Everyday AI Podcast – An AI and ChatGPT Podcast - EP 276: AI News That Matters - May 20th, 2024

Episode Date: May 20, 2024

OpenAI and Reddit’s data partnership, will Google’s AI plays help them catch ChatGPT, and what’s next for Microsoft?  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 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. Key Partnerships and Deals in AI2. Google's New AI Developments3. Microsoft's Upcoming Developer Conference4. Apple's Future AI ImplementationTimestamps:02:00 Reddit partners with OpenAI for AI training, content.04:28 Large companies lack transparency in model training.06:58 Reddit becoming preferred search over Google, value in partnerships.12:08 OpenAI announced GPT 4 o and new feature.14:48 Google announced live smart assistance, leveraging AI.18:19 Customize data/files, tap into APIs, virtual teammate.21:06 Impressed by Google's new products and features.26:33 Apple to use OpenAI for generative AI.29:08 Speculation around AI safety, resignation raises questions.32:22 Concerns about OpenAI employees leaving is significant.34:20 Google and Microsoft announce AI developments, drama at OpenAI.Keywords:Jan Leakey, smarter than human machines, Reddit, OpenAI, data deal, model training, Google, AI project Astra, Microsoft's Build developer conference, AI developments, Apple partnership, safety concerns, everydayai.com, Ask Photos, Gemini Nano, Android 15, AI powered search, Gemini AI assistant, Google AI teammate, Microsoft developer conference, Copilot AI, AI PCs, Intel, Qualcomm, AMD, Seattle, Jordan Wilson, personal data, Reddit partnership, Google IO conference.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. What does OpenAI's new partnership with Reddit mean?

Will Google's new AI announcements place them ahead of everyone else? And how will Microsoft respond this week? We're going to be going over those things today and more on Everyday AI. What's going on, y'all? My name is Jordan Wilson. And I'm the host of Everyday AI. And this thing is for you. It's for me.

It's for all of us. but it's to help everyday people learn and leverage generative AI to grow your career, grow your company as well. So if that's you, welcome. If it's your first time, maybe listening on the podcast. Thanks for tuning in. Make sure to check out your show notes, as always for more information. And if you're joining us live, appreciate you doing this as well.

So every Monday, even when I'm on the road, like today, we try to bring you the AI news that matters because there's a lot, right? Like you would literally spend hours every single day, try to be. to understand what's going on in the world of AI and how it impacts your career or every Monday you can just join us for our segment of AI news that matters so if you're tuning in live like Dutch here or Brian or Victor or Tara joining us from Nashville thank you would love to hear your thoughts sometimes we you know feature our favorite comments here from the live stream in our newsletter so we'd love to hear from you as well so as a reminder if you haven't already

Go to Your EverydayAI.com and sign up for that free daily newsletter. All right, because, yeah, we're going to be recapping the AI News That Matters and a lot more. So let's jump right into it and go over the AI News that Matters for the week of May 20th. All right, let's start at the top. Pretty big deal here, y'all. So Reddit and OpenAI have struck a deal on data for model training. So Reddit has struck a partnership with OpenAI to include. corporate Reddit posts into chat GPT and other AI tools, granting OpenAI real-time access to

Reddit content. So this follows Reddit's decision to charge for access to its data API, sparking some controversy and pushback from users and third-party apps. The partnership raises concerns about the use of user-generated content for AI training and a potential impact on content quality. So, you know, this, you know, there's obviously a lot of concerns that have been raised about the use of user generated content for AI training and the potential impact on content quality on Reddit. And a lot of publishers, you know, such as the Associated Press and Financial Times, have also struck some pretty lucrative deals with Open AI to allow Open AI to use its content for training and to allow its content to be used in chat. Sh pt. So Reddit also reached a similar

partnership with Google in February, which allowed Google to train its AI models on Reddit to data. And that deal between Google and Reddit was reportedly worth $60 million per year. There's no report right now how much this new Open AI and Reddit partnership is worth yet, but it is important to note the timing of all of this. So Reddit just went public about two months ago, right around the same time as that Google partnership was announced. And after this new partnership with Open AI was revealed, their stock shot up 10% instantly on this Open AI news. So it's, this is actually a very deep topic, one that I could probably talk about for hours,

right? So I'm a former journalist, and, you know, there's been a lot of high profile journalism outlets It's in news corporations suing open AI and other large language model makers for using its data to train its models, right? So without going on to a short tangent, you know, here's essentially how large language models work, right? And I'm oversimplifying, but they are essentially built on the open Internet and the closed Internet, right? So, yeah, large language models have a lot of copyrighted information that ends up in its models, right? And if there's one thing that, you know, for all the great things and, you know, the bright futures that these, you know, large companies are helping us paint, you know, they don't really share a lot of information on how their models are trained, right? And again, I'm oversimplifying here, but essentially they have web scrapers and they ingest the entire open internet that they can get to, as well as there's these very large data sets, such as Google's C4 data set, right? And that essentially scrapes the 15 million most popular websites on the internet.

But a lot of those websites are websites that probably aren't the best, right? So such as there's websites out there that, you know, kind of collect all of this copyrighted material, you know, that are essentially kind of like piracy websites, right? So you have to think at what is at the base of these models, right? And I've been saying this now for a year and I've really leaned in to this heavily over the last six months here. on the everyday AI show that, you know, 2024 is going to be the year of a lot of lawsuits, right? So, you know, a lot of people say that these companies, you know,

such as Open AI or Anthropic, and even to a certain extent, Google and Microsoft are, you know, kind of really, you know, just exploding in terms of, you know, open AI and anthropic, you know, going from, you know, small quote unquote startups to now companies that are bringing in, you know, reportedly hundreds of millions of dollars in revenue. And then you have companies like Microsoft and Google, right, that are seeing all-time high valuations meta as well. And a lot of this, a lot of people are saying it's because of these large language models and this generative AI movement. But it goes back to the data, right? Where's the data coming from? Who owns the data? Right. And we're going to be seeing a lot of lawsuits. And I've said for a long time, the New York Times versus Open AI and Microsoft lawsuit is going to be one of the first big dominoes to fall. However, in the interim, what we've seen is partnerships like this, like with Reddit in the Associated Press and the Financial Times.

You know, a lot of times we're seeing partnerships that are worth tens of millions of dollars each and every year or more, right? When we saw that $60 million kind of Reddit Google partnership, I said Google just got the steal of the year in terms of data, right? And if you aren't super familiar with Reddit, I mean, it is a gold mine in terms of data, right? It is essentially the public internet, you know, playing out in real time. But, you know, a lot of people, especially over the years, have started to just search on Reddit versus searching on Google because it's real humans helping real humans and answering, you know, it's real smart humans answering other smart humans questions, right? And in providing context and relevant data. So, you know, that data right there, you know, even just within Reddit is just, just, largely, largely valuable, right? So this is going to be interesting as we see more and more of

these partnerships announced, because there's really, if I'm being honest, there's really only two ways you can go. Either you're going to be, you know, partnering with an open AI or Microsoft or a Google on a, you know, probably 10, 50, 60, 100 million dollar a year partnership. Or, you know, maybe if you're a smaller outfit, you're going to be suing. Or if you're the New York Times, you're going to be suing. So there's really only two ways that this is going to go. And I think people don't really talk about this a lot, right? We all just kind of blindly benefit from these large language models, myself included, right? And we love them and all of the capabilities that they can give to ourselves and to our business, but we don't get back to, okay, we're they

legally trained, where they ethically trained, right? And a lot of companies don't share. You know, they just say, hey, we, you know, crawl the open internet and we have humans, you know, who, you know, kind of comb through all this data and they trained models. But, you know, ultimately there is a lot of copyrighted material in all of these large language models. So pretty, pretty big news here from Open AI, presumably the most important player in the generative AI or large language model space in Reddit, which is probably the holder of the most unique data on the internet, right? But this also goes to beg a bigger and different question, right? Let's say you are an avid, you know, Reddit user and you know, maybe you've put a lot of personal data out there. You know, you've done so

anonymously, right? You're not putting your, your full name on your Reddit profile, but you're probably maybe sharing a lot of either your own personal data or maybe your own personal knowledge and expertise. And, you know, now it's like, okay, that's kind of for sale, quote unquote, or it's already been sold, right? So this does beg a bigger and bolder question that, you know, will definitely be tackling in the coming weeks and months here on everyday AI. Yeah, and I'd love to hear from our audience. What do you think of this, this Open AI and Reddit partnership and just in general, right? I'm curious, do you feel like I do?

And maybe, you know, I'm a little bit on the extreme side, but I don't see any other way, right? I either see large lawsuits, right? Because let's say you're a medium-sized publisher, right? You might not be one of these first, you know, outfits to get a, 10 or a $60 million partnership. So, I mean, what's your other alternative? If you feel pretty confident that these large language model makers have used your copyrighted content, right?

Now, yeah, you can put kind of blockers up there that block your website in the ongoing content that you're creating. You can block it from these different large language models. And you can block kind of those agents that go on your website and scrape. But then at the same time, even in the case of Google, right? You want your content to show up on Google. that's presumably how you make money, right? But then is it going to end up in Google Gemini and all these other, you know, you could say

the same thing with, you know, you want to show up on Bing, you know, Microsoft search, but then, okay, is it going to show up in co-pilot? So, you know, we are in this tricky, right? This tricky day and age of like, okay, putting out good content that helps people, that's valuable, that serves your customers and your readers. But, you know, at what point is it your data in your data alone or your content and your content alone versus, you know, now it's going out to all these large language models. And especially, as we talk about a lot here on the show, as the lines between, you know,

traditional internet search and content consumption and AI, as those lines blur, you know, there's, I think there's a lot of big issues to tackle there. Yeah, love what Woozy said here. He said, helping, you know, in quotes, and smart humans might be a stretch on some of those Reddit threads. Yeah, I agree with that. Yeah, I, I mean, obviously there's a lot of, you know, nonsense there on the Internet and Reddit included. But yeah, there's obviously a lot of very great helpful information out there on these publications as well. All right. Our next piece of AI news.

Yeah, Google's I.O. conference. Kind of a big deal. So let's talk about what they announced on Tuesday and because it was literally like AI everywhere. But first we, or sorry, before we even go through kind of the bullet points of what, was announced and why it matters. I mean, we have to talk about the timing. 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.adobie.com. So Google's I.O. Developer conference was May 13th, right? So six days ago. But Google announced that date two months prior, right?

It wasn't the last second thing. They'd been planning this event for two months. They had that date out there for two months. And then about two days prior to that announcement, OpenAI announced a surprise event, right? Which happened the day before. It happened last Monday, you know, about an hour after we wrapped up our live stream. And they called their, Open AI called their event, just spring event. Okay.

So in that event, OpenAI announced not only its new GPT40 model, which was available immediately to paid users, but also a new feature that kind of competed directly with Google's big feature that they were coming out with. So Open AI didn't really name their feature, but essentially it's this ability or capability to kind of see for the AI, the model, so GBT40, in the iOS app, to kind of see things live, right? And we've talked about those demos and we showed those demos here on the show before. But, you know, essentially you could hold your phone up and work on a math problem live and ask chat gvt like hey what do you see what's going on here help me with this right so it can see it can interact it can talk to you with very low latency right um human

times so uh it's interesting because that was kind of google's big uh big piece here so we're showing here on the screen about project astra so about how a lot of people are saying that this is the future of AI um so it's kind of uh you know the timing when we talk about The news that matters, I mean, Open AI just swooped in, you know, kind of had this surprise announcement, literally 24 hours on the clock before Google announced theirs. So Project Astra was this from Google. It's their kind of new multimodal AI assistant. That's similar to the one that Open AI demoed 24 hours before that you can kind of interact with in real time. It can see the world, multimodality, answer questions, help with tasks, etc.

So kind of in Google's demo, which we hope was live, right, unlike their Gemini snafu that they, in December, they infamously shared a heavily edited video showcasing Gemini's capabilities, which it did not have those capabilities. So unlike this one, this demo for Project Astra was presumably live, right? And they said it was, you know, an unedited one-x speed demo. but it kind of showed someone walking around an office with kind of Project Astra on a Google smartphone and, you know, asking questions about things in real time that were coming up on the camera. So you can think of it like this. Think of if you're, you know, FaceTiming, right? If you ever have to face someone, FaceTime someone who's maybe, you know, you need help with your computer or, you know, you're showing someone on your house and something's broken and you're saying,

how do I fix this, right? Maybe you have an uncle who's a handyman and you're facetiming them and there's a leaky pipe and you're saying, how do I fix this? Right. So that's kind of what these new kind of smart assistants are, right? It's in live. You can show them something and talk to them and converse, right, in real time. So that's kind of the big piece of, you know, Google's announcement at I.O. last week.

But before we get even too much more into that, I do want to kind of detail kind of some of the big pieces here because there was a lot. Google, right, announced a slew. like they're literally sprinking, it's like sprinkling AI everywhere, right? Like the salt bay, just sprinkling salt on a stake. I mean, they're AI this, AI that. So let's talk about a little bit about what I think are some of the bigger and will ultimately be the more impactful AI announcements

that came out of Google's IO conference on Tuesday. So they did announce some enhancements to their Gemini models, including Gemini 1.5 Pro with a longer, context window between one million tokens and two million tokens. They now also have an ask photos powered by Gemini, which was revealed for Google Photos, providing enhanced photo memory summaries. That one looked really cool, right? So essentially, if you use Google Photos, I do, you can essentially just ask the same thing, right? So they gave the example. I think it was someone's daughter and her progress of swimming. And you can say, hey, how, you know, show me everything, you

my daughter's swimming progress, right? And then this kind of ask photos kind of goes throughout the years of all the photos. It uses kind of AI and computer vision to say, okay, here's, you know, this is your daughter. You know, maybe even if you have multiple daughters, you can ask about one. It uses kind of this facial recognition technology, kind of creepy, kind of cool. And it says, all right, here's here's everything that your daughter. Here's kind of a photo montage and kind of progress on how she's been swimming, right? So a lot of really cool use cases, if you're okay, handing off all of your data and all of your photos, you know, to Google.

I don't personally care. I do this. But Ask Photos does look like a pretty unique feature. And I'm wondering if, you know, as we gear up for Apple's announcements at WWDC, I'm wondering if they'll have something like that as well. I don't think they will. So I think that's a pretty unique in a really strong feature that Google announced. Also, Google announced the rollout of their new Gemini Nano model with multimodality on their pixel phones. Android 15, their new operating system was presented with AI-powered search.

Gemini as the new AI assistant, obviously, and on-device edge AI for new experiences. Pretty huge piece there, right, bringing kind of the power of large language models locally on devices, pretty big. Also, another, you know, maybe this is a straight competition piece with OpenAI, but Google showcased gems, which are essentially customized versions of Gemini AI for specific tasks. So just how, you know, you have your GPTs with OpenAI, you can create a customized version of a model. You know, you can put in your own data, your own files, your own, you know, kind of configuration instructions, you know, tap into AP, third-party APIs and bring in your own kind of specific capabilities. It's kind of what Google is going for here with gems, which are just literally smaller customized versions of its Gemini model.

So pretty much a one-to-one, I don't know if we would say clone, but more or less cloning the capabilities that you have with GPTs in Open AI. And then the fact that this is like seventh on my list here is pretty remarkable, but Google also demonstrated their new Google AI teammate, which is a virtual teammate for work task. So you literally can give a virtual teammate, kind of a space inside of your Google workspace, you know, and it just performs actions, you know, on your behalf. You give it access to the data that needs to, but literally thinking it is like, okay, this person has a seat on my team. So kind of thinking with Google's new AI

teammate as a literal virtual teammate, not just this big, you know, idea of, oh, you can do, you know, you can do all these things with AI, but no, like my AI has a seat on my team. It has access to all of these files. So pretty cool announcements there from Google. And I'm curious, you know, for our live stream audience and, hey, if you're listening on the podcast as well, what's your reaction between, you know, what Google announced on Tuesday versus what Open AI announced the day prior, right? clearly, I mean, we're kidding now, actually back to back to back announcements from three of the biggest players in AI.

But I'm curious what you thought. You know, my thought is the Google announcement. They obviously, I mean, they're a much larger company, right? Open AI technically startup, you know, one of the most powerful startups ever, but still technically a startup. And Google is, you know, legacy tech company, one of the largest companies ever, right? So, you know, we're kind of comparing apples and, you know, banana pudding here. But still, the Google announcement, and I watch them both, the Google announcement seemed just too, too scripted for me.

It seemed too polished, right? And I think especially as we talk about generative AI, I think people like authenticity. I think they like, you know, that's kind of why I go live here every day. I don't edit things. You know, I sometimes, you know, mess up or stutter or say something. I didn't mean to say, it's like, I keep it in there, right? Because I think if we're talking about artificial intelligence, you know, it should be real. It should be live, right?

It should be, you know, raw. And I think that's what we saw from Open AI, whereas, you know, Google's announcement seemed robotic, right? It seemed very robotic. It seemed overly scripted. That's just me, obviously, from a product standpoint, super impressed with what Google rolled out. I haven't personally been impressed with Google Gemini as a model, right? But as you start to see all of these other features and capabilities that Google rolls out tied to its model, right?

I mean, you have to start paying attention. So, you know, I'm curious if you're paying attention to Google maybe a little bit more now than you were last week before this big announcement. Yeah, hey, Brian, Brian here's, you know, kind of on top of the breaking news. This didn't make it into our AI news that matters, but he's asking, you know, about if anyone saw that Open AI is pausing the, you know, of Sky Voice. So yeah, we'll be talking about that in our newsletter, but they did say, yeah, that was a little too close. They said to Scarlett Johansson. So they paused the use of that one. All right. Speaking of these big companies, right? So we talk now about, you know, kind of OpenAI's announcements with their GPT40 that came out right before Google's. And then Google, all of Google's

IO announcements and some of them competing directly with Open AI. And then here we go, you know, Microsoft sitting over here in the corner being like, hey guys, you can't. can't forget about us. When it comes to legacy technology, like Microsoft is there. All right. So Microsoft is, it's Microsoft Build Developer Conference, is set to kick off now kind of unofficially here in a couple of hours. So the official build conference, though, from Microsoft kicks off tomorrow.

So that's Tuesday, May 21st. However, in a couple of hours here, Microsoft is slated to announce a few new products today ahead of its keynote tomorrow with C.E. CEO, Sadia Nadella. So Microsoft is holding an AI event today to showcase their AI vision and upcoming surface hardware in Windows software changes, all based around AI. So the event today will kind of set the stage for what Microsoft is calling the year of the AI PC. And will also include updates on Microsoft AI initiatives, such as co-pilot, right? Powered by GPT and Microsoft's reported 10 to $13 billion investment in OpenAI for.

for reportedly 49% equity. So let's talk a little bit more about Microsoft Build in what we might be expecting this week, right? Huge week in tech, my gosh, y'all, like having from the three biggest companies, having their yearly announcements, more or less, pretty big news here. So Microsoft, obviously, will be discussing

their AI vision and upcoming changes. So starting today, the event, so Microsoft, the Build Conference, will also be taking place right after the event. Like I said, where further updates on AI and co-pilot will be shared. The conference is also obviously going to be heavily discussing AI PCs, right? Kind of this new term, but, you know, Microsoft is saying, hey, we are AI, we are an AIPC company, right,

baking AI into the hardware, and they'll have a neural processing unit to handle AI tests and are expected, so the new AI PCs are expected to arrive with Intel, Qualcomm, and AMD chips. also so many sessions right so the conference is going to feature you know keynote speeches product announcements and over 480 sessions with 251 offered online in 175 recorded for later viewing so yeah if you want to check out that free conference yeah you can view it for free you can obviously attend it to the event is in Seattle and the event does run for three days technically starting on Tuesday but if you do want access if you want to check out some of those sessions that are live and free and available to anyone make sure you check out the

newsletter and we'll be dropping a link there to that registration page who a lot a lot going on y'all I got to I got to take take a sip of water so so much big news you know in the past seven days from open AI from Microsoft from Google so let's talk a little bit speaking of big companies there's one that we haven't technically really talked about yet, and that's Apple. So OpenAI and Apple reportedly have a date. So we did talk about this last week on our AI news that matters that, hey, Apple, it looks like, is choosing OpenAI over Google to break generative AI capabilities to its next iPhone and iOS updates. So according to Bloomberg reports Apple and Open AI are preparing for a major partnership announcement at Apple's

WVC. So OpenAI is working to ensure that their platform can handle this influx of new users. Yeah, we'll see about that from iPhone. And iOS 18 is expected to feature AI-driven messages and Safari summarization capabilities. Hey, I can't wait for that. I'm so bad at this, y'all. I don't know if it's just me. At the end of the day, like, I kind of look at my text messages and I have, you know, sometimes 50 unreaded, sometimes 250.

I've met a lot of group chats. But, you know, I actually can't wait for the AI summarization where I can just ask, you know, Siri and hopefully Siri will be a little smarter if it is maybe powered by Open AI to say, hey, catch me up on all of these text messages, right? Or if that comes to email, you know, on the iOS device, but then also for Safari as well, right? That'll be a huge time saver. Obviously, you know, something that we've had capabilities to do on the desktop for, you know, multiple years. And if you have, you know, Google or Android phones, you've had capabilities to that as well. So, you know, personally, you know, I have an iPhone. So I'm looking forward to kind of some of these generative AI features coming reportedly to the iOS 18 and Apple's next iPhone as well.

But we do now know that there is an announcement planned for June 10th to kind of talk more about this OpenAI and Apple kind of partnership. So again, like we talked about this last week, we said, hey, it does look like, at least according to Bloomberg reports, that Apple has decided on Open AI, right? There was literally reports going on back and forth so much flip-flopping over the last three months on, you know, oh, Apple is going to develop their own model. And, you know, we saw them release a couple research papers. And then they said, no, they're going to Open AI. And then there's a new report that said, no, they're definitely going Google. But, you know, we really, at least in the last week, we haven't seen any conflicting or contradictory reports, but it does look like not only is Open AI and Apple going to be partnering here in the future. But we will see an official announcement from the groups on June 10th, you know, kind of right there at the front end of Apple's Worldwide Developer Conference there in June. So pretty big, pretty big news. All right.

Speaking of pretty big news, our last piece of AI news that matters, well, there's some more drama at OpenAI. You know, you can't get through more than a week without some trauma. But this time it comes to their former head of alignment. So we did, yes, talk about this on the show last week as former Open AI head of alignment and super alignments teams Jan Leakey resigned last week from Open AI. So we did talk about that. but he kind of put out some more information, I guess,

or kind of a statement, so to speak, on Twitter. So we're going to be breaking that down here. But, you know, if you don't know what super alignment is, but in short, you know, Jan Leakey led the team that insured future AI and artificial general intelligence. So AGI models were safe for society. So a pretty important role there at OpenAI or any company. So if the head of the kind of department that,

ensures that, you know, these models are safe for the public, kind of resigns suddenly. It's pretty big news, right? And there's obviously a lot of speculation, not really going to talk about that a lot. But you can imagine, right, if the head of, you know, making sure that AI models are safe at one of the biggest players in AI in the world suddenly resigns, yeah, you might have to raise an eyebrow and be like, oh, what's going on there? You know, are there bad things happening? are new future models safe, right?

This did kind of align with GPT40. I don't think that it's related to GPT40, the new model from OpenAI, but you do have to assume that there is going to be another much more capable model coming very shortly from Open AI was kind of Jan's kind of announcement related to that. I'm not sure, but let's just kind of read over some of the things that he said in his announcement. So kind of quoting him here. So he said, over the past few months, my team has been sailing against the wind.

Sometimes we were struggling for compute, and it was getting harder and harder to get this critical research done. Jan also said, building smarter than human machines is an inherently dangerous endeavor. Open AI is shouldering an enormous responsibility on behalf of all humanity. And then he said, but over the past few years, safety. and culture and processes have taken a backseat to shiny products. So just huge, I mean, just some huge one to punch there. So essentially saying, hey, we weren't able to get kind of the compute power that we needed to make sure all that these models were safe.

But then also coming out and just saying plain as day, hey, it was all these, you know, kind of he said that the company is more concerned about all these, shiny features and they're not cared about the model's safety. So pretty eye-opening, you know, whether whether you follow, you know, AI safety and, you know, ethical AI or not. Pretty, pretty big, pretty noteworthy here. And Open AI's CEO, Sam Altman kind of said in response, said, you know, kind of said he's super appreciated of Jan's contributions. But then he said, he's right we have a lot more to do we are committed to doing it i'll have a longer post in the next couple of days so yeah pretty eye-opening right um when the former head of alignment at you know and

you can't just look at you know oh you know whether your company's using chat gpt or open ai or not the reason why this is super important is because there are thousands uh i don't know if open a i gives a number, but I would venture to guess that there's tens of thousands of companies that use open AI's technology in their products, right? So there's a good chance, you know, let's just say you have a handful of, you know, vendors that you work with at your company and, you know, different capabilities. Maybe it's a CRM, maybe it's a ERM. You know, I don't know. But just think, there's probably so many of them that use kind of an API connection to Open AI. So you're probably using a lot more of this GPT technology from Open AI than you may realize. So it's not like,

oh, okay, well, if models aren't safe, then, you know, I just won't log on to chat GPT or, you know, my company will, you know, switch over. You know, it's not like that, right. Presumably, there are tens of thousands of other, you know, enterprise pieces of software, products, services, et cetera, that tap into Open AI. So, you know, pretty, pretty concerning here when you have a pretty sudden resignation, but not only a sudden resignation, but, you know, generally when open AI employees leave, and we mentioned this in the newsletter, last week, there's some pretty strict rules around that. And, you know, there's some reporting, you know, and Sam Altman came out on Twitter as well and said like, oh, you know, this is my mistake.

And it's embarrassing that we had all of these kind of stipulations, you know, that open AI, you know, former Open AI employees couldn't, you know, say certain things after they left. So, you know, generally if a high-ranking person from Open AI leaves, you don't really hear anything. It is radio silence. So it's actually pretty telling here. And I would assume pretty, you know, pretty brave and bold move here as well from Jan Leakey coming out and saying like, hey, you know, it's, this is an inherently dangerous, right? He literally said, building smarter than human machines is an inherently dangerous endeavor. So, you know, essentially saying this is, this is dangerous.

Like, this is dangerous work that we're working on. We don't have the compute power to make this all happen. But then also saying, like, hey, our work and this important work that we're doing was taking a backseat to safety as well. So we're going to have a lot more from that breakdown in the newsletter. Yeah, agree with Michael here joining us from YouTube saying, yeah, so much going on today. So, yeah. Hey, as a quick recap, here is the AI news that matters for this week. So, yeah, we talked about Reddit and Open AI have struck a deal on data for model training.

No specifics yet on the amount. But presumably tens of millions or $100 million. Google's I.O. conference was last week, and they just announced AI literally everywhere. Probably the biggest piece was their project Astra. Then Microsoft kicks off, unofficially kicks off their build, Microsoft Build Developer conference today. but that runs, the official conference runs from starting Tuesday for three days in Seattle, and we're expecting to see kind of what's next with AI at Microsoft. Then Open AI and Apple now reportedly have a date on the calendar to announce officially on June 10th,

their partnership. And then last but not least, some little bit of drama at Open AI as its former head of alignment speaks out about some of his safety concerns. All right. That's not it. That's just the recap. So if you haven't already, make sure to go to your everyday AI.com, sign up for that free daily newsletter.

And hey, if this was helpful, let me know about it. If you could, you know, we spend, you know, hours every single day, you know, going through what matters in the AI news to keep you up to date so you can grow your company, grow your career. So if this was helpful, we'd appreciate if you could, you know, if you're listening on the podcast or, you know, share it with your friends. If you're listening here on the live stream, we appreciate your support. You know, click, click that.

little repost button if you're here on LinkedIn. We appreciate that. And I'd appreciate it if you 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 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.