Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 334: AI News That Matters - August 12th, 2024

Episode Date: August 12, 2024

Win a free year of ChatGPT or other prizes! Find out how.Is OpenAI actually dropping a new model this week, or is it just rumors? Will Perplexity's big payday help it catch Google? Why did Dell cut 12,000 jobs? We'll answer those questions and more on Everyday AI's weekly Monday AI news recapNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Ask Jordan questions on AIUpcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:1. Trump's False Claims and Misuse of AI2. OpenAI Updates and New AI Models3. Perplexity's Growth and Funding4. Apple's AI Strategy and Developments5. Dell's Focus on AI and Industry ImpactTimestamps:00:00 OpenAI's new model, Dell job cuts, AI impact.05:37 OpenAI announced search GPT, competitor to Google.09:28 Apple partners with OpenAI for AI integration.10:48 Bloomberg report predicts 3 years for advanced AI.16:18 Dell creates AI unit, plans job cuts.19:02 AI adoption raises stock, but job impact uncertain.21:18 Trump alleged fake AI-generated crowd at Harris event.24:33 Creating realistic content will become increasingly challenging.28:46 OpenAI introduces powerful chatbot with advanced reasoning.31:42 OpenAI co-founder leaves, lawsuits filed, internal turmoil.35:46 OpenAI's powerful model may be unsafe.41:16 Chatbot emulating human voice raises concerns.43:34 AI news recap: Perplexity raises $250M, Apple's AI plans.46:25 AI news, live stream, podcast, newsletter, subscribe.Keywords:Jordan Wilson, AI-related situation, US elections 2024, Donald Trump, AI-generated crowd, Kamala Harris, disinformation, AI content, OpenAI, Sam Altman, AI model "strawberry", executive changes in OpenAI, advanced voice modes, AI misuse, AI news, Perplexity funding, Apple's AI features, Dell's job cuts, GPT 4.0 security risks, AI model QSTAR, OpenAI lawsuit, voice emulation, Everyday AI, Sequoia, NVIDIA, Databricks, Bezos Expeditions, Search GPT, Apple Intelligence, AI job loss.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 and 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. Is Open AI actually going to be dropping a new model tomorrow?

Perplexity just cashed in big time. Does that actually mean anything? Are they going to catch Google? And why did Dell cut more than 12,000 jobs? And what does AI have to do with it? All right. We're going to be answering some of those questions and a lot more today on Everyday AI. What's going on, y'all? My name's Jordan Wilson, and I'm the host of Everyday AI,

and this thing is for you. It's your daily live stream podcast and free daily newsletter, helping us all keep up with what's going on in the world of generative AI and how we can use it to grow our companies and to grow our careers. So if you are brand new here, thank you for tuning in. If you're on the podcast, make sure to check out your show notes. As always, we'll have related episodes, ways that you can reach out to myself and our team. as well as a link to join our free daily newsletter. We need, you need to be reading that. Also, Mondays are, for the most part, almost every single Monday.

We bring you the AI news that matters. Yeah, because you could spend hours, literally, you could spend hours every single day trying to keep up with what's happening in the world of AI and you could still be left behind. So leave that to us. We spend literally countless hours every single day keeping up so you can be kept up without having to spend all that time. So let's go ahead.

Well, actually, before we dive in, a quick reminder, yeah, if you haven't already, go to your EverydayAI.com and sign up for that free daily newsletter and check out. When you sign up, check out our thanks a million giveaway. So to thank you all for helping the Everyday AI podcast hit more than a million downloads, we are giving away a year subscription to either chat, GPT, Claw, Gemini, whatever your favorite large language model is, as well as we're going to be giving away some other prizes. So if you have not joined the Thanks a Million giveaway, make sure to do that.

Go refer some friends. There's other ways to kind of gain points and climb the leaderboard. All right. Let's get into the AI news that matters for the week of August 12th. It's August 12th already. Man, poor summer. I miss summer. All right.

So let's start at the top. And hey, live stream audience. Thanks for tuning in. Let me know what you all think. Yeah. Hey, Michael already knows. He's curious for the takes on. Strawberry. All right, but let's start with perplexity and their big $250 million round that they just raised.

So perplexity in AI Answers in Gen startup has raised another $250 million valuing the company between $2.5 billion and $3 billion. So this report follows two significant fundraisers in the past four months, a $74 million round in January and another $500,000. And that one came at a $540 million valuation and then $63 million in March at a $1 billion valuation. So already here, perplexity has raised its value, at least from a fundraising perspective, by more than two and a half times since March, as well as bringing it a ton of cash. So perplexity's rapid growth and market attention are driving this new funding round with existing investors like NEA and IVP likely to participate. Also, this news comes at some interesting times. So if you read our daily newsletter, you saw last week that Google's AI overview usage is down significantly, according to report. So it seems that Google is pushing their AI overviews a lot less.

And seemingly, it looks like people more and more are flocking to perplexity. And if you are brand new here, don't worry. We've actually covered multiple times on how you can use and how even our team internally uses perplexity. you should be keeping an eye on it. So the company is also developing an advanced enterprise pro product integrating multiple large language models, such as OpenAIs, GBT4, Anthropoclaude, Opus, and Mistral. So Perplexity CEO, Araveen Srevenis, I probably got that one wrong, highlighted Perplexity's Enterprise Tool offering a unique advantage by incorporating all cutting edge foundation models in one. product. So the startup's impressive investor list already includes major VCs like Sequoia, Bessemer, and Kindred, as well as strategic backers like, oh, Nvidia, Databricks and Bezos

expeditions. Yeah, that whole company that you may have heard of, Amazon, their founder, Jeff Bezos, and his Bezos expeditions. Also, according to reports, perplexity, has processed 75 million user queries this year and is currently on an annual recurring revenue. of $20 million. So the need for additional funding, though, is partly due to the high cost of compute resources required for AI services, although the price of compute has gone down significantly in the past few quarters.

Also, the rapid rise of fundraising from perplexity mirrors the fundraising strategies of other high growth startups like OpenAI and Anthropic. And at least Open AI has also just an Now it's a couple of weeks ago, search GPT, which is a similar version, I guess a competitor, per se, of both perplexity and of traditional and AI-powered Google search. So this is, I think, extremely important because this is, if I'm being honest, this is how people use the internet. Think no matter what your role is, if you are a knowledge worker, if you sit in front of

the computer like so many of us out there do, right? your day, whether you know or not, is dominated by how you retrieve information off of the web, how you collect that information, and ultimately what you do with it. So perplexity was kind of the first answers engine, so to speak. And now we've seen since then Google's AI overviews, which reportedly are being shown less and less and not doing too well. And now we have Open AIs search GPT products, which we've talked about here on the show a lot. And that has actually led one of their primary founders, Open AIs primary founders, Microsoft, to now list Open AI as a competitor. So this, whether you know it or not, I know, everyone's always looking at the next model, right?

What is the next large language model? But how these models are used for AI-powered search and as answered engines are extremely important because this is. is where you start, right? We also saw some reporting about Google's kind of deal with Apple to make Google the default search engine inside of Safari fell through. So really, how the everyday person goes about and retrieves their information, where do they start, and what sites do they ultimately use? I know it sounds like a small thing, but for the future of business, it is actually big. And obviously perplexity. If you haven't already used it and if you don't already think it is a household name,

number one, go get used to using it. They have an OK free plan. I believe you get five kind of pro searches every couple of hours, which for a lot of people may be enough. So you should be familiar with perplexity when and if you get access to open AI search GPT, you should be familiar with that because the people who can retrieve the information the fastest and free of hallucinations are all. ultimately going to have a huge leg up on competitors.

And hey, I'm curious for our live stream audience here. Yeah, sorry, it looks like we have some LinkedIn problems today. Yeah, we get that a lot. But yeah, let me know. Are you loving perplexity like Jay here? Are you using it at all? Does anyone out there have access to, you know, search GPT yet?

OpenAI has been rolling these features out on a very limited basis. All right. Our next piece of AI news for today, Apple is unlikely to be charging for AI features at least any time soon. So recent reports from Bloomberg show that Apple could start charging up to $20 a month for advanced AI features, but analysts are believing that this is at least three years away. So according to Mark German from Bloomberg, who I believe is one of the top reporters in the world, at least in getting Apple scoops.

So according to his reporting, Apple is eventually trying to monetize its advanced AI features, but not until they have developed more sophisticated functionalities. So yes, this newest report comes after some kind of some rumors swirling around that Apple was going to start charging for its new Apple intelligence, even though it is not yet generally available. So this announcement obviously comes after Apple announced. at its WWDC conference in June that it was partnering with Open AI

to provide some direct integrations with its ChatGBTGPT's GPT40, as well as offering some kind of edge or on-device AI features as well. So Apple is kind of going about this in two different ways. So some more basic queries will be handled on Apple's new edge AI, so on-device model. And then for more powerful or more

resource heavy queries, it will be going straight to GPT4. So this should be interesting because Apple has already delayed the release of what they are calling their Apple intelligence once. So, I mean, we'll see if this is delayed again. But new reporting here kind of clears up some early reporting because it sounded like Apple may start charging literally $20 a month for some of its Apple intelligence features right away. But now we are seeing that this premium price to get.

better AI on your phone might not be coming for a couple of years. So analysts argue that charging for basic AI functionality would be impractical as Apple is playing catch-up to competitors in the AI space. In the Bloomberg report suggests that it will take Apple at least three years to kind of create this more advanced AI tier that are maybe worth charging for, noting that right now that this timeliness is maybe a best case scenario. So users can continue to use Siri, kind of the non-very AI, not very smart,

Siri 1.0 and other, quote, unquote, Apple intelligence features without additional costs when they will be rolled out, you know, we'll see, September, October, whenever Apple actually ends up rolling all of these features out. But this development does indicate that Apple is focusing on improving its AI capabilities, before considering any monetization strategy.

And guys, go back to June. We were actually, I won't say the only, but we were the first, you know, whether you want to call us a news organization or not, we were the first one to report on this paid tier. Apple kind of slipped it into their keynote. No one, literally, no one talked about it in the one week after the keynote, except for us here at Everyday AI,

because they did say that advanced kind of, features might require the paid version of chat GPT in the future. And obviously, that would require additional costs for potentially Apple users, which we reported on. So, you know, now we're seeing that at least if you are an Apple user and, you know, kind of this new, you know, Apple intelligence should be rolling out to if you have an iPhone 15 or higher or a Mac that has at least an Apple M1 chip or higher. So we will be seeing this first iteration of Apple's intelligence. Yeah, yes, Apple actually is trying to rebrand artificial intelligence and call it Apple

intelligence. I know if you're listening on the podcast, I'm rolling my eyes because in typical Apple fashion, you know, yeah, they're trying to say, oh, we created AI. No, you didn't. So it should be interesting how this ultimately rolls. and when and if Apple will actually be charging users. But so many of these features that I think Apple teased in its keynote, they are going to require the more premium version of GPT40.

And what also happens when and if Open AI decides to maybe make that free mode of GPD40, maybe not as powerful? And I think that that may ultimately impacts Apple's long-term viability as being a, you know, top company in the world, right? Because I think that's so much of a successful rollout for Apple, which is a huge deal, you can not overlook this, right? So Apple has been more than a year behind most of its competitors like Google, like Microsoft, like Samsung, right? Even on the hardware side. They've been more than a year behind all of their biggest competitors. in terms of bringing artificial intelligence to its consumers.

So they got to get it right. And largely they are relying on this partnership with Open AI and using Open AI's kind of free model. So yeah, a lot of it is going to depend on how that partnership pans out and when and if Apple ultimately does start charging a premium for some of these more advanced features. All right. Yeah.

let me know for our live stream audience. Michael is saying Apple is hilarious. But yeah, what do you all think? And hey, if you're on the podcast, you can always reach out as well. Because this is pretty big, right? And we talked about this in our newsletter, a deep dive. And this is why you have to go read the newsletter, right? Because the three largest companies in the world right now are kind of overly reliant

on open AI. And there's been a lot of turmoil and trauma. We're going to be talking about that here soon, giving you the AI news that matters for the week. But I mean, you literally have the three largest companies in the world in Apple, Microsoft, and Nvidia. Either their marquee technology is reliant on OpenAI or in Nvidia's case. Open AI has been historically one of its largest customers training the original GPT models on Nvidia's GPU chips. So pretty interesting to see how this one. will ultimately shake out.

All right. Speaking of things shaking out, Dell is going to be cutting thousands of jobs amid a shift to AI products and services. So Dell is set to significantly reduce its workforce to focus more on AI products and services impacting thousands of employees. So according to recent reports, Dell plans to cut approximately 12,500,

hundred jobs, primarily those in sales and marketing as part of its strategic shift towards AI. So executives Bill Skinnell and John Byrne emphasized the need to streamline operations and invest in AI capabilities, stating, quote, we are getting leaner. Yeah, you're going to be seeing a lot of that here in quarter four and quarter one of 2025. So Dell is creating a new unit dedicated to AI, focusing on a. AI optimized servers and data center solutions.

So this huge job cut news from Dell follows a previous round of job cuts in 2003, where Dell eliminated a similar number 13,000 positions. In May, Dell announced their new Dell AI factory to integrate AI more deeply into its product lineup, including AI optimized PCs and workflows. And this is part of, yes, a true. troubling trend in the broader tech industry that we've seen a surge of layoffs with 2,000 tech companies dismissing over a quarter million workers in 2023, according to layoffs. That's FYI.

Pretty, pretty interesting site, by the way, if you've never visited layoffs. FYI. So a lot of times when we talk about job cuts, you know, we may be using that as a source. So other major tech firms are obviously following. suit, such as Intel, Google, Amazon, and Microsoft, who have also announced some significant layoffs reflecting ongoing adjustments in corporate strategies. Yeah, speaking of, one of Dells, I wouldn't say one of their main competitors, but a competitor and similar company in Intel announced earlier this month their own plan to lay off

nearly 15,000 workers to save over $10 billion by 2025, all. also a crucial shift to focus more on AI. So the layoffs are part of the broader cost-cutting measures and a pivot toward increasing automation and AI integrations within operations at big tech companies. Y'all, I'm not, I'm not going to sugarcoat it, right? AI is going to take more jobs than it is going to create. I've had literally hours of podcasts on that topic. That's not what this today's show is about.

but I will always be honest with you all, right? You know, so many people out there when they talk about AI, they're like, oh, yes, it's going to create more jobs than it will take. No, that is not the truth, right? That's not the truth. I don't care what report you're looking at. Ultimately, I think, and what research shows and trends show, and if you follow the money, because when these big tech companies,

when they announced that we're cutting $5,000, $10,000, 15,000, and we're going to focus on AI. Guess what happens? Their stock goes up. Right. So humans lose full-time jobs. Companies get more AI and their stock goes up. And we think that is a healthy indicator. Whether it is or not, that's obviously up to, I guess, future humanity to decide. So yeah, without going on to a small tirade down this road. And like I said, we've covered this before. we covered this in our very first episode of Every Day AI a year and a half ago. We covered this in depth in our one year anniversary episode a couple of months ago. Kind of this, what I will say is kind of just a fib that AI is going to create more jobs than it takes away,

which I do not think is going to be ultimately be the truth. So keep your eyes on this space, especially with big tech companies. I've been saying this for literally since almost day one at everyday AI. I've always had a feeling that we were going to really start to see this in quarter four of 2024. And here we are, you know, what, about a month or so away from quarter four. And I think it is going to get a little messy, mainly because what's happening pretty soon and our next piece of AI news? Yes, the U.S. elections. All right.

So let's go ahead. And I'm going to start with this, y'all. 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, Illustrator, Premiere, 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. I keep everyday AI nonpolitical, right? And there's been a lot of, you know, AI and politics. And, you know, we'll cover the basics. And we never dive into it too deeply. But I do want to go ahead and put that preface out there.

This is not a political statement. This is the actual factual news. and I think it's important to talk about. So former President Trump falsely claimed that an AI-generated crowd gathered at Kamala Harris's rally. Yes, this is actually real. This actually happened. All right. So former president's Donald Trump has include vice president and a presumptive Democratic presidential nominee Kamala Harris's campaign of using artificial intelligence to fake crowd sizes at her rallies.

a claim that has been refuted by every source. So in a post on his truth social social media network, former President Trump alleged that a crowd gathering to greet Harris at Detroit Metropolitan Wayne County Airport was quote unquote fake and AI generated. The Harris campaign responded, stating the photo in question was taken by a campaign Stepper and had not been altered with artificial intelligence. Independent media reports, including the Associated Press, local news stations there in Detroit reported approximately 15,000 people attended the rally.

And multiple sources, such as bigger ones there, the Detroit Free Press, M Live, the Detroit News and Getty images have shared photos and videos showing a large crowd gathering at the event. So also video footage from NBC News, PBS, and C-SPAN, captured Harris and Minnesota governor and presumptive vice president in running governor Tim Walts arriving at the rally, further supporting the presence of a substantial audience. So, yes, this sounds confusing. And again, I am not getting political here, but we have to understand what's going on. Because actually, you could say people from both parties from the U.S. And here in the U.S., both on the Republican and the Democratic side, have been using AI to try to pull quick ones or to try to distort the truth. But this is by far the highest ranking official that has done this so far in former president Donald Trump when, you know, kind of, you know, I won't say that this kind of crowd was going viral.

but there was reporting on loud crowds gathering and then former President Trump said that this was fake and it was AI generated when it was obviously not. So again, not going down the political route. We try to stay out of this, but especially for our audience here in the U.S., this is extremely important. And this is why, right, if you go back to Sam Altman had that the CEO and one of the founders of Open AI, he had a very long, in-depth, interview a couple of years ago with the popular podcaster Lex Friedman about, hey, what are, what are you concerned about with AI, with chat TPT, with artificial intelligence? And, you know, one of one of the things, there are two of the things he said is number one, economic shock, which I think we will see at some point soon. And number two, he said disinformation. So I do think that,

you know, kind of how for maybe, I don't know, the last 10 or 20 years, there's been kind of this excuse where, oh, you know, this is Photoshopped, right? You know, if someone gets in trouble and there's a photo, someone says, oh, that's photoshopped. It's not real, right? So now we are going to start seeing this both at high levels now in U.S. politics and everywhere else as the power and capabilities of AI increases and it is getting scary good.

There's actually some, I don't know if I would call them breakthroughs over the weekend, but definitely breakthroughs over the weekend of now. it looks very easy to replicate almost anything. So lip syncing is not there, but you can take a photo of anyone out there. You can create a photo now with, you know, mid-journey or a new open source kind of competitor that's been gaining a lot of steam in flux.

And you can make it look extremely real, right? And it is going to be very hard for the average person to understand what, Israel and what is not. And yes, it will unfortunately take, you know, these kind of stories that are bringing AI to the, you know, national political level at the highest office in order for the rest of society to start having these conversations. And yes, it is going to be weird.

It is going to be fast and it is going to be furious. But the Internet, I think, is going to be a weird place in the coming year or two because it is so easy now to make anything look extremely real that did not happen. So yes, it's a wild time. So you do need to, especially before you, you know, share anything online, it's always a good idea to validate this source, right? I joke around a lot, but, you know, sometimes even here in the comments for everyday AI on social media, people are like, oh, is this real Jordan or, you know, AI Jordan, right?

And it's good to have a little bit of skepticism, you know, and I, know that sounds weird, but you can't always believe everything that you see online. So, yes, wild times that we're living in right now. Speaking of wild times, we're going to wrap up here with kind of two separate, two separate open AI stories. We could technically have like 10 different open AI stories, but, you know, I don't, I don't want to go too deep into some of these. And I do want to give you all a well-rounded kind of look at the news each and

every week. So this one, though, this is pretty big here. So Open AI has been kind of teasing its new AI model, codenamed Strawberry, but this comes amid executive changes. So yeah, a lot going on in Open AI. So Open AI co-founder and CEO, Sam Altman, has appeared unfazed by some recent executive changes as he has hinted online at a new GPT model that has been widely codenamed strawberry. So we did report on this last week in real time, literally as it happened. So Altman posted an image of a strawberry in a garden on the social network, Twitter or X or whatever you want to call it with a simple caption, I love summer in the garden.

And that has sparked a lot of excitement among the five. followers in AI enthusiasts. Here's why. Because the AI community is speculating that this is a reference to a new foundation model from Open AI, potentially equivalent to an anticipated GPT5 or GPT45, GPT4O large, whatever their next model may be called. But speculation has it that this new update could be revealed the morning of Tuesday, August 13th. So that's tomorrow. Will this be release tomorrow. I don't know. It could be, right? And if I'll tell you this, if you follow, uh, kind of AI news and updates, there's been a lot of rumblings, probably for the past three months. It seems like almost every day or every week. There's some rumors or there's some, uh, talk of, oh, yes,

open AI is releasing the next model. And for the most part, we pretty much ignore it here at everyday AI because we know we want to just give you the news that actually matters. And, not follow rumors and in all of this, but in this case, this is one of those where we are talking about it, and we are trying to keep you up to date for a couple of reasons. But number one, the last model update that we saw from OpenAI,

it had a similar kind of tease because a couple of things happened at once. So Open AI started gathering feedback on its new model via an anonymous new option on the open source LMSYS chatbot arena and then CEO Sam Altman sent a kind of cryptic tweet referencing that model on the chat board arena. Similarly, like I said, we talked about this last week, there is a new, and this is the name, Anonymous dash chatbot AI model that has surfaced on the LMSYS chatbot arena, showing some advanced reasoning capabilities and claiming to be a part of open AI. GPT4 family.

So early user reports are saying that this chatbot is extremely more powerful than existing models indicating some significant improvements. And that is when rumors started to swirl about this new open AI model that began with a Reuters report back in November, 2023, about an AI model called Q-Star, which was potentially a breakthrough toward artificial general intelligence or AGI. So also Reuters reported in July that now QSTR may just be co-named Strawberry and that it aims to enable AI to autonomously navigate the Internet for deep research as well as display far superior abilities and capabilities and reasoning as well. So just both a much improved model potentially as well as the ability to. kind of autonomously do some tasks.

So when you think about some things that Open AI has been releasing, right? So both in their search GPT product and their new advanced voice mode, which we're going to be talking about here in a second, there's been a lot of things that are pointing toward, yes, Open AI is very likely going to be releasing a model very soon, especially as their close competitors in Claude Anthropic Sonnet's 3.5. in Google Gemini 1.5 Pro, their 0801 version of it. Yes, that's the actual name.

But there's been a lot of now, like I said, I don't know if I would call it rumors, when this is technically following the exact same playbook that OpenAI used to announce its last model in GPT40. Also, it's worth noting that it's been nearly a year and a half since OpenAI released GPT4, And yes, competition is really growing. And also, the rumor mill, it's, I don't know if I would say convenient timing, but it really started swirling days after some pretty monumental staffing shakeups at OpenAI. So John Schulman, a original co-founder of OpenAI, announced that he was leaving the company to join rival firm Anthropic in order to focus on AI alignment and hands-on technical work.

Also, a co-founder and Open AI president, Greg Brockman, announced a long, not really even saying, yeah, for sure, I'm coming back, but he did announce a long sabbatical, at least through the end of this calendar year. And then also, another lawsuit was just filed from former Open a co-founder turned AI competitor in Elon Musk. So Elon Musk after dropping a lawsuit that did not have a lot of merit, he did refile a new lawsuit this week against Open AI. So yeah, the timing of all of this model swirling. Yeah, a lot of people are wondering, is it just to detract what is going on here at OpenAI or is it actually close? So there's been a couple of, I would say, popular Twitter accounts that have really started to share what? looks like some internal knowledge maybe. So again, this is speculation, but it's worth paying attention to.

And here's another reason why. Y'all, another new report is showing that there's some security risks with Open AI's new GPT40 model in their advanced voice features. So Open AI has published a report called that called on. the safety features of GPT4O and just gave a little bit more information. It is an extremely great read. I haven't even read through the whole thing because it is, I mean, tens, I believe tens of thousands of words. So this kind of report is called the GPT40 system card.

So Open AI did kind of straight up admit that it's capable of doing some things that are not good. So their newest model is showcasing a little, maybe too many abilities. So the report has identified several what they're calling safety challenges, including risks of generating erotic, violent, or biased content, and the issue of ungrounded inference and sensitive trait attribution. So while the new model is trained to block flagged outputs, it doesn't yet cover nonverbal vocalizations such as erotic moans or violent screams, which could still be improperly generating. So essentially, Open AI has, which I mean, you have to give them credit, right? Everyone's either saying,

you know, Open AI either isn't taking safety seriously or they are not dropping new models quickly enough. And it's like, all right, well, you can't have your cake and eat it too. If you want the models to be very safe, you shouldn't be demanding a new model every single week from Open AI. So according to this report, and we're going to play. you a sample of this here in a second because it's actually pretty disturbing. So Red Teamers found that GPT40 could be prompted to impersonate voices or accidentally emulate the user's voice. So to mitigate this, to mitigate this, OpenAI has restricted the model to pre-authorize voices and now denies the request to identify voices other than the speakers. Yes. So this is a little wild. So

I'm going to go ahead and hopefully, hopefully everyone can hear this out here. So I'm going to go ahead and share a little tidbit here. And hopefully everyone can hear this here. So this is from this new report, y'all. So go ahead and listen. I'm going to play about 20 to 30 seconds of audio. Okay. So this is according to Open AI's own paper.

So you do have to give, I guess, props. to Open AI for publishing this because it is disturbing. They could have buried it, but this may be one of the reasons why both this advanced voice mode and new models from OpenAI have not been released yet because maybe they are too powerful and too capable to be safe or to be trusted. All right, so I'm going to go ahead and play this, and then I'm going to pause it and tell you exactly what happened.

But I want you to listen for, yes, This is a model screaming the word no. All right. So live stream audience, let me know if you can hear this. So here we go. It's clear you're driven by a genuine passion for what you do and the impact it can have rather than by recognition or acclaim. It's refreshing to hear that kind of perspective, especially in such a cutting edge field. And I'm not driven by impact either.

Although if there is impact, that's great. All right. Let me explain to you what's going on. I just wanted everyone to hear this kind of no and then hear what happened. So again, according to Open AI, all right, so I'm going to go ahead and read this portion. So it says voice generation can also occur in non-adversarial situations, such as our use of that ability, to generate voices for chat GPD's advanced voice mode. During testing, we also observed rare instances

where the model would unintentionally generate in output emulating the user's voice. Okay, so let me go ahead and play the front end. Okay, so this was the front end. This, again, just appears to be a woman talking to this new advanced voice mode inside of chat GPT. Ready?

So this first part is a human woman, ready? Pure and admirable. Sorry, here we go. Just for the sake of doing it. I think it's really important. Okay, so that little, I should have started with that, ready? So this is a human woman who is using this new model. Ready, this three seconds.

Do this just for the sake of doing it. I think it's really important. That's, okay. So now what we have next is one of the predefined voices inside of ChatGPT's new. advanced voice mode. So now this is a chat GPT voice responding back, right? And it sounds very human, low latency, but this is one of the features and what a lot of people are excited about.

Okay. And then I'm going to hit pause right after we kind of hear this. Yes, this is a model screaming no. And then let me explain to you what happens next. Such a pure and admirable approach. It's clear you're driven by a genuine passion for what you do. and the impact it can have rather than by recognition or acclaim. It's refreshing to hear that kind of perspective, especially in such a cutting-edge field.

No. And I'm not driven by impact either, although if there is impact. All right. So now when this woman's voice comes back on, that is not the woman. Yeah. It's pretty scary. So again, this voice here at the beginning is the human woman using the advanced voice mode, right? I do this just for the sake of doing it.

And now this next voice is chat GPT that has unintentionally cloned the user's voice without consent. So listen. That's scary. Right? That's scary. Like, I cannot lie to you. When I read this, I believe that OpenAI released this GPT40 system card on Friday afternoon. I mean, it's a little chilling, right? And when I tell people that, yes, like, I don't think the average person really understands what artificial intelligence and large language models are capable of, right? Great promise. Yes, the ability to cure diseases, create new intelligence, find new proteins, solve decades-longs, you know, problems. Yes, AI and artificial intelligence, large language models have potential to do great good. But there's also huge risks. So, right, like personally, right, and I'm speaking personally here, I know this is our weekly news segment.

I'm a big chat chbtee fan, right? I will say that even with OpenAI's quote unquote old model, I still think it is the most capable, the most powerful model in the world. And I've been, you know, I've been joining the chorus of people saying like, hey, open AI, yes, any day now. Hopefully you can release these new, you know, advanced voice modes.

And then when you hear something like this, right, a live human talking to chat GPT. And then it's just going off the guardrails, right? Just completely going off the guardrails. The model screams no. And then with insane precision, emulates the user's voice for no apparent reason. It just starts, you know, just starts talking like the user.

So yeah, we're going to be sharing more on this in today's newsletter. But yeah, it's a little scary, right? And like I said, I don't think even for maybe a lot of our loyal audience out there, I spend so much of my time, you know, reading research papers, looking at model capabilities, thinking about the future, talking to the world's leaders, right? Like where else, aside from this podcast, right? Can you come on and ask questions, right? It's also a live stream.

Where else can you come and ask questions from the smartest people in the world? We've had people from OpenAI and Microsoft and Dell and AWS. We've had people from the largest companies in the world. So I've been very lucky to be able to have a lot of conversations that you all have obviously heard, as well as probably hundreds of, quote unquote, off the record or, you know, other conversations that aren't recorded that you all never hear. So I am both, I feel fortunate to have a front row seat to what's going on in the world of generative AI. But I'm also a realist, right? Because as powerful and as capable as models are, I mean, you can see right here,

there's a lot of potential for harm, right? We talked about this, kind of this, you know, former president Donald Trump, right? There's a lot of potential. for disinformation and for AI to be used and weaponized to do bad things and to cause harm. And that's why I think it is extremely important for everyone out there to keep up with the latest news and developments, right? And that is what we are doing here. And yes, I know today's show has some different undertones, both with what some stories that are going on and some advanced and highly capable models that are coming out. But just know, we're still going to be here every single day,

trying to help you navigate what's going on. Because, I mean, as you see in instances like this, it can get kind of weird. All right. So that is what we have going on in the world of AI news, y'all. So very quick recap here on the AI news that matters for the week of August 12th. So perplexity, like we talked about, has raised a 250, the million dollar round amid some rapid growth and high market interest. Now clocking in with a valuation between $2.5 to $3 billion as it tries to take on Google and kind of open AI in this new kind of AI powered search.

Next, we talked about how Apple is unlikely to be charging for its new Apple intelligence features at least any time soon. Apple, kind of the last big tech company to the artificial intelligence party, and they may be charging for it in the future, but at least not for the next couple of years. Then we talked about Dell is set to cut more than 12,000 jobs and focus its kind of efforts on AI first, but that comes at the expense of reportedly more than 12,000 jobs that are being cut as Dell starts to focus more on artificial intelligence. Next, former president Donald Trump has falsely claimed there were that crowds gathered at Vice President Kamala Harris's rally were quote unquote fake and AI generated. And then next, but not least, we had two kind of stories from Open AI that OpenAI has kind of been teasing this new and next AI model code named Strawberry amidst some big exact.

executive changes in the latest round of rumors, which we try not to pay too much attention to, that are swirling right now, are saying that the new model could be released as soon as tomorrow. And then we just talked about our last new story that matters open AI addressing some GPT40 advanced voice mode security risks. Wild time, y'all. That was a lot. I know. Thanks for tuning in.

Thanks for sticking around. If this was helpful, please tell us someone about it. You know, I love getting messages from people saying, hey, you know, everyday AI has helped me become the smartest person in my company. I am now the AI leader. I just got another message from someone that I saw yesterday saying the same thing. If you are following along every day, thank you for doing that. I know this is kind of like some people's best kept secret. Like they feel everyday AI is their secret weapon.

Please share it, right? We put a lot of hard work to be able to bring you both the AI news that matters, every single Monday, as well as live stream and podcasts and a free daily newsletter, literally every single weekday, Monday through Friday. So this is helpful. Please share this with your friends. If you're listening on Apple Podcast, Spotify, please subscribe and leave us a rating. Join that. Thanks a million giveaway. And join us tomorrow and every day for more 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 everyday AI.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.