Everyday AI Podcast – An AI and ChatGPT Podcast - EP 423: AI News That Matters - December 16th, 2024

Episode Date: December 16, 2024

Did you see Google quietly released real-time AI. Apple might be coming out with the cheapest AI phone to date. And OpenAI and Google were like two heavyweights going 12 rounds, blow for blow with new AI releases. Yikes. We bring you the 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. OpenAI Updates2. Apple Intelligence Update3. Eric Schmidt on AI4. Klarna's Use of AI5. Google AI updates6. Microsoft AI modelsTimestamps:00:00 AI week: OpenAI vs. Google updates battle.06:16 New chat tool added, Python only, limitations.09:53 OpenAI's ChatGPT launches project organization features.11:05 Create chats, access documents, update projects easily.16:50 Eric Schmidt concerns self-improving AI, suggests monitoring.19:27 Klarna still hiring, despite AI job replacements.21:49 Devon enhances code workflow with automated tasks.24:45 Future: Specialized small models using synthetic data.31:03 Sign up for Google's trusted tester program.34:27 Google's Android XR: Smart glasses with AI assistant.37:03 Google launches NotebookLM Plus with paid tier.40:30 Google's Project Mariner: AI web navigation extension.41:53 Only works in active Chrome tabs.47:35 Google AI Studio offers advanced features now.48:35 Become your company's AI expert; share widely!Keywords:OpenAI, ChatGPT, projects feature, Apple Intelligence Update, Siri, Eric Schmidt, AI innovation, Klarna, workforce automation, Cognition's Devon, Microsoft's PHI 4, Gemini 2.0, Google Agent Space, Project Astra, Android XR, NotebookLM Plus, Project Mariner, Jordan Wilson, AI podcast, Notebook LM audio feature, Deep Research, Perplexity, Google AI Studio, advanced voice mode, Everyday AI, AI week, Sora, Canvas tool, video generation, screen share capabilitiesSend 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. It was never actually called this, but everyone has been calling the last week in AI, AI week.

Mainly because Open AI in Google have been in a straight up slug fest going punch for punch in releasing very impressive new updates to its respective chat GPT and Gemini products. But there was a lot else going on this week in AI news. And we're going to be covering it all today. What's going on, y'all? My name is Jordan Wilson. And this is Everyday AI. Welcome. This is your daily live stream podcast and a free daily newsletter,

helping everyday people like you and me, not just keep up with what's going on in the world of AI, but how we can all actually get ahead to grow our companies and to grow our careers. So if that sounds like you, you are in the right place. Thank you for tuning in. So everyone from Brian to tune it in from LinkedIn, Michael Phillip, joining on YouTube. Thank you all.

We do this almost every single Monday going over the AI news that matters because you can spend literally hours every single day trying to keep up with AI news, but you can't, right? That's what I do. So that's what you pay me for. Well, you don't pay me anything. You just tune in live and spend on Mondays. 40-ish minutes and you become the smartest person in AI at your company because you know everything

that's going on and why it actually matters. So if you haven't already, please go to your everyday AI.com. You know, you learn here, but you leverage it there with our free daily newsletter where every single day where we recap the show insights that you really need to know to grow. I didn't mean to rhyme. That was an accident. But you can do that there as well as go. to and watch and read more than 420 episodes that we've had of the everyday AI show. So no matter what area you're in, marketing, sales, healthcare, doesn't matter. We've had expert guests from around the world there. So without further ado, let's get straight into the AI news that matters for the week

of December 16th. All right. Let's get after it, y'all. And we're going to do it a little different this week because there is literally so much open AI and so much Google news. We're going to sandwich it like this. We're going to start with everything OpenAI. We're going to get to everything else that's not Google.

And then we're going to get to all the new Google AI news at the end. Yeah, it was one of those kind of weeks. I think we have five stories from Open AI, like seven from Google and a lot more. So it's going to be fast and furious, y'all. But to our live stream audience, thank you for tuning in. Let me know which one of these. shows or sorry, which one of these news pieces is going to impact you? I love going through after the show and reading comments from our live stream audience. Thank you so much. So let me know what

you guys think of all of this that's going on. All right. First, and like I said, this is going to be a very, very fast roundup of each story because we got so many to get to. So Open AI has released SORA. Yeah, this happened like three hours after our show last week. So we had to fit it in. But Open AI has released SORA, their new AI tool for generating videos from tax accessible to premium chat GPT users. So whether you are on the $20 chat GBT plus or the new $200 chat GPT pro, you will have access to SORA. And they did reopen signups. So they did have to shut signups down because their servers were legit melting. Signups are back up.

So the new tool can create high quality videos like whatever. Sumo wrestling bears, right? But you cannot create videos of humans if you are on the $20 plan. That's important to know you do have to be on the $200 chat GPT pro plan in order to do that. So right now the company is prioritizing the prevention of harmful content and is working with artists and policymakers to address concerns because there's a lot of concerns when a AI video tool as powerful as SORA hits the market. I'll say this, there's plenty of other AI video tools on the market that have had way more time out,

way more training data than SORA, such as runway, such as cling, such as Luma Labs. Open AIs with SORA, their ceiling is higher. So for every single generation, right, you can run the same generation 10 times and get 10 different outputs. If you do that with every single video generator that's out there, SOR is going to be the best, right? There's a in the same way that there's Elo scores for chat bots, right? You know, you put in a prompt and you get two outputs.

Same thing for video. And guess what? SORA is wiping the competition. So Sora by far has the highest ceiling. It doesn't mean its floor is the highest, right? If you do 10 of those, you know, some of the worst ones might be SORA. But by far, the best ones are also going to be SORA. But this is the worst it will ever be.

All right. We got to get to all the chat, GPD update. They launched updates to their Canvas tool, and they also made Canvas available to all users, including free users. So Canvas, if you haven't heard of it, it enhances the user experience by allowing real-time content editing while conversing with ChatGBT. And also the other big updates, so number one, it's now available to free subscribers. Number two, it can run Python directly within the tool.

Pretty cool. So this new integration, well, this updated integration with GPT40 enables automatic activation of Canvas because it is now a tool. So previously it was a mode that you had to select from a drop down menu when starting a new chat. Now it is kind of near the chat bar as a tool. So pretty cool. But, you know, the most powerful models such as the O1 model, Canvas does not work there just yet. also the Python thing that's a shot at Anthropic y'all you know I'm going to try to give you my two cents on all these news stories Anthropic has had a huge one huge advantage over every other large language model I would say by far with its artifacts feature which can render all kinds of code right in the browser

so this is kind of a shot at Anthropic and there's another one coming here soon from OpenAI but right now at least with Chad ChbT's canvas you can only run Python, where in Claude Artifax, you can run all kinds of code. And even if you're not a technical person, that doesn't mean anything, y'all. Like, you can use Python. You might not even know. You can upload a huge spreadsheet and just be like, yo, create me some graphs, some visualization, right? That's Python.

So, you know, if you don't think that you might use this new Python development, you're a chat TV subscriber, go ahead and upload a spreadsheet in there and toggle on the canvas mode and say, like, hey, create me a video. visual that could be helpful, create a dashboard that could be helpful, and it will do it. All right. More Open AI news, video has come to advanced voice mode. So OpenAI has enhanced its advanced voice mode with video and screen share capabilities, allowing it to see through a phone camera, and see as in quotes, as users perform tasks like,

you know, whatever you may be doing and wanting to show. advanced voice mode. So this updates comes just after Google announced it's Gemini 2.0, which we're going to get to, which also had some of these live video features. So this new chat GPT feature, essentially advanced voice mode has been out since September. And Open AI demoed this advanced voice mode with video back in, oh, it's spring event. So like, what, that's like six, six months ago. So what this means, you now have these.

advanced voice mode inside of chat chbt. We did a quick review on our YouTube channel that we also shared in our newsletter. I had some problems with it at first. I've tried it since. It's gotten a little better. You know, like I said, open AI's servers were getting crushed last week. But essentially, think of the advanced voice mode if you've used it before, but now with video, right? You could be working on a math problem, on a sheet of paper. You could be, you know, drawing something out on a whiteboard, you know, something strategizing for your business. And you automatically, have a smart AI assistant that can quote unquote see and collaborate with you in real time. Michael says, I love the video feature, but haven't discovered much usefulness to it yet.

Interesting. So far, he's just used it for fun. Dr. Harvey Castro says, I love the screen share and video open AI features. It remembered my dog's names. Yeah. Hey, podcast audience, too. Hit me up. I always have, you know, in the show notes, you can email.

us, you can text the show, although I can't text you back, FYI. Also, I put my LinkedIn in there. So I would love to hear use cases for those of you that are using advanced voice mode with video and have found great uses for it. I would love to hear. All right. More Open AI. So another shot or maybe taking a page out of Anthropics book, OpenAI has introduced projects

in chat GPT for, well, just better file. organization and some new features. So OpenAI has launched projects in ChatGPT, a feature that allows users to organize files and conversations similar to, well, Anthropic Claude's projects and also similarly-ish to Google's notebook L.M. So right now, this is initially available to ChatGPT Plus Pro and Teams subscribers. This feature enables users to create folders, customized project settings, and and upload documents.

This has been one of the lowest pieces of hanging fruit, I guess, right, for, since chat GPD came out, right? So many people are like, it's hard to organize everything, right? Why can't we have folders? Well, you finally have folders with projects. But it's a little bit more than that because you can also upload documents that are shared between all the different chats. So again, if you've used projects inside of Claude Anthropics Claude, it's the same

thing, right? So think of it like this. You can have a project and you could, let's say, upload five documents inside of that project. And then you can create new chats, separate chats that all live within there that share access to those documents as well. So I'm sure there's going to be updates to projects in the future. And right now, unfortunately, at least as of like eight, nine hours ago, the new 01 model, you cannot put those chats or work with those chats inside projects. Also, good to know, you can move old chats into new projects. All right. So you can kind of right click or option click on your chats on the left hand side and add them to a new project.

So very useful is a feature that honestly, I'm surprised that took this long, but I'm glad it's out. Yes. Mariam from LinkedIn says, thank goodness for projects. We can tidy those chats up. Yes, but also you can upload it's it's it's it's not you know, uh, you know, kind of like what Dr. Harvey is saying here, you know, saying this is kind of like mini rag. It's kind of, right. Uh, it, it is a way to bring your companies, uh, or your, uh, first party, first company, your data into chat, TBT and to work with that first. I have been doing some testing and so far, uh, actually, it seems like project does this, uh,

document retrieval process, actually a little bit better than custom GPTs. So there's still huge benefits to custom GPTs. We'll save that for another day. All right. Next piece of open AI news, probably the smallest one of the week. But actually, it should have been huge, but we've just heard about this for so long. Apple intelligence. So yeah, Open AI and Apple got super official with the release of Apple intelligence powered by chat.

GPD. So Apple has rolled out a significant update to its Apple intelligence platform focusing on AI powered everything. So essentially now it has access to chat gbt. Siri is hopefully a little smarter. So this new chat GPT integration works on the newest operating system. So that's iOS 18.2 or if you are on a Mac, that's Mac 15.2. So the update introduces a chat GPT integration with Siri, which is kind of funny. Siri has always been our AI assistant, but hasn't been very smart. Now Siri has an AI assistant because essentially when you give a query to Siri and Siri doesn't know, Siri just like says, yo, chat GPT, can you help me out with this? That's literally what happens for complex queries.

Siri slash Apple intelligence just essentially calls on chat GPT, right? And there are settings within Apple's new updates where you can kind of have that process go automatically. Otherwise, it will prompt you. Yeah, funny, right? you're giving a query to what is supposed to be a smart Siri, and instead Siri prompts you to use another AI. So meta, and I'm not talking about the company. So like I said, users can now prompt Siri to use chatDBT to answer complex queries

and improve productivity across iPhones, iPad, and Mac devices. So yeah, and then, you know, all the new writing tools are out, which I don't know. All the Apple intelligence stuff to me, it's a big fat, me. You know, it's a nothing burger. This is all stuff that we had through other tools pre-chat GPT, right? We've had a lot of access to what we now have access to and Apple intelligence, aside from some of the new stuff that I think is marketing at best, right?

Like, oh, AI gen moji. I don't care. I don't need to send custom emojis. I'm already bad enough at texting. So, yeah, interesting. What do you all think? Yeah, Jack, love what Jackie here says. She says, Siri is the middleman.

series of middleman or middlewoman now, just, you know, passing off all our queries to chat GPT. That's funny. All right. Now we're technically pivoting away from we got through OpenAI news. Now let's get to non-open AI, non-Google news. So there may be a cheaper iPhone that can handle Apple intelligence. So according to reports, the upcoming iPhone SE4 is, set to include Apple intelligence, leveraging Apple's A18 chip to bring advanced AI features to Apple's budget smartphone. So here's why this is pretty important. Well, right now, if you want a lot of these new Apple intelligence features on your iPhone, you have to have one of the most powerful iPhones. You have to have an iPhone 15 Pro or higher. And in most cases, that's more than

$1,000. So Apple has, for many years, had a kind of budget version of its iPhone called the S-E, which I believe is that standard edition? Does anyone out there know? I think it's standard edition. Anyways, now this rumored new iPhone SE4 may have the, essentially, the chip required to run Apple intelligence. So reports are saying that the phone could be either between $4.99 and $5.99. So essentially, if you do care about having Apple intelligence, but you don't want to pay, you know,

somewhere between 1,000 to 1,500, 1,800, if you want one with a lot of storage. There's a budget option. So pretty cool there. And I do think that that is actually going to move the needle. All right. Moving on. Can we unplug AI? I don't know.

Former Google CEO, Eric Schmidt says it's not too late. So in a recent interview, the former CEO of Google, Eric Schmidt, has raised concerns about AI systems that can self-improve, suggesting it might be necessary to quote-unquote unplug them once they reach this level, the level of being able to self-improve. So that's according to an interview that he was on in ABC's this week. So as the AI field rapidly evolves, Schmidt highlighted the unprecedented scale of innovation, warning of the possible unforeseen dangers and the need to carefully manage AI. He emphasized the importance of America leading the global AI race, particularly against China,

and suggests building a secondary AI system to monitor the first AI system for safety. Yeah, what happens when AI goes rogue? Well, you've got to create another AI that is supposed to keep the AI from going rogue, right? So he also predicted that AI systems capable of independent decision making could emerge within two to four years, right? These are things that 10 years ago, people said we're 50 years away, the ability for AI to self-improve, self-heel, essentially when the next version of AI is built by AI. So someone that knows a thing or a million, the former Google CEO, said that that could be two to four years away. All right. Well, here's another thing that is really changing in the AI space.

Klarna is making news again for not high. human and instead giving all of the human work to AI. So, Klarna's CEO,

whose name I'm definitely not going to get right, Sebastian Simeiatowski. I didn't get that right. I'm sorry, Sebastian. So we'll just say Sebastian. So Sebastian

said that the company in an interview has reduced its workforce from 4,500 to 3,500 over the past year. So essentially what's happening is he,

said, Sebastian said, well, there's natural attrition at any company, including Klarna, and instead of hiring 20% of its workforce, so he said in general, maybe about 20%, you know, you might have to rehire year over year. Instead, Sebastian says, ah, we're just not hiring people anymore. When people leave or when they retire, or when we maybe terminate their contract or their position, we just don't rehire anymore. So they're down and y'all, going from 4,500 to 3,500, that was a big drop-all. So, despite the CEO's assertion that AI can replace human jobs, Klarna is still at least posting for over 50 roles right now. So indicating that there is still some reliance, whether they're actually hiring for those or not, we're not sure. So Klarna's hiring

activity is mostly focused on backfilling essential positions in engineering rather than expanding its workforce. So as Klarna prepares for a potential IPO, the company is showcasing its AI integration to appeal to investors. The broader AI adoption remains gradual across industries. So yeah, Klarna was one of the bigger companies earlier this year that essentially said, nah, we're just giving all or as many human roles over to AI as possible. So Klarna was one of the big, you could say, AI case studies of just essentially unabashedly

handing over human roles, being like, nope. Nope, we're not hiring humans anymore. We're just giving all these human roles over to AI. So it should be interesting to see. Number one, if Klarna does IPO, are they going to hire more people, right? Are they going to continue to reduce their workforce and rely more on AI? Well, time will tell. 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 Premier, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations.

Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adop.com. More big price tags, right? If you thought that $200 chat GPT Pro price tag was a lot, well, how about Devin from Cognition, which is now available for $500 a month per user? So Devin is an AI tool for engineering teams.

It is now generally available at $500 a month, offering no seat limits and integration with Slack, different IDs, and APIs. So essentially, if you haven't heard of cognitions, Devin, what they're really kind of positioning it as, well, it's a junior developer. All right. So I think there's so many great tools out there, right? You got to tip your hat.

Claude is great at coding. You have GitHub copilot. Now you have 01, right, from opening I. You have all these large language models that are great at individual coding tasks. So Devin aims to be a little different. It's more of like, hey, you don't have to keep prompting. Just give your files, give your commands, come back later, it'll be done kind of thing. So it kind of grabbed a lot of headlines when it was first teased many months ago. So now it is generally available. So a lot of companies have been wanting to get their hands on Devin from

cognition, but they haven't. Now you can. So recommended uses include handling small front end bugs, creating first draft press releases and making targeted code refactors, enhancing workflow efficiency. So Devon has successfully assisted in real world scenarios like resolving issues in open source projects, adding features to libraries, and fixing bugs in various repositories. I don't know. Any people in software development out there, I would love to hear directly from you all. Is this, is Devin?

Is it shaking up your industry? right? If you are in software development, if you're a software engineer, if you're code or web developer, et cetera, I don't know. Is Devin something you're excited to? Is it kind of like Klarna in the dark, right? Or you're like, this thing's going to maybe take my job or I'm only going to be using Devin. Yeah, let me know. Super curious. Yeah, Dr. Harvey was saying my business partner was telling me about Devin AI, Tara saying, I wonder how it compares to Replit. Yeah, Replitt's got a great AI agent. I mean, there's so many now in this space.

You have windsurf, which is another newer kind of tool. You have cursor. I mean, there's so many great kind of AI coding tools now that essentially connect right to your database. So it's no longer you have to copy and paste two ways, right? Copy the code into a large language model, work with it, go back and forth, copy that code out, put it back into your repository, into your software development stack. No, it just connects directly now.

So should be pretty one worth keeping an eye on there. All right, Microsoft very quietly unveiled Phi-4. So that's PHI-4 in case you're looking it up. A new language model, a small one. So Microsoft Research introduced Phi-4, a 14 billion-perimeter language model designed for efficient reasoning tasks offering a competitive edge over even larger models on certain benchmarks. So, yes, this mini, mini, model, right, a 14 billion parameter model is outpunching GPT40, Lama 3 in certain benchmarks like math.

All right. So just for reference, GPT40 is reportedly 1.8 trillion parameters. So this 14 billion parameter, that is a fraction. That is like 1% of the size. And it's already outpunching. This is where AI is going. I've been saying this for a long time. Small models.

I do think the future, we are going to be working with hundreds of specialized small models. We're not going to be working with a jumbo model. Or I think all that jumbo model is really going to do is it's going to handle some general tasks. But eventually it's just going to pass your query on. I think these jumbo models are going to have hundreds or thousands of smaller models house within them that are built for specialized tasks. So 5-4 utilizes high-quality synthetic data.

So going over the trend like meta using synthetic data or AI generated data to help create it. So the model's post-training refinement includes direct preference optimization, enhances output accuracy and usability, making it practical for real-world applications. So like I said, by four excels and benchmarks like GPQA, math, and human e-val, showing its advanced problem-solving capabilities and valid. its utility in real world math competitions. All right. I got to take a sip of the coffee because here we go. We are done with the open AI news. We are done with the biggest news.

That is not Open AI and not Google. And now we are officially on to the Google portion, y'all. So Open AI, I think, did a great job, right? They had this, you know, 12 days of Open AI. So we are seven days in. we have five days left to go. Google didn't really create any marketing, any messaging, seemingly any real strategy around what was dropping.

So their head of development, Logan Kilpatrick, I tweeted at him like two weeks ago, and he essentially said, yo, we're going to be releasing so many new updates in the coming weeks. I didn't really think of it. Like, I didn't think of anything, right? because the first kind of week of Open AIs or the first couple of days, right, we didn't hear anything from Google. And then the middle of last week, Google went bananas, B-A-N-A-N-A-S, like they went wild. And I will say this.

A lot of the stuff that Google, quote-unquote, released, it's not released. So in typical Google fashion, we got some great stuff. And then we got some, you know, some, some, some, some, some, some, um, some, um, some, um, some, um, um, um, um, some teases, some updates. So I think Open AI brought us more things that we can use today. Google, though, had, I think, their best three days in the last three years. All right, let's go over it very quickly because we got a lot. All right.

So first, Gemini 2.0, big jump. And Google is entering what it calls the agentic air. So Google DeepMind has launched Gemini 2.0, a highly advanced AI model designed for the evolving, quote-unquote, egentic area, offering significant enhancements in multimodal capabilities, including native image and audio output. So the new model that's being released initially is the Gemini 2.0 Flash. So we don't have the big boy. We don't have Gemini 2.0 Pro. We don't have Gemini 2.0 Ultra. We have Flash, which is supposed to be the cheap, fast model.

guess what? It is already outbenching the Big Boy 1.5. So even though it is a flash model, which is supposed to be similar to Open AI's mini models, right? You think of it as, ah, not very powerful. It's just the fast to cheap version that's great for API use. No, this thing is a banger. All right. So the new model, Gemini 2.0, is also being used across all of its new products. So we're going to talk about a couple of these or most of these, but like Project Astra, Project Mariner, and Jewels. So Jewels is essentially a new AI coding tool. We're not going to get too much into that today.

But big news there, we do have Gemini 2.0, at least the Flash version. So what I would assume in the coming months that we're going to see the Gemini 2.0 across the pro or the ultra, as do all companies, Gemini Google does have some unnamed new months. bottles that are being tested out in the wild on the LMS Arena chatbot leaderboard. So we're going to have more than just flash pretty soon. Yeah, thank you. Tara got my Gwen Stefani reference. I don't know if I aged myself.

I don't know if anyone else got that. It was accidental. I promise when I'm making these random cheesy quips, they're not planned. I'm just a dork. All right. Next, well, if you are a dork, you're going to like this from Google. they launched agent space for enterprise AI solutions. So Google Cloud has introduced Google Agent Space, a multimodal search agent aimed at enhancing

enterprise operations by integrating advanced AI reasoning and search capabilities. That was a mouthful. So this platform allows businesses to create a company branded search agent, right? This is wild. All right. Providing conversational assistance and proactive support. through integration with tools like Google Drive, Confluence, and Microsoft SharePoint. Yeah, you can even work with your Microsoft tools over there in agent space.

So employees can access AI agents and use low-code tools to build custom expert agents with embedded features such as Gemini for advanced reasoning and integration with image and video generation tools. So not everyone can get it, right? Yeah, Google, here we go. Typical Google go to market. I'm not a fan. think open AI is crushing it and go to market.

Outside, you know, I think the SORA left a little bit to be desired, the advanced voice mode with video, right, having to wait multiple months. But everything else, you know, opening I go to market, great. Claude just ships. They don't even really bring much marketing around it, right? But if you want this agent space from Google, sorry, you can sign up right now for early access. Are you going to get it?

Probably not. But you can go at least sign up. Who knows? Maybe, maybe you're all. already in Google's good graces. I know Google does have their kind of trusted tester program for individuals. And I think they have something similar for enterprise organizations. So who knows?

You may be able to go get it right now. And we'll see over time if this is kind of a one-to-one competitor with Microsoft, Microsoft 365 co-pilots, Copilot Studio. It does look like that's what it is. But we don't really have any great information on this right now because so few people have this, right? I'm trying to see a bunch of reviews about agent space and there's nothing out there. So I don't know if, you know, two companies have access to this, if 200,000 companies have access, but you can go at least in Google fashion, you can go put your name on a wait list.

All right. Speaking of waitless, there's updates to Project Astra, but it's still not available. All right. So Google did showcase some new updates to Project Astra. Astra, its advanced AI assistant that offers a glimpse into how AI can assist in navigating the physical world. So Astra, well, let me just say it in basic terms. You're going to probably in the end, you will be wearing glasses.

Are glasses coming back again? I don't know. Well, Google, Meta, everyone else, they're really going all in on these glasses, even though the Google glasses, I don't know what that was 10 years ago didn't really work. But essentially, Google showcase some updates to Project Astra, which think of it as a live Gemini, right? But it can see what you see. So at its I.O.

Conference, Google first demoed it mainly with the app, with the Google Gemini app. So if you are one of Google's few trusted testers, you'll have access to that, which not many people are. And I do believe it does require a new Google smartphone from San Francisco. Samsung as well. But for everyone else, you can still go get on a wait list. But think of it like this. It looks like what Google's trying to do here is bring the glasses back.

I don't know if glasses are the form factor of AI. I think for limited use cases, great. My wife actually just got me the meta glasses. I've been so busy. I haven't even been able to try them out yet. I think for certain spurts, they're great. But I don't know. I don't know with these Google ones, right?

the way that they're kind of marketing them is, oh, you should be wearing them all day, right? Because they can navigate you. You can have like, you know, essentially, you know, projecting things onto the screen of these glasses. So I don't know how realistic that is, right? And these are a little heavier, right? The thing I like about the meta raybans, they look just kind of like raybans, right? They don't look like these big, fat, thick things, right?

A lot of these newer, quote unquote, smart glasses, the ones from meta as well, the other ones not the Rayband collaborations. I think that one's the Orion. They're big, fat, thick things. So I don't know. I don't know if people are going to want to wear around these, you know, super thick, heavy glasses. I don't know. Tara says she wants them.

Dr. Harvey says, I think meta-AR-V-R glasses is more of the future. We shall see. Speaking of glasses, well, these two Project Astra and Android X-R kind of go hand in hand. because when I'm talking about smart glasses, that's kind of where we're headed here. So Google in collaboration with Samsung and Qualcomm announced Android XR, a new platform for extending reality devices, including both headsets and glasses.

So this brings the AI-driven Gemini assistant, right? Straight to your eyes, straight to your ears. And it is central to the Android XR platform, being able to understand user intent and assist with tasks such as planning, and research through conversational interaction. So Android XR will debut first on headsets with Samsung's Project Mujan expected next year, offering immersive experiences like virtual big screens and for various apps. The platform invites developers to utilize familiar tools for creating diverse XR experiences,

aiming to build a robust ecosystem for new devices. So yeah, obviously Apple's Vision Pro. that thing flopped. Before it came out, I said this thing's going to flop. I said, this is going to be the least successful Apple device ever in. It literally, no one, no one bought it, right? There's reports that they're, you know, that they're, you know, now really, you know, slowing down production and they might not be updating it as much as they originally thought,

because who would have thought? There's not a bunch of people with an extra four grand in their pocket that want to wear a 20-pound headset on their head. So, I mean, we'll see with the Android XR. It's obviously way cheaper. It looks way lighter. But again, I think maybe for certain people, right, if you're working at home, maybe something like that could be great.

But for out in the real world, I don't know. I'm not sold yet on wearing around, you know, an AR, XR mixed reality headset, right? So what that is is, you know, it's a headset. You wear it and it projects, you know, you can see both what is happening in the real world, but then it has this mixed reality, this X-R element to it. And then it brings in this conversational agent with Gemini. Couple more stories, and I'm saving the three biggest ones for last. Yeah, we're still on the Google segment.

My gosh, I said they went all Gwen Stefani on us. So notebook LM. Yes, one of my favorite tools. What I would say has to be in the running for AI Tool of 2024. notebook LM from Google has some big updates that are rolling out. So the biggest one is a new paid tier, right? So it's been free. Now there is a paid tier.

So Google has launched Notebook LM plus enhancing its popular app with features aimed at enterprises, teams, and individuals who use the app's research tools extensively. So I believe I did, I've been. chatting a little bit on Twitter with some of the team there from Google's Notebook LM. And it seems like if you are already on the paid version of Google Gemini, that you will have access to this. So it does seem like it is both a personal premium plan, but also a team premium plan as well. So being able to share this with businesses, or sorry, within your team members. So I have obviously multiple paid Gemini accounts, both on my personal Gmail and on my Google

workspace plan. I haven't seen this roll out yet. So who knows, it may be rolling out soon, but it does look like it's already being gradually released. The big feature that I'm really looking forward to is the updated audio feature. So essentially now you can quote unquote call in or buzz in, right? So if you haven't to use notebook LM, it is a state-of-the-art rag model. Right.

If you don't upload your data, you literally can't use it, right? Which is really cool. I like that. I think more AI models should at least have an option to operate like that. But there's always been this kind of deep dive podcast, right? So you can put in millions of words, literally you can put in millions words, millions of words of something you're trying to learn about your company's data, whatever.

Click a one-click audio overview and it creates a cool, personalized podcast, right? with two hosts that seem human-esque, right? So now the last update about two months ago, you could customize or give instructions to the AI hosts. So now you can quote unquote call in, which looks like a groundbreaking feature for learning. All right. So essentially, as you're listening to the deep dive,

you can essentially interrupt them and ask a question. You can say, hey, what does this mean? Or, hey, could you explain that a little more and maybe use a basketball reference on the fly. So again, my brain hurts, and I thought about this a lot since it was announced. I don't think people are talking about this enough. I think this one little feature, this wasn't even the big update, right? The big update here is now there's a pro plan and you get five times the limits.

That's great, right? And you can share all this with your team. That's great. There's a new, you know, writing pain to create content. You know, so now it's going to this three-tier pain, which I think is really. cool. So now it's also turning into a content creation tool and not just a learning tool that you can ask questions on. But I think this kind of call in feature is one of the most exciting small features that everyone is overlooking. So don't sleep on that. Yes, Jackie. Now, hey, now the

live stream audience is playing along with me. Jackie says, call in question mark. That's bananas. Yeah, old school radio style. All right. Two more. And I think, again, I saved the best three for last. Next, don't worry about the headline here, live stream audience. This is Google's Project Mariner, a new AI agent navigating the web. So this was previously codename Jarvis.

We talked about it on the show a couple of times. But now this is, again, it's released to quote unquote trusted users, which I think is like hardly no one, but it is starting to roll out. So Google's new Project Mariner is powered by Gemini 2.0, and it is an AI agent that performs internet tasks through the Chrome browser. So it is a Chrome extension that just does things for you on the web. So very similar to Anthropics computer use, which we demoed here on the show a couple of months ago.

However, computer use super buggy. It's very technical. you actually have to download and install multiple programs, right? You have to download or, you know, you have to grab a bunch of information off GitHub. So it's not for non-technical people. You have to be pretty technical to use Claude Anthropics, or sorry, Anthropic Claude's computer use.

So Project Mariner, it's a Chrome extension, right? And then you essentially say, yo, Mariner, go do a bunch of this stuff. And then it goes and does it. The big caveat here, though, or the big downside is it only works in your active Chrome tab. Okay? So, I mean, what this means is I'm not sure if it's going to be able to work with, like, as an example, split-screen monitors. But once this comes out, I'm going to be using it all the time because I have a bunch of extra computers sitting around and lying, collecting dust. Right.

So one of them, if you can only work with an active tab, so you can't really necessarily do a lot of other work, at least inside Google Chrome. You can always open up Edge, which I like Microsoft Edge browser based on Chrome, based on Chromium. But an AI assistant coming soon, powered by Gemini 2.0, that can essentially just do whatever you tell it, right? Go do this research, you know, go to this website, find the price on this, right? Make sure these criteria are met and it just does all that for you. All right.

Let's see. Do we just have one more? Oh, no. We have multiple. All right. Deep research. I didn't do a good job at updating my headlines today for our live stream graphics. But Google also announced deep research.

And I'm telling you, y'all, this thing, perplexity, gosh, perplexity is on notice. All right. So deep research is a new AI tool available to Gemini advanced subscribers. So the paid plan. It is designed to generate detailed reports by scouring. the web for relevant information. So the tool uses Google's Gemini bot to create a multi-step research plan, allowing users to edit or approve the process as it finds and compiles key information from various

sources. Then once the research is complete, users receive a report with key findings and links to original sources with the option to expand on specific areas or export the report to Google docs. I use this. the minute it came out, not the minute, well, the minute I saw, I'm like, wait, this seems very much like perplexity. I went and used it.

It is so much better than perplexity. Perplexity, I've talked about it very recently. It's gone downhill recently, I think. A lot more hallucinations, the quality, I think, has gone downhill, especially since they now introduced the new shopping feature, right? So much of what I use perplexity for. It's comparing different products and services, right? That's something I would generally go to a lot of different websites for.

And now, instead of doing that work, perplexity with its new shopping feature, just shoves products down your throat. And it doesn't actually always adhere to the prompt that you give it, right? A lot of times I'm trying to use it to research five different products, you know, make me a chart, you know, show me the pros and the cons. Show me who's it for, who's it's not for. And instead of doing that, perplexity now with this new shopping mode, that is way too, it just, it needs guardrails, right?

But it's essentially, instead of answering questions consistently, instead just shoves products down your face, right? But this new deep research, y'all, I'm not kidding. I use this to help me research one of my shows last week. A single prompt inside, again, you have to have the paid plan. It visited, all right, and I'm not exaggerating here. 169 websites. All right.

I gave it one prompt to do a bunch of research for me. It visited 169 websites. It took about two or three minutes. Can you, can you guys imagine that? Right? We've been blown away, right? Rightfully so.

Perplexity, great. You know, perplexity might go anywhere from, you know, six to 20 websites. Chad GPT with a new GPT search. Pretty good. You know, can handle five to ten websites. This did 169. All right.

Saving what I think might be the best for last. Google AI Studio has released. Yes, released. No wait list. Released. Real time, vision, and voice with its new multi, excuse me, multi-modal live options. So Google AI Studio has introduced stream real time,

allowing users to interact with Gemini via voice and vision, providing spoken responses and visual analysis of your screen or camera feeds. So this new feature positions Google ahead of competitors like OpenAI by offering a fully integrated vision-enabled voice mode, enhancing user experience on both desktop and mobile platforms. So alongside this AI Studio also launched some starter apps, including Map Explorer and Video Analyzer, showcasing the capabilities of the Gemini API,

and also available for exploration on GitHub. So I played around with this a little bit over the weekend. Let me know if we should cover this a little bit more. Without getting into a rant, this isn't available on Gemini, right? I still don't understand. Luckily, Google Gemini on the front. front end finally got updated, right? So you do have this deep research that I just talked about.

Thank you for bringing that to Gemini. You have the Gemini now 2.0 flash. But previously, Gemini, really all it had was old models, right? Google itself said that usually those models are between three to nine months old, which in AI years is like so far behind. So most of Google's new, you know, AI features and innovation, you have to go into Google AI Studio, not its front end Gemini chatbot. But I would encourage you to do this because all these things that we've been waiting for from OpenAI with advanced voice mode, right, the ability to share your screen, the ability for it to interact with video.

Well, now Google AI Studio does this already, right? So we don't even have it yet on desktop. I do think that that may be rolling out this week or next for OpenAIs chat GPT advanced voice mode. But we have it now for Google. So that's it, y'all. I can't even re-recap these because we had like 100 stories. But I hope this was helpful. If so, please click that repost button, share this with your friends.

We put in so much work, making sure you are the smartest person in your company at AI, making sure you can outsmart the future with us. So please, if this is helpful, share this with your friends. if you're listening on the podcast, please, you know, there's all the nice little share buttons. But first, you know, make sure you subscribe and follow the show on Spotify or Apple Podcasts. Leave us a rating if you can, but share this, share this episode with your friends, family, coworkers, your neighbors, babysitters, boyfriends, dog walker, whoever it is, because we all need to understand AI.

And that's what we do at everyday AI. You don't have to have a PhD in machine learning to stay ahead. You just have to tune in with us every day. Thank you for tuning in. Make sure to go to your everyday AI.com. Sign up for the free daily newsletter. We'll see you back 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.