EP 349: AI News That Matters - September 3rd, 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.

What is Orion from OpenAI? Are Apple and Nvidia teaming up and for what? And will Amazon's Alexa finally be getting smarter? Well, we're going to have those answers for you today and more on Everyday AI. What's going on, y'? All? My name is Jordan Wilson. I'm the host and this is for you. Everyday AI is a daily live stream, podcast and free daily newsletter helping us all learn what's going on in the world of generative AI so we can leverage it to grow our companies and to grow our careers. And usually once a week on Mondays, except when there's a federal holiday here in the US we bring you the AI news that matters. Yes, because here's the reality. Your business, the future of your business, whether you succeed, whether you get caught, whether you catch your next competitor. If I'm being honest, for the most part it depends on how you use generative AI in your day to day. So this is the one time a week with the AI news that matters, that we just recap it all. We say, here's what's going on and here's what you or your business need to do about it. So we do this once a week, but we do everyday AI every weekday. So make sure to join us for that. Make sure to go to your everyday AI dot com. Sign up for the free daily newsletter. All right? And while you're there, we, we, we decided to extend this our, our Thanks a Million giveaway. I was a little sick last week so couldn't wrap this up. So make sure you go check that out. Sign up for our Thanks a Million giveaway. Celebrating a million Downloads at everyday AI takes about 10 seconds. You get a link after you sign up and then you refer your friends with that link. So we're giving away whether a year of your favorite large language model, whatever that is. We're doing a 90 minute consult with our team. We might drop one or two other prizes here in the last week, so make sure you go check that out. All right? With that, here is what you need to know for AI news that matters for the week. And hey, for our live stream audience, Rolando and Brian and Marie and Tara and Jay. Thank you everyone for joining us. I'd love to hear from you all as we go along what you think of this week's news. All right, so let's start it off here Amazon's new Alexa is set to launch with Anthropics clock Claude AI amidst some challenges and competition. So Amazon is gearing up to release an upgraded version of its voice assistant, Remarkable Alexa, which will be powered by Anthropic's Claude AI model. So this development is significant as it, as it marks a shift from Amazon's in house AI efforts which have faced criticism for their performance. I don't know. And for the fact that it seems like you can't even talk to Alexa, right? Like if you've been chatting with a large language model for the past two years, trying to talk to Alexa is pretty much impossible. So initial versions of the new version of Alexa have reportedly struggled with user prompts and clarity, leading Amazon to seek external expertise from Anthropic, in which it has invested, yes, $4 billion. So a lot of competitive regulators, including in the UK are currently investigating Amazon's minority stake in Anthropic, highlighting the scrutin surrounding this partnership. So this name, Remarkable Alexa, reportedly has faced numerous technical and bureaucratic hurdles since its announcement last September, as noted by Formazon former Alexa AI scientists. So the new version of Alexa is expected to outperform its predecessor, which, my gosh, that shouldn't be too hard, with Amazon stating it'll utilize a variety of models to enhance user experience. So anticipated features for Remarkable Alexa include daily AI generated news summaries. So I mean you can do that, but don't replace us with it, right? A child focused chatbot and tools for conversational shopping. As if you weren't spending enough money on Amazon. So here's the downside. There are plans to reportedly introduce a subscription model potentially charging between $5 and $10 per month, while still maintaining the classic version as a free service. A demonstration of the new improved version of Alexa is set to take place at Amazon's annual Devices and Services event, which is typically held this month in September with an expected launch in mid October. So y', all, I mean is it just, is it just me or is this long overdue? You know, I'm curious for our live stream audience and hey, even for our podcast listeners, make sure to check out your show notes. I always leave my, you know, my LinkedIn URL, so feel free to reach out, send me a message though at least I know who you are. But you know, I'm curious, are you even using these, you know, Siri and Alexa anymore? If I'm being honest, I can't. Right? If you've been using a large language model, chat GPT, Claude Gemini, etc. For the last year or two trying to use something like Alexa or Siri is. It's like I literally find myself like, like the face palm emoji. I'm like, this is impossible. I can't get the simplest answer. And then I go back and think why and how did the rest of humanity find this useful the last decade? Because it is infuriating. So I'm personally excited to see that Amazon, I think is doing the right thing here and partnering with Claude. However, five to $10 a month really and also depends on what is the classic version of Alexa. Is the free version going to be what we have today or is it going to be a dummy down version of this new and improved Claude? It seems like it is the former and not the latter. So I don't know. Granted I'm not the head of a trillion dollar company, but if that is the case, if Amazon is saying to get any of this new Alexa you have to pay 5 to 10amonth, I think that is a major fail from Amazon if I'm being completely honest. I know, hey, normally we do the Monday on, you know, the AI news that matters on Monday, it's Tuesday, so I got some hot takes in me. I don't think that would be a good move. I think they should keep it free because Competitors, Google Open, OpenAI is a competitor now. They're going to be keeping, you know, they're going to have a version of this that's yes, it will probably be included with their paid tiers, but I think most people are going to have a paid tool of choice. Right. So whether that's a Google Gemini, whether that is a chatgpt and I don't know if you have that. Are you going to pay an additional five to ten dollars for Alexa? I'd say probably not. Yeah. Jay says another subscription play. I'm out. Tara says, so excited to plug my Alexa's back in. Yeah, I have a couple that are currently unplugged in my house. All right. Marie says I use Google Home to play music, ask what the weather is for today, stuff like that. Not looking to upgrade at any time. Yeah. All right, let's keep it going. So here's a big one. Nvidia and Apple. Yes, Nvidia and Apple are reportedly looking at going in together on a funding round for OpenAI. So OpenAI is reportedly in discussions with Nvidia and Apple for a new fundraising round that could drive the value of OpenAI to a staggering $100 billion. So according to the New York Times Thrive Capital is expected to lead this newest funding round for OpenAI if it materializes, following reports from Bloomberg and the Wall Street Journal recent regarding Nvidia and Apple's involvement. So despite generating a reported $3.4 billion in annual revenue or ARR earlier this year, OpenAI is projected to incur losses of up to $5 billion by the end of the year. Again, that's according to reports. Are those correct or not? I don't know. So OpenAI's reliance on Nvidia's GPUs for training it's GPT models highlights the interconnected nature of the tech industry, particularly in AI development. Apple has shown a commitment to integrating ChatGPT into its iOS platform, so there's obviously some pre existing relationships there, further solidifying the partnership between these tech giants. So this one is pretty interesting here and it kind of actually relates to our previous story because like we talked about, Amazon has reportedly invested up to 4 billion with a B dollars in anthropic and Apple and Nvidia are the two biggest companies that don't have a multi billion dollar investment in a large language model maker right now. Microsoft, you know, has a reported $13 billion investment from a few years ago into OpenAI, which has them a reported 49% ownership stakeholder in OpenAI. So you know, it is actually curious to me that Nvidia and Apple don't already have a large multi billion dollar ownership stake in one of these big large language model makers. So we'll have to see how this shakes out. But I do see, you know, if, if I had to guess, I would see either Apple or Nvidia at least coming with a multiple hundred million dollars or billion dollar investment in this next round. This one was wild y'. All. Did you see this? So Google's gaming gen has marked a breakthrough in AI driven technology. All right, we'll leave, we'll leave a link to a video in today's newsletter, so make sure if you haven't already, go to your everydayai.com Sign up for the free daily newsletter. But Google researchers have achieved a significant milestone by developing Game Engine, a neural network capable of generating real time game gameplay for the classic shooter Doom without a traditional game engine, yes, literally it can generate the game in real time, which is mindboggling. So Game Engine runs Doom, the classic shooter game from, I don't know, was that the 80s or the 90s? So it runs it at 20 frames per second on a single TPU or Tensor Processing units, demonstrating the ability to simulate complex 3D environments efficiently with generative AI. Also, this marks the first instance where an AI has fully simulated a video game with high quality graphics and interactivity. Moving away from traditional game engines that require obviously extensive coding. The implications for the $200 billion gaming industry are obviously profound as AI driven indentures engines could drastically reduce development time and costs, making game creation more accessible to smaller studios and individual creators. Y' all did. Hey, livestream audience. Did anyone else see this? We did drop the news piece to our in in our newsletter last week. So not sure if you caught that or read it, but this is wild. Yes, and thank you, Jay. Doom 1990s, not 1980s. Thank you. But y', all, the fact that this is possible. Let's think about this. It literally these video game studios, I mean, they are huge outfits, right? Billion, multi billion dollar companies to create video games. And yes, all this is doing right now, all this game engine is doing is, is, you know, it's creating a video game in real time. But at the time when this Doom game came out, this was the best leading video game in the world. And you know, there's still a cult following around Doom. So the fact that an AI engine can create a video game in real time, Bonkers, y'. All, I cannot underestimate this, right? And the implications for actually what this means in the long run. But I'll take you, I'll take you here. So we were invited out to Nvidia's GTC conference this year. And this is one of the questions that came up to Nvidia CEO Jensen Huang in a closed door session, essentially saying and asking, hey, what does the future for video gaming look like? Nvidia obviously one of its first big areas where it played and thrived before the generative AI and GPU boom was in the video game sector, right? So Jensen essentially said like he could see this happening in a couple of years, right? Where this real time video game creation. And even me, someone that follows generative, generative AI literally every single day, I scratch my head on that one. I said, wow, okay, that's shocking. But here we are, right? This hasn't even been five months since the Nvidia GTC conference, where I think even when he said that a lot of other people that were following generative AI were kind of taken aback by that, but here we are, right? The fact that an AI system can generate a video game in real time, what does that mean for our future? Right? Well, I mean, obviously if, if you're in or work around Video games or play video games. You can see what that means. But what about for everyone else? Are we going to see personalized video games? Are we going to see personalized short films? Right? I mean, because think if this can happen for a video game that, you know, you obviously interact with in real time, think of what it just has to the doors this opens up for just interactive media in general, right? You know, I think back to when I was a kid. There was all these, you know, kind of choose your own adventure storybooks, right? In print. Think of this in real life, think of this as a movie, right? Depending on, on where you look on the screen or, you know, being able to talk to a movie or something like that. I mean, this is realistic and it's possible. You know, the technology literally exists today. So pretty huge breakthrough there from game in general. All right, let's go to the other side, the ugly side of AI. So Klarna has planned some major job cuts as AI investments reshape their workforce. So Klarna is the, if you haven't heard of it, is the buy now, pay later firm is making headlines in a not so great way with its plan to cut nearly half of its work's workforce in response to advancements in artificial intelligence. So Klarna, according to reports, is aiming to reduce its workforce from 3,800 to 2,000 employees by leveraging AI in marketing and customer service, following a previous reduction from 5,000 to 3,800 over the past year. So, yes, over about a year and change they're looking to go through from 5,000 employees to 2,000. And much of that is because of AI. So Klarna's CEO stated that the job cuts would allow the company to increase salaries for remaining employees, highlighting a shift in workforce dynamics due to technical technological efficiencies. So if you haven't heard already, we talk about this a lot on the everyday AI show. But the very famous International Monetary Fund, the IMF study has projected that AI could affect nearly 40% of all jobs, potentially exasperating existing inequalities in the job market. So Klarna's recent financial results showed a 27% revenue increase reaching $1.3 billion, attributed again to efficiencies gained through AI investments. And also this comes as Klarna is reportedly preparing for a public listing. This one's interesting, y'. All. And I think Klarna could actually be one of the earlier case studies in what happens when you start to, quote, unquote, replace traditional human jobs with AI. So Again, this next kind of round of cuts is going to be slowly rolled out and we'll see how it impacts their revenue. And if they do go public, however, they already had one round of this and it led to a 27% revenue increase. Right. And I do think this approach here is going to be a common one. So when companies cut jobs due to AI, I mean, what usually happens if they are public companies? Their stock goes up. Wall street hates employees. Wall street loves AI. And we're probably going to hear this kind of, I'm not going to say excuse, we'll say this reasoning or rationale, right? Oh, this allows us to pay our employees more. So yeah, get ready for that spin. We're going to be seeing a lot of it. 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 chat, GPT 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 everydayai.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, here's another big one, y'. All. So some AI giants are collaborating with the US government for safer technological development. So according to reports, OpenAI and Anthropic have announced agreements to provide the US government with early access to their upcoming AI models. And this marks a pretty significant step in the evolution of AI technology in terms of its risk and safeties. So this collaboration is particularly newsworthy as governments worldwide are increasingly focusing on regulating these powerful AI algorithms. Because, yeah, what happens when these AI algorithms come out and these, you know, large language models are updated? Companies are implementing them and cutting jobs. So both companies have reportedly signed a Memoranda of understanding, or mou, with the US Artificial Intelligence Safety Institute, part of the National Institute of Standards and Technology, or nist. So the US AI Safety Institute was established following President Biden's executive order on AI, emphasizing the government's commitment to a more responsible AI development. Anthropic has previously shared some of its models with the UK AI Safety Institute, whereas there's been reports, which we're going to talk about here in a minute, of OpenAI doing the same. So, yeah, live stream audience what do you guys think about this? Should these private companies in the US be required to first either share or get their technology approved with the US Government? Yeah, this is one of those sticky ones. And I don't think this is a political conversation necessarily. But, you know, I'm curious because there are some great safety and risks associated with these new generative AI systems and large language models. And, you know, I guess that calls into question what the focus of the government should be in these instances. Right. Should the government be involved in some ways, shape or form? Because this could be so potentially devastating to the employment, especially here in the US where we don't have really, laws and we don't have really regulations. So speaking of laws and regulations, California could be the first. All right, so California's SB 1047. Let's talk about it. So the California State assembly has passed SB 1047, a significant proposal aimed at regulating advanced AI models. It passed with a vote of 48 to 16, by the way. So the legislation has garnered attention for its potential to shape AI safety standards in the state and beyond. So, yeah, if you are listening overseas. Well, first of all, thank you. We have listeners in hundreds of countries, so thanks for that, y'. All. But here in the US if you don't know, California really just drives a lot of policy here in the U. S. So, you know, that's why this is actually worth paying attention to. It's not. You know, very rarely do we cover what's going on in one single state. But a lot of times California does set the tone for the rest of the US or some of its policies and laws that, that it enacts does force companies that operate internationally to change how they operate, which is why this is extremely important. So the proposal, and we talked about this last week, it's divided, right? It's divided all of Silicon Valley. But it has passed. Right, but it's not done yet. So the bill still faces strong opposition from major tech firms as everyone's awaiting to see if California's governor signs this into law. So opposition has come from OpenAI and Meta, as well as a lot of very prominent VCs who argue that it could stifle innovation and also hinder open source developments. However, polls indicate that a substantial 70% of Californians support the bill. And that's, that's important, that's important to talk about here. So another thing is, despite its popularity, Governor Gavin Newsom. So California's Governor Gallon Newsom hasn't yet hinted on a decision on whether he will sign the bill, as there's still significant lobbying efforts from the tech industry. So here's, here's what this all boils down to. Okay, so, so the legislation for SB 1047, here's what it's all about, y'. All. It includes provisions for whistleblower protections and mandates safety plans for AI companies spending over $100 million on model training, aiming to balance those safety concerns with California's need for a more regulated startup ecosystem. So critics have mischaracterized the bill, claiming it imposes excessive penalties and criminal liability on AI developers, which has contributed to confusion and fear surrounding the bills or surrounding its implications of potentially being passed. So the bill's passage could set a precedent, like we said, for AI regulations across the US and influence global state standards as other countries look to California for guidance on AI safety laws. If signed into law, SB 1047 would mark California as the first state to implement such a robust AI liability rule, potentially impacting the future landscape of AI development and the tech industry. So it's pretty big. Yes, it is a bill for California, y'. All. But if this is passed, if the governor does pass this and if it's not vetoed. Right, because that could always happen. Yeah. I used to cover politics back in the day as a journalist, so there's a lot of ways that this could go. It's. It could still drag on. Will this even happen before the elections here in November? I don't know. Because that could impact races there as well. Because this has, you know, and not just politically, this has really divided Silicon Valley, which is where actually a lot of the world's innovation happens. Right? So essentially all of these companies that make large language models are based in California. All right, so if this does pass into law, you know, I don't know how this could work. Right. But it could. We could in theory see something like companies releasing two completely different versions of a model. Right. Or they might have to kind of give you, give us all this watered down version and release it everywhere. That's just speculation, but it's something that in theory could happen. Let's keep this train rolling. Choo choo. All Right. So Google has launched some pretty significant updates to its Gemini AI platform with Gems and updates to its base model in Gemini. So Google has unveiled some major updates to its Gemini AI platform, introducing features designed to improve user experience, experience and accessibility in the ever competitive AI market. So first, let's talk about this. So there is the new Gems feature, which is similar to OpenAI's GPTs. And this feature allows for users in over 150 countries to create specialized AI assistance for tasks like coding and marketing, democratizing access to advanced AI tools. And y', all, we did do a complete show on Gems last week and I'm curious, did anyone try Google Gems? Did you like them? I had a whole show on them. So you can go watch that if, if you want or listen to it on the podcast or read about it in the newsletter. So the introduction of task specific assistance aims to improve the relevance and efficiency of AI responses, addressing limitations in those more general purpose models. Right. So you know, essentially you have to spend a lot of time when prompting, you know, doing some swift prompt engineering to get good results out of a lot of large language models by default, which is where some of these customized or smaller versions of these models really come in handy. All right, that wasn't the only upgrade that Google rolled out last week. So Google also upgraded its image generation capabilities with Imagine3 promising higher quality images and implementing ethical safeguards like Synth ID watermarking to combat misinformation. And y', all, that's not going to solve any problems, let's be honest. All right. Also, it updated its base models as well, but these models are only the ones available kind of in its cloud version and not on the front end. If you're using Google Gemini, so you're still using an older version, but Google's. If I'm being honest for developers, Google's been crushing it. So they announced updates to their Gemini, all their Gemini AI models, including a smaller Gemini 1.5 flash 8B or 8 billion parameter. So a very capable smaller small language model there and improved versions of its big boy Gemini 1.5 flash and Gemini 1.5 pro. So these developments come in response to similar updates and offerings from competitors such as OpenAI Microsoft, which both released some large language model updates and small language model updates in July and August. All right, Jay says haven't tried Gems yet. Looking forward to that. All right, hey, if anyone else did, let me know. I wasn't a fan. I wasn't a fan, if I'm being honest. All right, here we go. Here is the big news, y' all gotta save the best for last. So open AI's new AI model called Strawberry, codenamed, is set to enhance training for the upcoming Orion. Let's explain. All right, so according to reporting from the information, OpenAI is preparing to unveil a new kind of AI. I don't even know if it's technically a new AI model. Let's call it a new AI mode called Strawberry, which is generating a lot of buzz due to its role in developing OpenAI's next large language model, codenamed Orion. So this release is very noteworthy as it promises to improve AI capabilities significantly. All right, before I get into my bullet points here, let me break this down. What it means. So previously we heard reporting, going all the way back to last year about something called Q Star, which was essentially this, this model or this mode that OpenAI had developed that was reportedly shockingly good, like kind of scary because it possessed a level of human reasoning. So this kind of Q Star project was renamed Project Strawberry. And now we are getting new reports from the information that is talking about. Strawberry may be technically slower and it may not actually be kind of the end model or the end play here. All right, so let's talk about that a little more. So Strawberry is designed to create high quality synthetic data that will aid in training for Orion, which could in theory be GPT5. It might be called GPT next, it might be called GPT Orion. But it does look like the Orion is the next model. And it looks like Strawberry is more of a mode that can help produce synthetic training data. Right, because there's been this kind of not talked about problem except in geek circles about, well, how can we keep making these models more and more powerful if all of the training data has already been scooped up? Right? So that is reportedly one thing that this new Strawberry is doing. So think, think of Strawberry like this. You have new cars that come out each and every year, right? That's not Strawberry. So reportedly Strawberry is going to be a new engine that is going to help create the next car, which is Orient. So the enhanced reasoning abilities of Strawberry are expected to address the current limitations in AI, particularly in tasks that require symbolic reasoning and content sensitive problem solving and pattern recognition. So the new Strawberry program will extend beyond training Orion or Orion. Not sure about the pronunciation. All right. It will also be capable of handling complex tasks such as devising market strategies and solving challenges, challenging word games, thereby broadening the application of AI. Again, this is all according to reporting from the Information. Here's the thing we kind of talked about a little bit earlier. But OpenAI has reportedly collaborated with government agencies to ensure that the development of Strawberry aligns with regulatory standards and addresses national priorities. So this partnership aims to enhance the safety and ethical deployment of AI technologies. So the development of Strawberry has been somewhat secretive. Somewhat, not so much. However, we have seen, you know, OpenAI CEO Sam Altman subtly confirming the project's progress through social media posts. He just posted a picture of a Strawberry, right. So kind of a subtle nod, which is something that Sam Altman has done multiple times in the past on Twitter. So the integration of Strawberry into Orion could lead to a significant LE leap in AI performance impacting various sectors including education, business, technology, marketing, communications, customer service, everything, y'. All. So this upcoming launch, whenever that may be, aligns with OpenAI's ongoing efforts to innovate and expand the functionality of AI technologies. All right, that is a lot, y'. All. Let me quickly recap. Here we go, bullet points, the AI news that matters for this week. So Amazon's new Alexa is set to launch with Anthropic's Claude AI as the engine. According to reports, Nvidia and Apple are looking to get in on OpenAI's next funding round, reportedly at $100 billion. Google's game and gen model has marked a breakthrough in AI driven gaming technology as it could create literally the classic video game Doom in real time. Mind boggling. Klarna has planned some major job cuts as AI investments are reshaping its workforce and its revenue is reportedly climbing. In the midst of that, AI giants are reportedly collaborating with the US Government for safer technology development. As we talked about anthropic and OpenAI, California's SB 1047 is moving closer to potentially getting passed as it awaits Governor Gavin Newsom's signature. Or it could be vetoed. We'll see and find out. And that has some pretty wide ranging impacts. Then Google has launched some significant updates to its Gemini AI platform as well as gems. And then last but not least, some new reporting and details coming out about OpenAI's new mode. Strawberry is set to enhance training for its next model, codenamed Orion. All right, that's it y'. All. That is the AI news that matters for this week. If this was helpful, I hope it was. Because hey, guess what? Now you are the smartest person in AI at your company, right? Let's be honest, if you tune in every day, if you read our newsletter every single day, you are now the person in your company. Or maybe you're looking for your next job, guess what? You are solidifying your next career moves and your trajectory. Because the future of business, the future of work is generative AI, Period. There's no denying it. So by tuning in at least once a week, but hopefully every every day, you are staying at the cutting edge. So if this is helpful, it takes y' all sometimes 3, 4, 10 hours to put together a single show. And we do this every single Monday through Friday. So it takes you 3, 4, 10 seconds to go ahead and click that repost button, that Retweet button, or to go ahead and click that subscribe button. So if you are listening on the podcast on Spotify or Apple, please subscribe. Leave Us a Rating Share this with a friend and join us 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.