Everyday AI Podcast – An AI and ChatGPT Podcast - EP 409: AI News That Matters - November 25th, 2024

Episode Date: November 25, 2024

Big Tech is fighting each other for every scrap of the AI pie.  ↳ Google clapped back at OpenAI ↳ Amazon invests another $4 billion in AI ↳ OpenAI is going after a bigger piece of the market �� Microsoft silently drops some of 2024's biggest AI updates 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. New Developments at OpenAI2. Apple's AI Developments3. Google’s Gemini 1121 model4. DeepSeek's r1 Lite Preview5. Amazon's Investment in Anthropic6. OpenAI's Legal Issue7. US AI Policy Shift8. Microsoft AI InnovationsTimestamps:00:00 Google's Gemini leads AI chatbot arena rankings.06:07 Google Gemini's latest model available via API.09:50 DeepSeek lacks API and open code release.10:44 R one lite preview is free, daily limit.16:21 Google may divest Chrome due to monopoly concerns.19:39 Siri to understand screen content next year.22:30 Amazon invests $8B in Anthropic, strategic partnership.26:23 Gemini Live connected; OpenAI data deletion issue.28:53 NYT misunderstands AI, OpenAI dismisses arguments.32:25 Global AI safety event hosted by Biden administration.33:52 Microsoft unveils 90 new AI features for Copilot.39:19 Microsoft 365 cloud subscription enables scalable AI computing.40:49 Opt-in service captures and retrieves PC activity.44:00 Sign up for daily AI insights newsletter.Keywords:Jordan Wilson, AI Models, OpenAI, DeepSeek, "o one" model, "r one light preview", chain of thought reasoning, DeepSeek Chat, Open-source AI, GPT-4 o model, AI chatbot leaderboard, Google Gemini update, OpenAI web browser, Google Chrome, ChatGPT, GPT search, Siri upgrade, LLM Siri, Apple AI models, MMLU, AI chatbot rankings, Gemini 1121, Amazon's AI investment, "Everyday AI" podcast, memory feature in AI, Claude AI, Claude Alexa, AI Policy Shift, global AI safety, Microsoft AI Innovations.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. Big Tech is fighting each other for every scrap of the AI pie.

I mean, Google just clapped back hard at OpenAI. Amazon invested another $4 billion into an artificial intelligence company. Open AI is going after an even bigger piece of the market. And Apple is bringing a smarter Siri to us all powered by large language. which models, but probably not as soon as you might think. We're going to be talking about those stories and a lot more today on Everyday AI. What's going on, y'all? My name is Jordan Wilson.

I'm the host, and this Everyday AI thing, it is for you. It is your daily live stream podcast and free daily newsletter, helping us all not just keep up with what's going on in the world of AI, but how we can all actually get ahead and use all of this information to grow our companies and to grow our careers. If that sounds like you, you are definitely in the right place. If you're joining us on the podcast, thank you as always. Make sure to please check out your show notes for a lot of more updated information on what we're talking about today, as well as a link to our website.

That's where it actually happens, your everyday AI.com. There you can first sign up for our free daily newsletter where every single day we recap our podcast and give you a lot more information. But while you're there, check out. We have like thousands of hours of free AI content from the top leading experts in the entire world. So make sure you go check that out at your everyday AI.com. All right, y'all, this is the one-stop shop. If you can only join us once a week, Monday is probably it. Monday is where we bring you the AI news that matters.

So if you don't have, you know, five to 10 minutes every day to read our newsletter, or if you don't have 30-ish minutes every single, single day to listen to the podcast. Monday is your day. Join us. We recap everything that happened. And not only that, but what it actually means, we break it down. Yeah, luckily, by talking to the smartest people in the world on AI every day, we can kind of cut through the fluff and, you know, sift through the noise and tell you what actually matters. So thank you to our live stream audience like McDonald and Marie and Woozy and Tara, Philip, Samuel, Brian, Joe. Michael, who beats me here almost every day.

Thank you for joining us. Yeah, if you're listening on the podcast, you can always join us live. We do this 7.30 a.m. Central Standard Time Monday through Friday, you know, where you can ask guest questions. So without further ado, let's get into the AI news that matters for the week of November 25th. So Google has launched its latest AI experimental model, Gemini, Experiential 1121, and it has grabbed the top ranking once again. So as of at least a couple of minutes ago, Google has reclaimed the top spot in the AI chatbot arena leaderboard. So right now, their newest model is accessible

through the Gemini API and Google AI Studio. So the new 1121 model quickly reached the top of LM Arena's chatbot arena rankings, tying with open AIs GBT 4-0 in some instances, but also improving in others, including the ELO score. So according to Logan Kilpatrick, the product manager at Google, Gemini 1121 features key advancements in coding proficiency, reasoning abilities, and visual processing capabilities. So these enhancements build upon the strengths of previous models, offering more sophisticated solutions for complex problems across various fields. So the release follows the success of Google's Gemini 1114. Yeah, a big whopping one week between releases here.

But Google talk about this here at a second. They released two big updates to Gemini because Open AI sneaked one there in between to reclaim the top spot of the leaderboard. I'll probably do a story about this in the coming weeks. But it seems like a lot of the big AI labs are maybe just overfitting their models for these arena benchmarks.

So if you haven't heard of the kind of AI chatbot arena is the unofficial name. It's where millions of users vote blindly on different models. So you can put in a prompt. You can see outputs from two different models. They're unnamed. You choose which one wins. And that more or less leads to what's called an ELO score.

So a lot of the big AI labs are constantly keeping an eye out on this because, yes, there are, you know, benchmarks, right? We have your MMLU benchmarks and all these other benchmarks. But, you know, ultimately it's what do humans prefer? So there has been a lot of competition as of late. You know, we've gone from Open AI to Google to Open AI to Google, topping the leaderboard over the course of eight days. So it's been a fury of action.

So, but that's not all with Google. So according to leaked screenshots, Google plans to release Gemini 2 potentially this year, continuing its trajectory of innovation in AI. Also, Google's Gemini. chatbot has introduced a new memory feature, similar to the memory feature in OpenAI for paid users, and it allows Gemini to remember personal information, whatever you type in there, right? Food preferences, tailored responses, whatever, right? But the difference is with Google Gemini, you can go in and

add this. So a couple of points here about the new Gemini 1121 model. It is not available on the front end. All right. So if you are a everyday user, like me, like so many of us out there, if you log on to, you know, Gemini.com, you are not using this new most powerful model. So for whatever reason, Google on the front end of Google Gemini is usually using models about six to nine months old. All right. So if you do want to use this newest most powerful 1121 model, you are limited to either using it via the API or Google's AI Studio, which is really more technical. The other thing that most people don't understand about Google AI Studio is you can't really opt out of training on your data there.

So it is more experimental. The AI Studio, it's more of like a sandbox. It is not for production. So if you're using it that way, you can't necessarily safeguard your data. If you are trying to combine the 1121 Gemini model with your data, just something to keep in mind. All right, next piece of AI news. There's a new reasoning model on the block, and it's not from OpenAI. So DeepSeek has unveiled R1 Light Preview, a new high-performance reasoning AI model.

So Chinese AI startup DeepSeek has launched the R1 Light Preview. Love the names of these models, right? So it's a reasoning-focused large language model. The model is exclusively available through DeepSeek chat and is noted for its high-level reasoning capabilities and it may be better than OpenAIs O1 model, at least according to the company. So according to DeepSeek, R1 Light Preview rivals and sometimes surpasses the performance of Open AI's O1 Preview model, which was released in September of this year. So the new model from Deepseek is designed for task requiring Lerner.

logical inference, mathematical reasoning, and real-time problem-solving. So deep-seek claims that the model exceeds open AIs, O-1 preview-level performance on benchmarks like AIME and math. So users can observe the model's reasoning process in real time. That's a big differentiator. So number one, this is from a Chinese company, not from the U.S. And number two, you can see its kind of chain of thought reasoning in a much more detailed way than you can with Open AIs O1 preview. So with Open AIS O1 preview, you can still kind of toggle to see what the model is doing under the hood. But you can't really see it in a detailed fashion.

It's more of a summary. So if you are brand new to reasoning models, that's okay. Almost all of us are. But here's what I like to tell people. So to really get the most out of large language models, a lot of human intelligence is required, right? You have to interact with the model. You really have to steer it. So I like to say, you know, traditional models.

So, you know, you could say GPT-based models, Gemina, Claude, et cetera, right? You can kind of give it a destination, but at the same time, you still have to build the road and also drive. With these new reasoning models, it kind of does it for you. Think of it more of like, oh, this is, you know, this is like a Tesla autopilot, right? So you give it the destination and it kind of does the driving and builds the roads in real time. So that may or may not get you where you want to go, but these models are very new. They are very fresh. So essentially, you know, the 01 model from OpenAI and then the new Deep Seek model, R1 light preview,

they kind of do this chain of thought reasoning under the hood. So as long as you still give it some direct guidelines, it can actually do these reasoning models can do a fairly good job, especially in some of those kind of areas that we talked about. So math, problem solving, anything that requires logical inference. That's huge. So DeepSeek has not yet released the full code for third party. analysis or provided an API for independent testing. That's huge. You know, at least with OpenAI

when we saw the O1 preview in O1 Mini releases, you also got it via the API. So right now you do not have that with this new DeepSeek product. So the new R1 Lite preview is accessible right now for free on DeepSeek chat with a daily limit of 50 messages in its advanced deep think mode. So if you are using or if you want to see the capabilities of the R1 light preview, you will use the deep think mode in deep seek chat. So many buzzwords y'all. So deep seek's plans to release open source versions of its R1 series models and related APIs continuing its support for the open source AI community. So previous models like deep seek V 2.5. so that is not its, you know, reasoning model,

but its normal model have been praised for their capabilities and language processing and coding. Yeah, what Michael says. 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, 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. China's coming in hot here.

Yeah. Philip, Philip saying large language models don't have emotions, right? Yeah, you know, that's another. episode for another day, Philip, but I will see, I will say this, you'd be surprised at the answer without getting into it a little bit much, or without getting it into a little bit more. So another question here from our live stream audience. Thanks for all the questions this morning, guys. It's always nice. It's always nice when we can get some questions in, even on the AI news. So McDonald asking, what is MMLU? That is the massive multitasked

language understanding. It is essentially a benchmark. It is a ACT4 large language models. All right. Let's keep this thing going. More AI news so much this week, y'all. So open AI also updated its GPT40 model, like we talked about with its 1120 release. And they've quietly rolled out a new creative writing update. for chat gpt in this new model enhancing its ability to generate more natural and engaging content. So the writing improvement was part of open ai's 1120 release update and they updated gpt 4-0 with some benchmark improvements and like we said very briefly reclaimed the top spot on the AI chatbot leaderboard like we said that lasted only a day with their new 1120 release as Google released their 1121 Gemini release.

And also, y'all, we called this. We called this a week before it happened. We said within a day, open AI would have a new model to outperform Google. We didn't see Google clapping back so hard immediately, though. So in addition to these creative writing enhancements, chat GPT40, now offers better handling of uploaded files, offering deeper insights and more comprehensive responses. So the model's ability to analyze and work with spreadsheets and images has also reportedly been significantly approved, at least according to improved, according to OpenAI. So users have noted the model's advanced capabilities with one user on Twitter claiming that it rivals, yeah, I'll have to find this tweet, claiming that it rivals Eminem in crafting complex rap lyrics.

Yeah, someone online. I wish I did this because I did this probably about a year and a half ago when new models came. out, I would have it create essentially a rap in the tone of M&M, you know, describing large language models. And it seems like someone did something similar and then took those lyrics into Suno's new V4. And pretty, pretty impressive. I listened to it. So the important thing here with OpenAI, this update is automatic.

So you don't have to, you know, just be using the API. You don't have to just be using OpenAI's playground. If you are using the front end of ChatGPT, you will start seeing the, uh, this new 1120 release update. So it's not going to be called, you know, GPT40 1120. So at least what Open AI said is their new kind of model naming is 11 or sorry, it will just be GPT40 latest. But then if you look on the back end, you will see the date.

So it is automatically refresh when Open AI does release their models, unlike Google where you're not using it on the front end, right? if you go into the chat interface. So let's keep talking about OpenAI. I'm sorry, y'all. They made a lot of news this week. So because they're not just trying to compete with large language models. They're reportedly trying to compete in the browser, y'all.

What about that? You know, Google claps back by retaking the top spot on the arena board. And then right thereafter, we see reports that Open AI says, oh, we're actually going to be releasing a browser now. Well, they didn't actually say that, but that's what report said. So OpenAI is reportedly developing a new browser, web browser, that could challenge Google Chrome, and that's according to the information. So this initiative is obviously newsworthy as it signifies potential disruption in the browser market,

traditionally dominated by Google Chrome. So this project involves ex-Google developers, Ben Goodger and Darren Fisher, who were part of the original Chrome team and have reportedly now joined OpenAI. So the new browser is expected, according to reports, to include built-in support, obviously, for chat GPT and GPT search, enhancing user interaction with AI. So despite these developments, the information reports that a functional version of the browser is not expected anytime soon. Google Chrome currently holds the significant market share with about 65% on desktop and 68% on Google as per stat counters October data. Here's why it's also extremely newsworthy and interesting because the U.S. government

has proposed that Google divest its Chrome business. So yeah, the U.S. government essentially said, yo, Google, you're too big. You can't own the Chrome browser anymore. So it should be interesting how that one actually plays out, but it's pretty timely here. So Google may have to divest its holding in the Chrome browser because the U.S. government said that kind of constitutes an illegal monopoly, which could impact the browser landscape. So just in time for this, we see the news or the reporting from the information saying that OpenAI is working on a browser.

Also, OpenAI is in talks with Samsung to integrate a. features into smartphones similar to the partnership that Apple and Open AI have. So speaking of that exact partnership, what the heck is going on? We heard announcements months ago from Apple that we were soon going to be getting a much smarter Siri. Well, you might have to wait until 2026. Yeah, that's not a glitch. 2026. All right. So Apple is reportedly developing a significant AI upgrade for Siri, aiming to make it more conversational and competitive with chatGBT's advanced voice mode and

Google Gemini's Live, as reported by Bloomberg. So the new Siri assistant is tentatively named LLM Siri. Oh, gosh. Anyways, I can't, I can't stand the name. I can't stand the name. naming today or this week. I don't know why y'all. So the new assistant is tentatively named LLM Siri, yikes, and it will utilize Apple's AI models to enable more natural-sounding interactions. So although Apple might announce these plans next year, the complete transition to the new system is projected for spring of 2026. All right, so Apple plans to enhance series capabilities, allowing it to handle more complex tasks and interact with third-part. apps through an extended use of app intents.

So the overhaul will include features like summarizing and writing text using Apple intelligence. So Apple has been cautious with its AI rollout, to put it nicely, only recently integrating chat GPT into Siri as part of the iOS 18.12 beta. So according to reports, this new LLM Siri may not be debuting until iOS. OS 19. So yeah, just buckle up. It's going to be a while. And upcoming features for Siri include understanding on-screen content and performing actions with apps, which that piece is expected to launch next year. Let me just be honest here. I don't know how Apple missed the boat by this long. It's not like they were sprinting seeing the boat leaving shore. The boat has

made like eight round trips back and forth and Google or sorry and Apple has missed all of them. So I do understand and you have to applaud Apple's approach to artificial intelligence. And yes, I'm calling that artificial intelligence and not Apple intelligence. Can we please not try to claim intelligence? Can we please, right? And there are reports that, you know, Microsoft was going to call its co-pilot to Microsoft intelligence. I'm glad I didn't see that at Microsoft Ignite or hear any rumblings about this.

but Apple, you are so late to the party. Bringing this by 2026, I mean, that's like, you know, a company saying, oh, we're bringing a touchscreen phone in, in 2023. It's like, y'all, like we've had it for a decade. Like, we're snoozing by now. It's, it's, this is, I mean, you could say, oh, better late than never. I don't know, Apple, you're actually embarrassingly late here. But like I said, you do have to applaud Apple for its approach.

I do think they take security way more seriously than the other big tech companies. Not saying they're not taking it. The other tech companies aren't taking it seriously. But Apple is going a couple of steps further, but at what cost? Again, so late here. So late, right? We've only seen, you know, very simple text summarization come out via, you know, some 18.1 in the iOS or MacOS 15.1.

And y'all, this is stuff that we had pre-chat GBT. This is stuff that we had in 2020. So, you know, Apple's latest AI that they actually have publicly available for everyone to use is technology that is more than four years old. It is the least impressive piece of AI that I've seen from a trillion-dollar company probably ever. to call it underwhelming is giving them a lot of credit. Sorry, I don't want to be harsh, y'all, but that's just the reality of it. All right.

Next, yeah, 20, yeah. Everyone in the comments this morning is saying 2026 question mark. Yeah, that was not a, that was not a glitch there, y'all. Joe saying Apple needs to reevaluate Apple intelligence for branding because nothing I have seen in the past year has been anywhere near intelligent, Joe. I would agree with you there. all right. Let's talk about another big trillion dollar market cap company that's taking a different approach to AI. Amazon's throwing money at the problem. Don't hate the approach. So Amazon has invested another $4 billion into Anthropic. So Amazon has just announced another significant investment of

$4 billion into Anthropic, the creator of Claude AI, bringing the total funding to $8 billion. So yeah, Amazon has invested $8 billion into. to Anthropic, which I believe makes it the second largest partnership for a large language large language model maker with Microsoft reportedly investing around $14 billion into OpenAI over the past few years. So now you have Amazon upping its kind of chips in the pot to $8 billion into Claude's pile. So this investment is part of a strategic partnership where Amazon Web Services, AWS from Amazon, will become Anthropics primary training partner utilizing AWS.

CREAM and Inferentia chips for future AI models. Yeah, Amazon's got some new AI chips that Claude will be using for training. So this collaboration aligns with reports that Anthropics Claude AI will power Amazon's revamped Alexa Voice Assistant slated for release in 2025. Can we, will we actually get a smart AI assistant by then? Maybe. All right, actually more on that here a second. So the improved Alexa is expected to offer a more conversational tone akin to OpenAI's ChatGPT Advanced Voice Mode, although early beta tests indicated some performance issues. So Amazon's investments and partnership with Anthropic underscore its efforts to remain competitive in the AI sector, particularly against rivals like Microsoft and Open AI. So according to the Verge,

the improved Alexa reportedly performed better with Claude AI powering it. And so, according to the Verge, compared to Amazon's in-house models. That's not hard. Suggesting a potential leap in user experience for Alexa users. Yeah, you can't really ask Alexa anything except maybe the time and temperature. And even there, you're rolling the dice. Using traditional, you know, quote-unquote smart assistance now, it's like pulling my hair out. So I don't want to do it because I still have my hair and I don't want to lose it.

So that's why I don't use Amazon or Alexa or Siri. So despite the advanced. some beta users reportedly said that the new Alexa struggled with basic tasks, such as turning on lights, highlighting ongoing AI, highlighting ongoing AI challenges and AI development. And let me tell you something, y'all. So, I mean, we just saw it there. Siri, the large language model theory, not until 2026. The new Claude powered Alexa, probably not until 2025, which, you know, I know that's a couple weeks, but it could be toward the latter. part of the year. Advanced voice mode is great. It is by far the most capable model if you are just

looking at its abilities on the surface to respond in real time and be a natural kind of voice, right? And, you know, Microsoft is also tapping into that voice model as well. So the new GBT40 kind of voice capabilities. However, the OpenAI and Microsoft versions lack real-time access to the web. Guess who doesn't actually Google Gemini Live? I finally found a consumer-facing AI products from Google, aside from Notebook L.M, that I think is a leader in the space. So I did do a review of that last week in our newsletter, I think on our YouTube channel. That's why you got to read the newsletter, y'all. I was actually extremely impressed. So even though the new Google, Gemini Live is not quite as natural sounding. The latency between how you speak and when the

Gemini Live speaks is not as good as it voice advanced mode from ChatGBT. Unfortunately, advanced voice mode is not connected to the web. So you are working with old data. Whereas the new Gemini Live, which is now actually available for the iPhone, for iPhone users out there like myself, it is connected to the internet. So that piece is huge. All right, let's keep the AI news recap train going. So according to reports, OpenAI has quote unquote accidentally erased crucial evidence sought by the New York Times in a copyright lawsuit, causing significant setbacks in the case. So the reported deletion occurred while attorneys from the New York Times and the New York Daily News

were reviewing two large caches of training data files provided by Open AI earlier this month. So these files were reportedly intended to support OpenAI's defyons. that publicly published articles were fair game for training AI models. The accidental deletion, though, wiped out all search data in one of the caches, forcing the newspaper's attorneys to retrace their steps and lose a week's worth of work. Brutal. So although OpenAI did manage to recover most of the data, the folder structure and file names were permanently lost, hindering the attorney's ability to track where the copied

articles were reportedly used in building open AI's models. The attorneys have requested that Open AI take responsibility for redoing the search as they believe OpenAI is best positioned to navigate its own data sets. So despite acknowledging the erasure was likely unintentional, the deletion has delayed the newspaper's efforts to build their case. OpenAI has indicated plans to respond to the attorney's request with a spokesperson stating to Wired, we, quote, we disagree with the characterizations made and we'll file our response soon. All right. End quote.

Yeah. I've been saying this since this, yeah, this lawsuit has been dragging on now for nearly a year. It was reported in December of 2023. So here we are almost a year later, still talking about this lawsuit. And I said it then, and I will say it again, I don't see anything else happening here except a massive settlement. that OpenAI is going to have to pay the New York Times, even though, if I'm being honest, the New York Times at probably one of the worst arguments

in their opening briefs that I've ever seen. Granted, I'm not a lawyer. I did stay at a holiday Inexpress once, but I read through the New York Times entire case, and it seemed at least at the time, they did not understand how large language models work, which is not making a good case when you are trying to argue against a large language models.

language model maker. And that is why OpenAI has essentially responded to Open AI in the last year by just kind of scoffing. Because if I'm the lawyers at OpenAI and the Open AI leadership team, I saw what the New York Times initially put out and I'm laughing at it because it shows a complete and utter disregard for understanding the basics of a large language model. And I also, let me be transparent here. I'm a former journalist, right, but I also cover AI every single day. and chat chbt is my favorite at least you know kind of online AI chat bot so you know putting putting all my you know all my biases uh you know or potential biases in the ring there the new york times has completely completely just dropped the bag on this case uh but a nice little new little wrinkle here

with this news all right a couple more pieces of AI news open AI has opened up its advanced voice mode to web users. So chatGBT's advanced voice mode, which I just referenced, is now available to paid users on the web and in the chat GBT app. So the expansion is currently available for chat GPT's paying customers, including Plus Enterprise Teams and EDU subscribers, enhancing the chatbot usability with more natural real-time conversations. So the feature was, the advanced voice mode feature was initially launched in iOS. and Android apps in September. But if you wanted to use the advanced voice mode,

you could only do it on your phone. So now, as of a couple of days ago, it is available in the chat GPT desktop app as well as online. If you're not seeing it online, here's what I suggest you do. Yes, I'm going to be your ITT guy here for a second. Log out of your browser,

clear all your browser, cookies, cash, et cetera. Then try logging in. If that doesn't work, open it in an incognito window or private window, and it should work. It worked for me. I have it on access to it on in the web and on the desktop version, but you will need to update the desktop version to get access. All right, moving on. U.S. President-elect Donald Trump has announced his intention to repeal President Joe Biden's executive order on artificial intelligence,

highlighting a potential shift in U.S. policy as Trump prepares to take office in January 2025. So this announcement is very timely as it coincides with a global AI safety effort as these talks were held in San Francisco this past week, where world leaders and tech industry giants gathered to discuss AI's transformative potential and the risk involved. So leaders from countries worldwide, including Canada, Kenya, the UK and the European Union met to establish international safety protocols for AI, underscoring the importance of global cooperation in addressing AI challenges. So this kind of global AI safety event was organized by the international network of AI safety institutes and hosted by the Biden administration, reflecting a commitment to integrating

safety and ethical considerations into AI development. So these discussions are part of broader efforts to ensure that AI advancements do not compromise public safety or privacy with significant backing from major companies like Amazon, Google, and Microsoft for codifying AI measures into law. So the differing approaches of Trump and Biden to AI policy could significantly influence the direction of AI innovation and regulation in the U.S. impacting companies and individuals involved in the field. So we did cover this a little bit last week as well, talking about how Trump's proposed tariffs could actually affect some of the biggest, not just tech companies, but everyone with those tariffs could be impacting prices. All right.

And our last piece of AI news for the day, y'all, I don't know why. So many people are sleeping on what Microsoft announced, which I don't understand because guess what? There's a good chance if you are listening to this right now for work, you use a Microsoft PC, right? And guess what? Microsoft just changed how PCs work. All right. So Microsoft announced more than 90 new AI powered features for co-pilot at the Ignite Conference in Chicago. marking a significant advancement in AI integration into workplace tools.

So first have to give a quick shout out to both We Communications and Microsoft for hosting myself and the Everyday AI show at Microsoft Ignite last week. Really cool birthday present for me being able to technically beat Microsoft with their own announcement. So if you tuned in on Tuesday, you were actually the first audience in the world that got access to everything that Microsoft was announcing before they even started the keynote. But here's a quick rundown of what's new with AI and with co-pilot with Windows announcements. So the biggest thing, new co-pilot studio autonomous agents that are able to automate routine

tasks, reducing manual intervention and increasing efficiency. SharePoint. Now you have one-click SharePoint agents that allow employees to gain insights. just from specific SharePoint content providing tailored solutions. Also, FYI, so many of these kind of new features that I'm talking about do require that your organization has or gives all its employees access. So, you know, this isn't, you know, the default. If you have, you know, Microsoft Office 365, right, you have to have co-pilot. So you have to have Microsoft 365 copilot licenses, which are $30 a month enabled for all.

all of your users to access all of these new features. So which is funny, so few enterprise organizations are actually doing this, which is mind-boggling, but that's another podcast for another day. All right, let's get back to what's new. So the SharePoint agents, I think, are great. That is literally bringing a state-of-the-art rag model with zero technical capabilities. So, you know, I kind of talk about Notebook El-I'm a lot on the show because I think it's freaking amazing.

I think this is kind of what SharePoint agents could do, right? being able to, first of all, all SharePoint sites now have an agent built in by default, which I think is really cool. But then you can also modify just certain files or just certain folders in its SharePoint site and create an agent like that with no coding and no really technical capabilities. I was actually talking with the SharePoint team, the people that were actually building it. They're showing me some of the behind the scene, some of the stuff that's coming and is pretty interesting.

I think you're going to want to start using it. So here's what else is new. Microsoft 365 copilot actions. These introduce pre-built shortcuts meant to simplify task delegation. So essentially what that is, think of it like a combination between an advanced prompt maker and something like Zapier or, you know, programmable AI. So essentially, you know, with the new Microsoft 365 copilot actions, it gives you a list of drop downs where you can select team members.

You can select different files. And then you can also schedule something. So let's say you get the same email every Monday that asks you to recap all these files from your OneDrive, from your SharePoint, from your Teams meetings, et cetera. And maybe that's like 70% of what you do for that week. Well, you can now literally with co-pilot agents automate all that or with Microsoft 365 copilot action. So it is a way to schedule ongoing AI co-pilot power. automations that are triggered by a certain event that you set. So it can be, you know, you can always just run it weekly.

You can have it set by any trigger as well anytime something happens in your organization. Co-pilot agents, yeah, more that's new here. We're not going to go over all 90, but new copilot agents in teams offer real-time, note-taking, interpretation in multiple languages, and enhance meeting experiences. Yeah, you can literally, again, if your team has access to Microsoft 365 copilot, you will soon be able to speak in another language in real time, which still blows my mind. So, co-pilot in PowerPoint, yeah, let's keep it going. It has introduced a new addition to its narrative builder for creating presentations from a single source document

with also the ability to instantly translate that PowerPoint, which you can build on your brand presets into 40 languages. Co-pilot pages now supports rich artifacts. like code, charts, and diagrams providing a collaborative document experience. And we're actually going to have a dedicated show on that where we talked with the person, one of the people in charge of co-pilot pages. That's going to be dropping here soon. Also, we saw some hardware powered announcements from Microsoft with the Windows 365 link.

So that is a new hardware device that connects to a cloud PC offering a cost-effective solution for keeping systems up to date. So yeah, it is a $349. Think of it like kind of like a PC, but it can't run on its own. It is a cloud PC, but a physical device for $349. I think this is going to be great for large enterprises to essentially cut down on the kind of your ongoing costs, right? So essentially you have to have a subscription. So you pay to give it power monthly.

But instead of, you know, especially if you're doing a lot of AI power tasks, instead of giving a thousand people, you know, a PC that costs $2,000. Instead, you can buy this new Microsoft 365 link for, I believe, $349, and then you pay a monthly fee depending on how much compute or how much power or how much memory each user needs it. So it's a very easy way, I think, and ingenious to quickly scale. But like I said, it's not a standalone device. It won't work without that. So think of it as kind of a cloud PC. also Microsoft release or announced cloud pilot analytics, which provides tools to measure AI's

business impact. So everyone's always talking about what's the ROI on AI? Well, Microsoft is trying to answer those questions with co-pilot analytics. And then we also saw, I dug in a little more. I found all the right people at the Microsoft Ignite conference and talked to. And I did find out the recall feature. Well, it actually just got rolled out Friday. I did find that out Friday or might have been Thursday, actually.

So the recall feature, which has gone, it's kind of been kind of highly criticized and people are skeptical of it. For me, I can't wait to use it. But think of it like this. If you do have a co-pilot plus PC, so it doesn't work on just any co-pilot plus PC and you do need the Microsoft 365 copilot subscription to use this. It essentially takes screenshots of everything that you do on your PC. It is opt in. Don't worry, privacy advocates. You do have to opt into the service.

and you can delete anything from the memory, but then essentially there, you can talk to co-pilot in real time. You're like, oh, man, what was that thing I saw on my computer three weeks ago? Was it in a team's meeting? Was it in a PowerPoint? Was it on edge? What was it? Well, you can just ask, what was that picture or what was that announcement about the new product,

the new coffee maker, right? And it'll bring it up, right? You don't even have to know where it was from. So pretty cool. So that is it, y'all. I do think the Microsoft Ignite announcements were pretty huge. I don't know why it wasn't getting talked a lot about maybe because it was in Chicago and maybe, I don't know, maybe only the tech conferences get the shine.

You know, when they go on the West Coast or they go in New York or something like that. But, I mean, just monumental announcements from Microsoft that I think are going to change how we all work. Because the overwhelming majority of enterprise companies out there are using Microsoft. Microsoft PCs. Microsoft is literally changing how PCs fundamentally work with Microsoft 365 copilot. All right. Real, real quick, let's go ahead and quickly recap our AI news that matters. So like we talked about, Google Gemini released a new model, Gemini 1121, that topped the rankings.

also they'll reportedly be releasing a Gemini 2 soon, as well as they did already announce memory. Deep Seek announced a R1 light preview model, which is a reasoning model, a competitor to OpenAIs, 01. Speaking of OpenAI, a lot of news from them, they've released a new kind of creative writing update under the hood for the new GPT40-1120 release. Open AI is also reportedly venturing into web browsers to compete with Google on an even grander scale. According to reports, Apple is set to reveal a LLM Siri, but not until 2026. Amazon has invested another $4 billion into Anthropic, uping its total investment into the AI company to $8 billion. $1. Open AI reportedly deleted some data in their New York Times copyright lawsuit. Open AI also has expanded its advanced voice mode to web users and to the desktop.

Trump's AI policy repeal has kind of caused some AI safety concerns and world leaders were meeting to discuss it. And Microsoft unveiled 90 plus new AI features at it. its Ignite conference. All right, that was Aton, y'all. I hope this was helpful. If you don't have time every single day to keep up with what's happening in the world of AI, Mondays, Mondays, Mondays, be there, be there, be there.

The AI news that matters, this is your time to get caught up. But if you want to keep up every day, if you want to be able to talk in real time and ask questions to some of the leading AI experts in the world, we do this every single weekday, actually Monday through Friday. So please make sure if you haven't already. go to your everyday AI.com. Sign up for our free daily newsletter. And we'll see you back for more AI.

Every day. 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.