creating this accord, kind of like an unofficial agreement between these large companies. So here's a little bit more on what this agreement means and why I think it's important to talk about. And also, it is a very timely announcement, especially considering what we just saw was possible. This announcement came hours after OpenAI's SORA text a video model, right? Again, I don't think this one was a coincidence because when you have 20 companies coming together to

agree on something, presumably it's been weeks or months in the work. But still, pretty interesting that this kind of alliance was launched just hours after we saw the unfathomable, unfathomable capabilities of Open AI's new SORA model. So this accord, these members agree to eight high-level commitments, which we're going to share about more in the newsletter, but they're mainly aiming to protect the integrity of elections and combat the risk of AI generated deep, deceptive content reflecting the industry's commitment to safeguard democracy. This is a pretty huge deal, y'all. Like, especially here in the U.S., it's unlike other countries or groups of nations.

We do not have any legislation around AI. Yes, we have an executive order from the White House, from the Biden White House. from the Biden White House. We have an executive order on AI. We have, you know, these kind of new, I won't even call them laws necessarily, but you have these new guidelines from the FTC as an example, banning AI robocalls. But there is no actual legislation, right?

This is something that people don't fully understand about the legislative process here in the U.S. It moves slower than a snail. There is technically not even. Even legislation, right? So again, we have to say there's rulings from governing bodies such as the FTC, there's executive orders, et cetera, but we don't even have legislation that has gone through our Congress and has passed out on social media, right, which has been around for 20 years.

You know, we're still debating section 230 of a law that was passed in 1996 about the internet. Right. So I do think that this is a huge announcement from these tech companies. This is essentially an unofficial handshake of sorts, right? And they're saying, hey, we all understand that it's important for the future of democracy here in the United States to come together and say that we're going to do certain things. We're going to make sure we're going to do what we can what's in our control to make sure that our systems are not used explicitly to create disinformation, misinformation around elections. Because here's the thing.

These text to image models like Mid Journey v6, you can't tell the difference between that and a real photo. You can't. I used to be a photographer of sorts. I've taken probably 250,000 photos. I have a bunch of DSLR sitting in a closet somewhere. You can't tell the difference. I don't care what anyone says.

You can't tell the difference between someone that's really good at Mid Journey. And if they're creating political images, you can't tell the difference, especially the average person can't. Text to voice, you can't tell the difference. The models are too good, dangerously good, right? And now we're seeing the first, what I think, is a usable model for text to video in SORA from Open AI. Can you still tell it's AI generated? Yeah, you can.

Right. But, I mean, if we were talking about models that existed last week, you know, runway and PICA, I don't think you would ever look at what comes from those programs and think that it might be real, right? Whereas SORA, you might. So again, we're not sure when that's going to become publicly available, but I think it's very timely. This agreement between these 20 companies, I think is extremely important for keeping the elections here in the U.S. as safe as possible. But don't get me wrong, it is still going to be sheer and utter chaos.

I've been saying this for more than a year that the 2024 elections, I still think, is going to be people's, a lot of people's first experience with generative AI. Because even though they're banning these use cases, people are still going to use it. And it's going to be hard to track down. And it's going to be hard to police their mismanagement. So I still think this is going to be extremely problematic for the 2024 U.S. elections. However, huge steps. So you have to give credit where credit is due to these large companies, all putting aside competitive differences, right?

Because one of these companies could have said, ah, no, you know, we're not going to, you know, sign on board with this. And that, in theory, could have given them a competitive advantage, right? Even if people were using it for bad purposes, you know. So I do love that all big companies in social media and in tech came together as part of this agreement to protect the elections here in the U.S. All right. Our last piece of AI news that matters this week. Reddit has just reportedly inked a new AI content deal. This is huge.

All right. So, let's talk about this. So according to a Bloomberg report, Reddit has entered a new licensing deal with an undisclosed large AI company estimated to be worth about $60 million annually. All right. So this move reflects a shift in how AI companies are acquiring data moving away from using open web scraping without permission to establishing formal agreements with platforms like Reddit. All right. I talked about this a couple months ago. The ongoing legal dispute between the New York Times and Open AI and Microsoft. So in this lawsuit, the New York Times alleges that Open AI has used millions of its pieces of copyrighted materials to train its models, right?

And to oversimplify this, right, because I don't have an hour to explain the intricacies, but essentially so many generative AI models are just trained off the history of the open internet, which does obviously include a lot of copyrighted work. So it is very hard for any company out there. there to make an argument that their large language models or their generative AI models are not trained on copyrighted works because most of them are. Don't get me wrong, I think there are companies that are taking a much more cautious approach. You know, I think Adobe is taking a much more cautious approach than others. But I think what we're going to see is a lot of deals

like this, a lot of deals like what Reddit just saw, where you're going to have these companies signing what look like, man, is this, this looks like professional athlete contracts. But if I'm being honest, I think whatever company signed this deal with Reddit, got a great deal. Got a great deal. $60 million annually, it might seem like a lot. But when it comes to reliable information and when it comes to SEO value even, if you use the internet, Reddit, I think has become a lot of times, if you want to say, you want to save time. And if you want to get reputable information, a lot of times people find that on Reddit,

right? The SEO world has become a mess. So you might read 10 articles, you know, when you do a Google search or you do a Bing search. You might read 10 articles and seven of those might just be, you know, companies that have just gained the SEO system and just a bunch of unrelated garbage. So the content on Reddit is extremely valuable. Right. So I think a. $60 million deal is actually a great deal for whichever large language model company sign this. So it could be OpenAI, it could be Google, it could be Microsoft, it could be Amazon, it could be anthropic. I would say it's one of those, but I'm sure we'll find out in the coming days and weeks. One other thing to keep in mind about this is it's going to set the stage, right?

whatever happens both in this New York Times versus Open AI lawsuit, which I assume will be settled out of courts. I don't think it'll actually go to a judge or a jury trial. I do believe it'll be settled. But I think that's going to set the tone for what large language models look like in the future. Because yeah, you're going to whatever happens between that case and what's happening here with this agreement between a large language model company. and Reddit, I think that's going to be what it looks like going forward. So so many publishing companies I've talked about this on the show before, there's essentially very limited options on what happens, right?

Either companies will just go out of business because they're losing traffic, because instead of people going to their website, they're just using AI search engines like perplexity, like Google's search generative experience, SGE, using co-pilot, or using chat GPT. So people aren't visiting these websites as much. So either one, these publishing companies are online, you know, information providers will either go out of business. Number two, they'll sue all these large language model and big tech companies for scraping and using their copyrighted data. Or number three, they're going to see partnerships like we just showed here with this partnership, this reported partnership between Reddit and whatever company this is.

So I think there's going to be a lot of individual partnerships like this, as well as. I've called for this. I don't know if a company's out there doing it. Hey, if you want to make a bunch of money, go do this. Create a union for the tens of thousands of high quality websites that are putting out great content online and create a union. Create a union of online content producers. And there, that's, you know, then you can work out large agreements with these

companies to use all of this data. All right. That's it. As always, there's a lot more in our newsletter. So make sure you go to your everyday AI.com, sign it for that daily newsletter, as well as if this was helpful, which I hope it is. Y'all, sometimes we spend two, three, ten hours researching and producing one single episode, which you can probably guess that's a lot for a daily show. So we would really appreciate it if you would repost this, share this if you're, you know, tuning in on social media, repost this, retweet it, tag your friends. keep them in the know because here's the thing.

I think with generative AI, there's a lot of bad information out there. So many people do not care about the facts. They don't care, right? They just say, hey, here's, you know, 500 chat GPT prompts to make you rich. And then they try to sell you some crappy software. That's not what we're about here at everyday AI. We care about you understanding generative AI. We want you to know how to use it, how to understand it, to grow your companies and to grow your career.

So if you could share this, we'd really appreciate it. And also make sure you tune in tomorrow. I think there's something, a big impact on this new Open AI SORA model that people aren't really realizing. Yes, the quality is amazing. It's going to change the text of video. It's going to change storytelling, but there's something much more than that. So thank you for tuning in.

Thank you for listening, watching, reading. And we'll hope to see you back tomorrow for that show and more on Everyday AI. Thanks, y'all. Meet Firefly AI Assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface.

You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com

and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.