EP 359: AI News That Matters - September 16, 2024

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.

Apple is rolling out its iOS 18 today, but it's not going to have any AI in it. OpenAI, speaking of rolling out, just rolled out a new model of its kind and I think people are getting it wrong. And Oprah had some of the most famous people in tech talk about AI. We're going to be talking about those things and a lot more today on Everyday AI. What's going on, y'? All? My name is Jordan Wilson and I'm the host of Everyday AI. This thing is for you. It is your daily livestream podcast and free daily newsletter helping everyday people like you and me not just learn generative AI, but how we can keep keep up with it, use it to get ahead in our companies and our careers. So if that sounds like you, if you're listening on the podcast, first of all, thank you. Appreciate it. Make sure to check out the show notes for a link to our website. That's your everyday AI dot com. Sign up for the free daily newsletter. Yeah, we do this every single day. Not only do do we do the live stream podcast, but we have a newsletter every single day. So if you miss it, you can always go back and catch it there. All right, speaking of our website, it is a free generative AI University. Hundreds of episodes. No matter what you care about, you can go find it there. All right, every single, almost every single Monday, we do this. This is the AI news that matters. Because if I'm being honest, you can spend hours every single day easily going into deep rabbit holes about, you know, what's happening in the world of AI. But you know, you should probably instead just save all that time and join us on Mondays when we bring you the AI news that matters. And yes, Jay, it is Monday already. Shout out to our live stream audience. Joe, Brian, Marie, Michael, Rolando, Madon, Christopher, Fred, everyone else, thanks for joining us. If you're a normal podcast listener, come hang out on the live stream. It's fun, meet some cool people, but let's just get straight into it and go over the AI news that matters for this week. All right, first, yeah, new iOS dropping today. Guess what? Won't be dropping any AI inside of that. So Apple has unveiled a timeline, according to reports, for rollouts of its AI features and its upcoming iOS 18, highlighting a structured approach to enhancing user experience with Siri and other applications. So iOS 18 will be available to the public today. But the much anticipated Apple Intelligence features will not arrive until October at the very earliest with the release of 18.1. If you follow the show. I've been saying this all along. So the iOS 18.1, that's the one you got to pay attention to. So that will introduce some key AI enhancements, including the improved Siri capabilities for understanding follow up questions and various other AI powered tools like like the notification summaries and suggested replies and messaging apps. I think those are going to be the things that most people are using. Kind of the smarter quote unquote AI powered Siri and then the contextual awareness in all of Apple's different apps. Then following this we will have 18.2 which is expected to launch in December, bringing two other features that Apple's really been hyping, but I don't really see the point of them. Image Playground, which allows users to create some cartoons tunish photos from images and descriptions and Genmoji, which generates emojis based on text input or existing photos. So yeah, I'm not sure why so many people are paying attention to those two things. An AI image generator on your phone. Okay, that's cool. We already have a lot of them that do a better job these. The Apple results look rather cartoonish, but I think that is by design. And then the Gen Moji also Apple plans to expand the availability of these features to additional English speech and English speaking countries such as Australia and Canada, starting with that 18.2. So yes, the 18.1 Apple Intelligence updates will just be in the United States right now. So looking ahead, iOS 18.3 is expected to be a more minor release with no confirmed new AI features for US users, while iOS 18.4 is slated for March of next year and will introduce a more personal sounding Siri along with potential support for additional languages. So yes, according to reports, even the Siri Apple intelligence rollout seems to be two tiered. So in the 18.1 we will presumably get the more contextually aware Siri. What that means is the version of Siri that will know kind of what's in your calendar, what's in your text messages. Right. As long as you opt into those features where the more neural voice Siri that Apple has been teasing will reportedly not land until March. And again these are at the earliest dates, right? These are if everything goes right. So yeah, I would love to hear from our live stream audience because I'm curious. I'm an Apple user right now. If you're on the live stream, I'm Actually using my iPhone. I'm on my, you know, Mac, on a Mac Mini here. Apple, I'm not sure about this strategy, right? They've waited this long. They've waited a year and a half, Apple after Microsoft, right? And then they're still rolling this out in tears, right? Which I get. It's a very calculated approach. There's a lot going on under the hood. But after all of this fanfare, right, the WWDC announcement in June, the Apple Glow Time event, you know, this September, and then it's going to be a very gradual rollout. I'd really be interested to see how their stock even responds, because a lot of people aren't like you and me, they're not going through the details of this, right? Essentially, what a lot of people have heard is, oh, new iPhone 16, Apple Intelligence, and we're going to see this new iOS 18, right, roll out. So a lot of people are going to be confused today when there's no AI features. And people will be even probably more confused right, when they start getting their new iPhone 16 and you can't get any Apple Intelligence features. Given that is, I think Apple is putting so much of their future, at least in hardware device sales, on Apple Intelligence, right, On their new AI features. Their new edge AI is, you know, they have a small language model on board that can handle a lot of these things on device, which is great. It's faster, more secure. But here we are, like I said, a year and a half behind Microsoft, a year and a half behind Google, and we're still getting a tiered approach. I don't think Apple did a good enough job kind of mentioning those things in its keynote address last week. However, we'll see. We'll see, right, if it all works great, you know, and I think people are going to forget the fact that Apple is so far behind. But, you know, even me personally, I'm like, I should probably be on a Windows computer right now. I should probably be using Windows Copilot outside of, you know, the Copilot versions that they have available for Mac. So, you know, yeah, Michael says team Android. All right, well, hey, we'll see. We'll see what happens with Apple. All right, our next piece of AI news. This one, y', all, people are sleeping on this, but Google actually has a great AI product for consumers. And if you like podcasts, which you probably might, since you're listening, you're probably gonna like this one. So Google has announced the new updates in Notebook LM and its audio overview features for enhanced Note taking. So Google and its Notebook LM product have launched a new feature called Audio Overview, which transforms documents, slides and charts into audio discussions, making it particularly appealing for students and professionals alike. So I would say this innovative tool aims to enhance the way users engage with their notes and study materials. So the Audio Overview feature allows users to generate audio discussions by clicking a new Generate button within their notebooks, facilitating a dynamic exchange between two AI hosts. Yeah, right. Why am I telling you this? Well, hopefully you don't just go copy and paste our website and, you know, have these two host AI host in Notebook LM take over my job. But I'll tell you this, it's extremely impressive, if I'm being honest. Aside from the plus version of Chat GPT. Right. I wasn't super impressed with the original Chat GPT that came out in November 2022. Chat GPT Plus, I was super impressed with right when we were able to access the Internet, you know, there was, you know, plugins, all of those things. I haven't been, if I'm being honest, this impress with a consumer AI offering since then. It has probably been one of the more impressive. And I've used hundreds, if not a thousand AI tools since that point. I am impressed with this. So yes, it is powered by Gemini, but it gives you great citations. So think, you know, taking a bunch of, you know, copy and paste text, a bunch of your files, links, and then being able to go in and ask right within an AI chatbot interface. So it gives you the normal kind of AI chatbot response. But then you can also generate a short podcast based on that conversation. Right. And it is very impressive. The voices are great. Yes, it is a newer product, it's in beta, so, you know, there could be a hallucination. So always keep that in mind. But this Notebook LM actually does a great, great job kind of citing where everything is coming from. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Gen AI. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too, just like some of the biggest companies in the world do. Go to your everyday AI.com partner to get in contact with our team. Or you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on gen. All right, let's keep this AI train rolling. So the White House has launched a task force to address AI power. All right, here's what that means. The White House so they just announced a new task force aimed at addressing the growing demands of AI infrastructure, highlighting the increased significant of generative AI technologies in the US economy. So this initiative comes as tech companies expand their data center operations, raising concerns about energy consumption and environmental impact. So this task force is led by the National Economic Council and the National Security Council and it will focus on developing policies for data center growth while balancing economic, national security and environmental priorities. So some high profile tech executives, including OpenAI's Sam Altman and Google's Ruth Peratt, have participated in a recent meeting to discuss the infrastructure requirements necessary for advancing AI operations. The meeting addressed challenges to clean energy permitting processes and workforce needs for the development of data centers and power infrastructure. Generative AI, as all of you all know, it's, you know, it's capable of producing anything, text, images, videos, and it's creating a lot of excitement for its potential, but also raises concerns about misuse and its environmental impact. Yes, all of these generative AI systems that we use are very resource heavy. So OpenAI, probably the most visible company in the AI space in the world, has expressed that infrastructure development is crucial for the country's industrial policy and economic future, reinforcing the belief that infrastructure shapes destiny. All right, speaking of AI tools, we saw some new previews from Adobe. So Adobe has unveiled its new generative AI video tool, including an image to video feature Adobe recently released and teased, it's sorry, recently teased, not released, its upcoming generative AI video tools, which promise to change how Adobe customers edit video into post production. So this development is pretty significant. Significant as it builds on the company's Firefly video model, showcasing advancements in AI capabilities that could greatly benefit content creators. All right, so here's some of the new features that Adobe has teased for its new Firefly video. So it includes features that can produce video clips from still images and generate videos using text descriptions. Also, the new tool updates will allow users to to simulate various camera angles, motion and shooting distances, enhancing creative control over the final product. So users can also choose preset filming styles, making it easier to create visually appealing content without extensive technical knowledge. Also, this one's pretty important. The Firefly model is being positioned by Adobe as, quote unquote, commercially safe as it is trained on openly licensed public domain in Adobe stock content, potentially alleviating copyright concerns that have come along with other AI tools such as midjourney, Runway, etc. So Adobe plans to release the Text to Video and Image to Video features in beta as standalone Firefly applications later this year, with further integrations into Creative Cloud, Experience Cloud and Adobe Express Applications. Here's where it's not going. At least we haven't heard anything, and this is important to note. I haven't seen anyone talk about this since April. So in April, when Adobe first announced Firefly Video, they previewed that they'd be eventually integrating with OpenAI's Sora model. So OpenAI's yet to be released text to video platform, which looks about three times as powerful and as good as every other AI video tool out there. But we haven't seen anything yet from Adobe on that since the initial announcement since April. So this could be a potentially a shift in strategy from Adobe. And I think they caught a lot of positive press back in April when they announced the potential SORA integration into its Creative Cloud products, but nothing yet. So, interesting one there. All right, next, let's talk Oprah. You get an AI, you get an AI, we all get an AI. All right, Oprah had her AI special highlighting some urgent insights on the rapid evolution of AI. So in a recent special with leading figures in technology and law enforcement, Oprah Winfrey addressed the fast paced advancements in artificial intelligence and their significant implications for society. All right, here's kind of a roundup of who said what. So Bill Gates, maybe you've heard of him. He revealed that he thought AI development is outpacing expectations, suggesting societal impacts may arrive sooner than anticipated. OpenAI CEO Sam Altman warned against complacency in the AI space, emphasizing the economic changes driven driven by AI that are still unfolding. And both Gates and Altman discuss potential job market disruptions, with Gates predicting a shift toward shorter work weeks. Oh, that'd be great if only we were all billionaires and could afford that. While stressing the importance of human skills in creative and interpersonal roles. You also had a lot of other people. So tech YouTuber Marques Brownlee showcased AI's capability to generate realistic content, kind of, you know, showing some examples of it live on Oprah's special, but also raising some concerns about misuse and misinformation. And speaking of misuse and misinformation, that is the crux of what FBI Director Christopher Wray talked about in his conversation with Oprah, highlighting some alarming uses of AI in criminal activity, including exploitation and enhancing criminal capabilities. So a recurring theme from Oprah's special was the need for regulation and collaboration between tech companies and governments, with Altman advocating for safety testing similar to aviation standards. Gates called for a shift in how taxes and corporate responsibilities are approached to focus on societal impacts. Also, despite concerns, the discussion also pointed to AI's potential to change education and healthcare, with Gates envisioning personalized AI tutors and improved doctor patient interactions. All right, I'm curious. Did y' all catch Oprah's special? I think this is worth talking about because this is probably, you know, I talked about this with Apple Intelligence. I think Apple Intelligence is going to be the first kind of hand in hand interaction with AI that maybe billions with an ass, billions of people have. Right? If you, if you've been listening or watching to, you know, everyday AI, you're technically in the minority here, right? Sometimes I feel that, oh, everyone in the world knows AI, everyone uses it daily, but that's obviously not the case. So I think Apple Intelligence will actually be a lot of people's first interaction with generative AI with large language models. But I think Oprah's special is actually nice if you've never heard of anything AI, Right? But that's not any of you. So if you missed Oprah special, I'll tell you this, you didn't miss any new information that you don't hear here every single day on everyday AI. But if you do have, as an example, if you have, you know, a parent, a relative, a friend, a coworker who knows nothing about AI, and obviously you told them first, well, hey, you should check out everyday AI to learn. But if they still haven't, maybe this is a good kind of an introductory to generative AI for everyone else that doesn't keep up with it on a daily basis. So let's keep going. There is more. And obviously we're saving some of our juiciest news for last. All right, so next, US Senators have called for an investigation into AI tools that that summary summarize content online. Yeah, this one is very interesting. So a recent push for an investigation into AI tools that summarize online content has been brought to light by US Senators citing some significant antitrust Concerns, especially regarding their impact on content creators and publishers. So U.S. lawmakers are urging the Federal Trade Commission and the U.S. justice Department to examine whether generative AI tools and features are exasperating existing challenges faced by journalistic outlets, which are already struggling with consolidation in layoffs. So major platforms like Google, Google, Google and Meta, yeah, we're live. This is live and unscripted, y'. All. We don't edit this. So major platforms like Google and Meta are noted for generating substantial advertising revenue from original content created by others. While these programs and tools that generate AI summaries tend to keep users on those platforms rather than directing them to the original sources, publishers face a difficult choice. To avoid having their content summarized in AI outputs, they must opt out of being indexed in those search engines entirely, leading to a drastic reduction in referral traffic. So yeah, it's almost like publishers can't really take that option because they still need search engines to survive and all of these other bots that, you know, essentially crawl their website. But then what's happening is now all of these tools are just summarizing the most important content and all of these major publishers aren't getting clicks in. This scenario obviously creates a troubling environment for content creators as they have little recourse to profits for from AI generated summaries that utilize their original work. So the ongoing discussions regulating AI's role in content monetization are timely as many in the media industry express growing concerns over the potential monopolistic behaviors of these large tech firms using these generative AI technologies. The outcome of this investigation, if there is a formal one, again, right now this is just US Senators urging the FTC in the Department of Justice to look into this practice. But the outcome of this investigation could have long lasting implications for how AI interacts with original content, potentially leading to new legislation that addresses the balance of power between content creators and major platforms. Yeah, this one is tough if I'm being honest, right? Even for me personally. I took a video of a website that I was on yesterday because there was so many ads. There were so many ads on it, right? I was, I was reading an article last night while preparing for this show today. There were so many ads on the website, I couldn't even get out of the ads. There were too many of them. I was on a smaller laptop, you know, like a 14 inch laptop, but there was a pop up ad, there was a full screen overlay ad underneath that there was a header and footer ad, multiple side, but I couldn't get to the Content, Literally I couldn't get to the content as hard as I tried. And that, you know, I talked about this probably about a year ago, how I think that it's actually the SEO industry that has killed the Internet and, and how all of these AI overviews will essentially maybe save us all. But I get both sides, right? So using tools like Perplexity, using tools like AI overviews from Google and using tools like Search GPT from OpenAI, I'll say, you know, once Search GPT rolls out to everyone, I think those will be the three main kind of AI summarizers out there on the market. And I even personally have mixed feelings on them. Right. Personally I love them. I honestly can't get by my day to day without using tools like Perplexity or AI Overviews from Google. Right. Because going to all of these websites is a major pain. But I mean even, Even your everyday AI, right, our website, your everyday AI.com, sometimes I, you know, will put things into Perplexity or AI Overviews by Google and I see it getting content from our website, right? It's kind of funny. I talk about AI so much, I forget probably more than I talk about. But I'll see our website come up in these AI overviews all the time. And then I think, okay, yeah, I appreciate that I didn't have to go to these 3, 5, 10 websites, but what about when one of those websites is mine, right? And someone instead goes to Perplexity and they get the answer on something about AI or about ChatGPT versus visiting our website and then discovering, you know, our free daily newsletter and discovering, you know, all these other podcasts that they can listen to in our tutorials. Right? So it is, you know, kind of a love hate relationship with this technology, but pretty, pretty significant that we have US Senators calling for an investigation specifically on tools that summarize content online. Does this have to do with Google and Meta losing money as publishers and these companies, you know, having great presences in dc? Yes, but it is something that, you know, we're going to be talking a lot about the rest of this year. And hey, for our, for our live stream audience, what are you, are you team? Are you team? Go to the website, read it. Maybe you use an AI tool to summarize something when you're on the actual webpage. Or do you not even go to websites anymore? Are you just, you know, go on Perplexity on, you know, Google's AI overviews? Maybe Copilot has some great, you know, inside Bing, they have some Great AI powered features. Yeah. What, what team are you on? What are your thoughts on that? All right, pretty big one here. Salesforce has launched Agent Force, a new era for autonomous AI agents in business. So Salesforce has unofficially unveiled. After a lot of reporting in the last two weeks, they've officially unveiled Agent Force, a suite of low code tools designed to help enterprise create autonomous AI agents capable of executing tasks in sales, service marketing and E commerce. So Agent Force combines three tools, the Agent Builder, Model Builder and Prompt builder, providing a robust infrastructure for developing autonomous AI agents. And unlike traditional chatbots, these AI agents can make decisions and take actions independently based on your company's data, enhancing efficiency and various business functions. So Clara she, the CEO of Salesforce AI, emphasized that while previous AI capabilities focused on content generation, Agent Force will enable users to handle higher order questions and execute them autonomously. Also, Salesforce is rebranding its Einstein copilot as an agent for developed. Sorry. Agent Force developed agent enhancing its capabilities to retrieve data, reason, plan and take action. So, out of the box, out of the box agents will be available, including a service agent that can manage various service issues without predefined scenarios, aiming to improve customer service efficiency. Also, additional agents are tailored for merchants and buyers and those recommendations and assistance with purchases. So Agent Force for services and sales will be available starting October 25th of this year with some components of the Atlas reasoning engine launching in February of 25. Here's one that's a little bit of a head scratcher for me. The pricing model for this starts at $2 per conversation with volume discounts available, reflecting a shift toward consumption based pricing and AI agents. So Salesforce has actually established an Agent Force partner network, collaborating with companies like aws, Google and IBM to develop agents and actions for the Salesforce app exchange. So we talked about this last week as well because the Salesforce CEO essentially said they were making a hard pivot, kind of going from a traditional CRM company, which they've been for many decades, and they are probably the biggest name in the world by far when it comes to CRM and managing customer relationships. But they've essentially said we are doing a hard pivot and we are now an AI company. Right? I've talked about this a lot and I've said you have to look at as an example, if you're wondering if your company should go all in on generative AI, I always said look at the six biggest companies in the United States, right? You have Microsoft, you have Apple, you have Amazon, you have Meta. You know, I'm missing one or two others, but Nvidia, Tesla as well. So all of these companies are, have already gone all in on AI, right? So some of the biggest tech companies in the world have already gone, investing usually tens of billions of dollars both internally and extern. And yet now we have some of the biggest names in business, right. In Salesforce essentially saying, hey, we're doing a 180. We're no longer a CRM company. We are an AI company. So we'll see how Agent Force is received. You know, they have. Salesforce has their Dreamforce conference this week. Also, if you are reaching out, we were originally going to be at the Dreamforce conference. It didn't work out. So if, if you're reaching out and wondering, oh, oh, why is everyday AI not there? Something came up with the scheduling, but it should be interesting to see if this $2 per conversation model, if that actually works, I'd be interested. Right. Obviously Salesforce is putting a lot of confidence that this is going to work. Right. Imagine your sales team, right? I mean first of all, having access to up, up to date data. That is the, the, the, the golden standard right now. So when we're talking about all of the out of the box large language models, right? So your chat, GPT, Gemini, Copilot, claudianthropic. Right. One of the biggest downsides is not having up to date dynamic data from your company or your clients, your customers. Right? So Agent Force here from Salesforce is looking to be that solution, right. I'd say Copilot is actually much better with that since it has, you know, it lives on your desktop. So Agent Force, if this works, it could be one of the most widely used AI tools out there in the business world. Right? So I understand Salesforce's pivot here. I don't understand their $2 per conversation model. If obviously I'm sure they tested this and know a lot more than I do, but I think a lot of business owners are going to look at that and they're going to say, okay, well if I have a team of, you know, a hundred salespeople and they're using this thing all day, it better pay off pretty quickly, right? Your sales better shoot through the roof because you could very easily all of a sudden be spending tens of thousands of dollars a week if you have a large sales team. So you have to be able to, number one attribute this to, you know, Agent Force versus what, you know, your agents would, your sales team would normally bring in. So pretty, pretty bold, pretty bold move there from, from Salesforce. All right, our last AI news story of the day. Open AI has launched two new models, 01 Preview and 01 Midi Mini. And I think these are paving the way for just how models will be built in the future. All right, so OpenAI has made headlines with the release of its new 01 Preview and 01 Mini models which were, which were released Thursday and are being touted as significant advancements in AI technology. So yeah, if you've been following along over the last months, we've been talking a lot about this Strawberry, right? Previously it was codenamed as Q Star. Essentially a model that reasoned in that kind of code name turned into Strawberry. So this is essentially Project Strawberry has been released and this is a model that reasons under the hood much like a human would. So, you know, if you're used to just going into a generative AI tool like Chat GPT putting in a prompt in literally within a second you might have hundreds of words already. This is not how 01 and 01 mini work. These are reasoning models. So I did a show on Thursday specifically going over this. It was actually a bonus show. We did two shows that day. So if you didn't check that out, make sure to go check it out. But essentially it thinks, right? It takes anywhere from, you know, 10 seconds to sometimes two minutes before it gives you a response and it essentially goes step by step and you can see a little bit of what's going on under the hood. And it thinks more like a human, right? It shows some ability to reason. So this is obviously a departure from the rapid outputs of every other AI model out there right now. So currently the O1 model is available to chat GPT plus and Teams users, while the O1 mini is expected to be accessible to educational and enterprise accounts soon, with free access to a one mini anticipated later. So yes, right now the only people that can access this are people on the either $20 a month chat GPT plus or the, I believe it's 25amonth chat GPT teams. But right now enterprise clients, educational clients and free accounts do not have access to these two new models. So these models represent a new class of reasoning based AI distinct from the existing GPT family and are considered a step toward artificial general intelligence. So, all right, I'm going to save, I'm going to save my little rant here and give you more of the details. So the pricing for, for these models for API customers is a little bit higher. So with the Zero1 model is costing $15 per million token input and $60 for output compared to the more affordable rates of, as an example, GPT4O or 4OH mini. But it is worth noting this reasoning model is actually much cheaper than than the GPT4 model was when it came out for API customers. So right now if you are accessing it via the front end though, you don't have to worry about pricing. Right. You're not paying anything additional. But it is extremely limited in the number of messages. So in terms of usage limit, users of the O1 model can only send 30 messages per week. Rough. While the O1 mini model allows 50 messages per week, which can obviously restrict frequent usage. Yeah, so I actually have multiple paid accounts, you know, we have enterprise accounts because we help businesses, you know, businesses hire us to help them, you know, use ChatGPT across their company. So I've been checking in, you know, different, you know, capabilities or access yet. So yeah, it is still only those chat, GPT plus and Teams account and it's very limited, but it is very powerful. So this is important to note as well currently whether you are a Tier 5 API customer and right now that's the only people who can access it. It is tier 5 API user. So if your company does use the API but you don't use it a lot, you probably don't have access to this yet. But it is just 01 mini and O1 preview because there is actually a more powerful variety of the O1 model that OpenAI has not released yet. So yes, there is technically three new models and they're technically in a class on their own. More on that here in a second. So right now we only have access to Zero1 mini in Oh1 preview. But if we're talking about just the Zero1 model model that is not available yet, that has far surpassed human level performance and it has achieved a score of 92.3 on the MMLU benchmark, which is essentially, I tell people it's the act for large language models and it has greatly outpaced previous leading models, indicating a significant leap in AI capabilities. Yeah, so for the past like six months, the most powerful models in the world. So GPT4O Claude Sonnet 3.5 Gemini 115 on this MMLU benchmark. We've essentially been stuck at this like, you know, this score of like 87 to 88. Right. And we've been kind of stuck there for the better part of, you know, three to six months and everyone's like, oh, you know, it's the AI Winter AI is not going to continue to develop and I'm like, y' all don't understand anything. And clearly 92.3 is wild, right? That's wild. So a lot of studies show that the average educated human, right, me and you, if we took the MMLU, we would probably, probably get a 40, you know, and even the smartest people in the world would not get a 92. So you know this is already bordering, right? Depends on your definition of what is artificial General intelligence, what is AGI. But this is kind of bordering it, right? And that's why people are saying this is a significant step, essentially a model that reasons is is pretty impressive. And also I've talked about this. I think the introduction of these models is now going to trigger a competitive response from other May I major AI players including Anthropic and Google. So here's what I think is going to happen probably after the election here in the US in November. Well, probably before that, Anthropic has got to drop their prices on the API right now. Because if you are using Anthropic's API, you're probably looking at yourself and saying like why you're paying this. Depending on what model, you're paying the same amount or maybe a little bit more to use Claude's API when right now Google's API with their Gemini is so cheap and so powerful and now with this new model from open AI, it doesn't make sense. So I do think that Anthropic is going to have to drop their API pricing and probably respond with a 3.5opus probably after the election, you know, whether they do it yet in late 2024 or early 2025. And I do expect Google to make some front end implementations as well. So Google, don't get me wrong, they've been great on the back end for developers on the front end, their Gemini technology on the front end. So if you're just using it, you know, as a quote unquote front end user, right? You, you log into your account on the front end of Google Gemini and you're chatting with the AI chatbot. It's working on a model that is like six months old, right? So I do think that Google is going to have to start pushing their more powerful models that they have in their sandbox to production. I think Anthropic is going to have to drop their prices and also release a new model because otherwise OpenAI is going to run away with this. Especially when we know that they are probably going to have a GPT 4.5 or maybe eventually a GPT 5, you know, up their sleeve. That's the other thing I need to talk about. People are getting this wrong. Everyone, even the quote, unquote, smart people in AI, everyone is calling this model GPT01, right? Oh, GPT. Right. The smartest people in the world, right, that are, you know, keynoting conferences for Fortune 100 companies. That's wrong. This is not GPT01. All right? It is a different model. It is a different class of model. OpenAI literally said this on their blog post. I don't know why people hate reading or hate educating themselves, right? So this is not GPT01. It is a separate class of model. OpenAI said as much. They are separating this and they are going to continue to essentially update. They are calling it open AI01. So that is kind of one class of models. And this is presumably will continue to be their reasoning class of models, and then they will continue to make updates on the GPT5 or, sorry, the GPT class of models, right? So right now we have four. Oh, so we may see a 4.50 or a, you know, eventually a GPT5, but they will be continued according to OpenAI to be developed independently. Right? And we talked about last week and the week before, some reports from the information that said Open AI was floating out potential subscription prices that would potentially be hundreds or maybe a thousand or $2,000 a month or more. And people were like, losing their minds. And I'm like, y', all, this is. This is what we're talking about right here, right? If you have a model that can reason under the hood, right? The oversimplified way of saying it is it does a version of chain of thought prompting on the front end. It does that for you under the hood, right? That I think if OpenAI does, you know, have a tier, right, that they're maybe charging a couple hundred dollars a month for, it is going to be this one, right? So that is why right now we essentially get. What. What does that boil out to? You get a handful of prompts a day. That's it. Right? So I do assume that once this model, the O1, once we have kind of unlimited access to it, I do see that 20amonth price, not really including any of it. Right. It's going to include very limited version, but if you're going to want kind of an unlimited version of this model or a version with way less limits, kind of like what we have right now, you're probably going to have to be paying a lot more. All right? And one Last small little tidbit on this, this is a very small study, y'. All, but I, I think it's worth mentioning because people are going to be talking about it a lot. So OpenAI's new Model 01 has demonstrated a significant improvement in reasoning ability, scoring a 25 out of 35 on the Norway Mensa IQ test, which suggests a real advancement in AI intelligence rather than just access to specific training data. So this testing, and we'll link to it, you know, in our newsletter today, indicates that AIs may now have an IQ range between 100 and 120, while a lot of other models, at least according to this test, the Norway Mensa IQ test, while other models are probably more in the, you know, 80 ish range, 80 to 90 range. So at least on this one very. And this is a small scale test. But I wanted to talk about it because people are going to be talking about it, right? Because for a model, a large language model to, you know, on IQ tests and this is a very small study, so take it with a big old grain of salt. But for it to be essentially having a 40% higher score on these IQ tests than other class leading models. Pretty significant. Pretty significant. How many Rs are there in the word strawberry onto NUS? Well, it depends on which model you ask, right? If it's two or three. All right, so that is it. Make sure to also while you're@your everydayai.com Sign up for the Thanks a million giveaway and let me quick give you a very high level overview of the news that matters this week. So so Apple is set to release its iOS 18 today, but you will probably not have any AI features until 18.1 which will likely come at the end of October and 18.2 which is expected to roll out in December. Next Notebook LM from Google, empowered by Gemini now has an audio overview feature which I think is probably one of the best AI tools I've used in a year. The White House has launched a task force to address this whole issue of power consumption with big tech companies and AI models. Speaking of big tech companies and AI models, Adobe has unveiled some new features for its generative AI video tool, Firefly Video. OpenAI or sorry, Oprah highlighted OpenAI, Bill Gates and a lot of others on her AI special on Thursday. Really just pushing a lot of concerns, but I think it missed the mark on a lot of things. Next U.S. senators have called into an investigation for AI tools. That summary summarize content online such as the perplexity the Google AI Overviews of the World Salesforce continues their hard pivots from being a CRM company to an AI agent company as they officially unveiled Agent Force Open AI last but not least has unveiled and released it is live for ChatGPT plus and ChatGPT Teams users. It's 01 and 01 mini and 01 preview models. So again this is the Q Star Strawberry this more reasoning model. It is out. It is live. Go check it out. Let me know what you think. Speaking of check things out, go check out our website, your everydayai.com check out the daily newsletter we'll be putting out right after right after this. And please join me tomorrow and every day for more Everyday AI. Thanks y'. All.

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.