well it's been a heck of a week in the world of AI and there have been a ton of announcements so many that I've actually made multiple videos about some of the announcements leading up to today just so this video wouldn't be an hour long one of the biggest stories of the week was that open AI finally rolled out their advanced voice mode to pretty much everybody that has a plus or teams account for chat GPT this is one of the things that I made a full demo video of testing various things with the AI voice assistant and we learned that if you didn't seem to get access to the advanced voice assistant on your phone The Simple Solution was to remove the app completely on your phone go back into the App Store redownload it and set it up once again and that seemed to work for most people it worked for me and it seemed like it worked for most people in the comments as well once you have access to it you simply press that little circle with the little lines on it hey Matt what's up how can I help you today and it looks like it'll start working hey I'm just uh doing a quick live demo of the advanced voice feature nice advanced voice features can add a whole new dimension to your content how's the demo going so far so far so good awesome to hear demos can be a bit unpredictable so it's great when things run smoothly since this has been rolled out and since I made my demo video walking through it there has been a few like interesting things pop up around this for instance this guy here managed to actually get it to do a duet with him for the the song Elanor rig by The Beatles [Music] El the window face that sheaves in AAR by the door it my Guin All the Lonely People super interesting you could hear for a second it started to say my guidelines don't let me do that but then it continued to start of sing along anyway we also learned that there are some rate limits I'm actually not 100% sure where these rate limits lie to here on X shared this screenshot saying that you have 12 minutes remaining of advanced voice mode if you reach the limit you can switch to standard voice quickly looking at the FAQ here over on open ai's website we can see for how long can I have voice chats your daily use of advanced voice for plus and team users is subject to a limit each day and daily limits may change we provided notice as you are approaching the daily limit plus and team users will be notified when they have 15 minutes left of advanced voice for the day so apparently it's a bit of a moving Target and they're not telling us what the limit is it's kind of a changing limit every day I I don't totally know how this works yet again I did a much deeper dive on this new advanced voice feature the day it comes out it looks like this ear but we did get some other news out of open AI this week we learned that open AI is going to remove moove the nonprofit control and give Sam Alman equity in the company according to this report open AI is working on a plan to restructure its Core Business into a for-profit benefit corporation that will no longer be controlled by its nonprofit board they're trying to make the company more attractive to investors the nonprofit will continue to exist but own a minority stake in the for-profit company the rumor is that Sam Alman himself will receive around 7% of this new four entity that's created and it's expected to be valued at about $150 billion that would make Sam ultman stake in it roughly $10. 5 billion now this is still just sort of rumors there hasn't been any sort of confirmation that I've seen that's come out of open AI yet but there has been some other interesting things that happened at open AI this week for instance the CTO Mera moradi one of the people that really stood up for Sam Alman back when he was sort of booted from the company almost a year ago now has decided to step away from the company she says after much reflection I have made the difficult decision to leave open aai my 6 and 1/2 years with the open aai Team have been an extraordinary privilege while I'll express my gratitude to many individuals in the coming days I want to start by thanking Sam and Greg for their trust in me to lead the technical organization and for their support throughout the years everything seems to be amicable mea is not making any statements about leaving for safety reasons or anything like that it sounds like she just wants to move on to something else you'll see on X all sorts of like conspiracy theories and people trying to figure out explanations of why she's leaving but quite honestly in some of these cases it might just be that they put in a ton of work over the last six plus years they got thrusted into the spotlight because open AI became such a massive company and chat gbt was such a big success and they just don't want to be in the spotlight anymore smokea away here points out on X Ilia announced his departure the day after the GPT for presentation Mira announced her departure the day after the chat GPT voice release I do think the timing was a little bit strategic miror probably knew for a little bit that she'd be leaving but didn't want to make waves before a big announcement so waited till after the big announcement and I'm sure it was the same kind of thing with Ilia personally I'm not buying into a lot of the conspiracy theories around open Ai and why all these people are leaving again I just think the company grew really really big really fast a lot of these people were thrust into the spotlight really quickly Mir has been on all sorts of TV shows and has essentially become famous because of this and that could quite honestly burn anybody out pretty quickly and I think that probably has a lot to do with it it doesn't seem like there's bad blood or she's scared of what open AI is creating or anything like that that's not what I'm getting out of this a lot of people on X and other YouTube videos will probably try to convince you that that is the case but I'm not seeing that miror wasn't the only one who left open AI this week their Chief research officer also left right after Mira left open ai's Chief research officer Bob McGrew and research VP Barrett Z left the company on Wednesday hours after open AI CTO mea moradi announced she would be departing Mera Bob and Barrett made the decisions independently of each other and amicably he said but the timing of Mira's decision was such that it made sense to now do this all at once so that we can work together for a smooth Handover to the next generation of leadership the way they're sort of angling this as well mirror was leaving so there was going to be a little bit of a shakeup in the leadership anyway might as well leave at the same time so they can sort of figure out the reorganization all at once instead of one person leaving reorganize another person leave all right let's reorganize again another person leaves now let's reorganize again they sort of seem to have coordinated it to kind of make things easier on open AI I don't know that's sort of the way they're angling it at least this week Sam Alman made a rare personal blog post over on his Sam alan. com website called the intelligence age he says in the next couple of decades we'll be able to do things that would have seemed like magic to our grandparents the phenomenon is not new but will be newly accelerated people have become dramatically more capable over time we can already accomplish things now that our predecessors would have believed to be impossible he goes on to talk about where he believes all of this is headed eventually we can each have a personal AI team full of virtual experts in different areas working together to create almost anything we can imagine our children will have viral tutors who can provide personalized instructions in any subject in any language and at whatever Pace they need we can imagine similar ideas for better healthc care the ability to create any kind of software someone can imagine and much more this is the part that I think is probably the most interesting and well interestingly worded he says it is possible that we will have super intelligence in a few thousand days wording it like that is interesting cuz it makes it sound like it's fairly close but a few thousand days could be you know anywhere from 3 years from now to like two decades from now it's an interesting read and I highly recommend it if you want to understand where Sam Alman the CEO and the person running open aai believes all of this is headed right now and finally in the last bit of open AI news for this week Johnny Ivy confirms that he's working on a new device with open AI Johnny Ivy if you're not familiar is a famous designer who worked at Apple who helped design some of Apple's most iconic products like the iPod the iPhone the iPad the iatch most of those were designed by Johnny we don't know much about what this new device that he's teaming up with open AI on are all we know is that he has given some sort of confirmation that there is something in the works so that's something to look forward to hopefully it's not another like rabbit R1 or Humane pin where it seems kind of cool in theory but in practice it's just not something that most people are interested in using but given his reputation with the products that he helped design at Apple I think he's going to Faire a little bit better than some of those products moving on to the next big massive Monumental thing that happened in the AI world this week was meta connect 2024 I was actually at this event and once again I made an entire video breakdown of all of the announcements they made at the event some of my thoughts around it and a little behind the scenes of my experience at the event therefore I'm not going to go too deep into all of the announcements they made because I have a whole breakdown video that looks like this right here that you could watch and see but here's the quick rapid fire overview if you just want the tldw they introduced the new meta Quest 3s which is very similar to The Meta Quest 3 but it is less expensive it's going to be starting at just $299 it's going to come with a new Batman game that's super fun I've played it myself I already have a meta Quest 3 but I will be buying that Batman game because the demo hooked me I want to play more of it they also announced a ton of new AI features and functionalities that are going to be rolling into Facebook Messenger Instagram messenger WhatsApp and you know all of the meta Suite of tools they announced a new meta voice feature I actually think that maybe open AI knew that meta was going to announce this voice feature and wanted to sort of front run their advanced voice announcement because that came out the day before this voice feature interestingly enough if you want to talk with meta's AI they actually have some celebrity voices that they got permission to actually use voices like Aquafina Dame Judy Dench John Cena Keegan Michael Key and Kristen Bell I find the Kristen Bell one absolutely fascinating because Kristen Bell actually spoke out quite a bit about AI at one point she actually made an Instagram post that said she opposed meta's AI to use her data but now she's one of the chatbots official voices you can see on Instagram she put this whole thing I own the copyright to all images and posts submitted to my Instagram profile and therefore do not consent to meta or other companies using them to train generative AI platforms this includes all future and past post stories threads on my profile one thing to note is by just putting this message on your Instagram it actually does not exclude you from anything that was in the terms and conditions that you agreed upon when you signed up like meta is not watching your posts to see if you actually consented or not I'm actually kind of a fan of Kristen Bell my wife and I used to watch Veronica Mars and I think the good place is like one of my favorite TV shows ever but this whole 180 that she pulled is kind of fascinating being sort of anti- AI at meta and then sort of flipping the script and becoming one of the voices and it makes sense to be honest they don't want to consent on using their likeness and their content without compensation and I'm sure this new deal with meta got her compensated quite well if I had to guess the new meta AI is now multimodal so you can actually upload images and it can understand what what's happening in the images you can even edit your images with text we can see in this example they uploaded an image of a cake and asked how to make the cake and it actually gave them instructions and a recipe on how to make the cake here's an example where they uploaded an image of a goat and then gave it a prompt to add a hat that says goat and we can see it put a hat on the goat with the word goat on it put them on a surfboard and it put the goat on a surfboard so this new multimodal functionality is going to let you have a little bit more fun with your images that you throw into one of their messaging platforms one of the more useful features I think they are rolling out is the AI translation and lip syncing I can make an Instagram real completely in English upload it and then have it translate it to Spanish and Japanese and whatever languages I want and it will recreate that same real with me speaking in the proper language properly translated and if I'm on camera speaking it'll actually sync up my lips so it looks like I'm speaking in that language that seems really useful to get a lot more reach on your Instagram reels and things like that they also showed off new Creator AI feature where you can create a sort of virtual version of yourself that's trained on your Instagram and threads and Facebook content so that it can speak like you and answer questions in the way that you would likely answer them and in this really cool demo a buddy of mine was actually the example they showed off this is Mark Zuckerberg talking to the AI version of Don Allen Stevenson here congrats on the new book that you just released you know what's the main thing that you're hoping that people take away from it thank you so much yeah the main thing I want people to take away from my book is the idea that you have the power to create your own opportunities by combining curiosity adaptability and resilience in a rapidly evolving Digital World they also rolled out a new version of llama 3.
2 the open source large language model and it is now multimodal and it is also available to use and play around with for free right now on hugging face you can upload images here type in text pretty much using in the same way they were demoing it inside of the various meta platforms they showed off some new features for the Rayban metag glasses glasses that I actually use as my sort of daily Wares I actually love mine they added a bunch of quality of life features to it like you can start talking to it by saying hey meta but you don't need to say hey meta every time you want to prompt it or ask another question after that first hey meta when it starts kicking off the AI conversation you just kind of keep talking to it normally after that you can also tell your glasses to play music for you from places like Spotify or apple music or audiobooks from places like Audible and it will just start playing them in the sort of built-in headphones on the glasses but in my opinion the most useful features that they're adding in are memory so you can say hey meta remind me in 10 minutes to do this and your sunglasses will remind you in 10 minutes to do the thing or the example they showed is they said hey meta remember where I parked right and it looked at the parking spot where they were took a picture of the number on the parking spot so that later on when they were looking for their car later it would remind them what parking spot they were in it's also going to have live translations so somebody could speak to me in Spanish if I'm wearing those glasses it can translate that to me directly into my ears in English it could scan QR codes and automatically open whatever it scanned on your mobile phone all you have to do is look at it with your glasses and it'll scan the code a lot of really cool features they also rolled out a new clear version where you can see all the electronics inside of the sunglasses I actually got my hands on a pair the pretty cool looking and honestly I think these metag glasses are only going to get more popular with all these new features they just rolled in the biggest announcement at the event was their Orion Project which is their augmented reality glasses that just kind of look like normal glasses I mean they're still a little bit big and bulkier than normal glasses like The Meta rayb bands but they're a heck of a lot smaller than like a meta Quest or an apple Vision Pro they look like glasses and they work almost like an apple Vision Pro where you can use hand gestures and put up videos in of you and move things around and play games in sort of augmented reality in front of you they're pretty dang mind-blowing everybody was pretty blown away when they were showing these off at the demo during meta connect you can see some examples of having phone calls in the glasses while having a browser open and a messenger open on the side it can look at ingredients on a table and give you a recipe straight into your eyes based on the ingredients that are there on the table a super super exciting project and once again I did an entire like 20 minute breakdown of everything they talked about at medic connect you can see that video here it looks like this and that's a super deep dive into all of the new announcements but again because I don't want this video to be an hour long I'm going to go ahead and keep moving on with the rest of the news from this week with something that I think shocked almost everybody the fact that James Cameron you know the guy behind Terminator and Terminator 2 and Avatar and Titanic and some of the biggest movies ever made as s on as a board member of stability AI he is such a huge figure in the film making world that it just seems a little bit shocking that he is joining forces with an AI company as you know most of Hollywood is kind of actively trying to fight against AI right now hopefully big names like his can sort of legitimize the use of AI and this emerging technology in Hollywood and I'm super excited to see what kind of Technology comes out of the team up of James Cameron and stability AI because James Cameron essentially invented new technology to make a lot of the movies he made like Avatar they had to create whole new cameras and new systems just to make those movies and with him pulling AI into the mix I can only imagine we're going to see some really really crazy sort of film making capabilities become more and more accessible to normal people that don't have the kinds of budgets that someone like James Cameron might have but while we're talking about Hollywood I got a to mention this real quick we've got only a couple days left for Gavin Nome to make the decision on whether he wants to pass or veto the sb147 bill this is the bill that will put model makers responsible for any catastrophic harms that are done with the models even if the model maker wasn't specifically involved with that catastrophic harm it seems like Hollywood is speaking up and telling him not to veto it you've got to pass this bill he recently passed a whole bunch of bills that really really help out Hollywood bills that help protect actors from their voices and likenesses being used in films without their consent but now they're getting behind a bill that doesn't really involve them too much it really kind of puts Gavin Nome in a tough place because the two most powerful Industries in California are the tech industry up in San Francisco and the film industry down in LA and right now those two industries are sort of at odds with each other the tech industry does not want sb147 to pass the film industry does want sb147 to pass both of these industries have huge lobbying power in the government and it puts someone like Nome in a tough spot like which industry do I piss off and which industry do I work closer with yeah that's I I don't want to be the one making that decision right now we got some new updates out of Google this week as well there's updated Gemini models reduced 1. 5 Pro pricing increased rate limits and a bunch more updates to the Gemini Suite of models here we can see the price reduction of using Gemini 1. 5 Pro this is specifically for API so if you're a developer you're going to be able to use Gemini for cheaper than you used to be able to but in my opinion the coolest thing to come out of Google this week is that they made some updates to their notebook LM platform if you're not familiar with notebook LM it's a platform where you can throw a bunch of documents or text files into a sort of folder and then it will help you summarize those you can chat with them it will even create audio podcasts explaining what's going on in those documents well now they just added new features where you can even add audio and YouTube videos into your folder and have it help you summarize those and create podcasts around those as well so for example I could come into notebook LM here create a new notebook and we have the option to add a YouTube link down here we can also upload PDFs text markdown audio like mp3s and have it actually use that as the context for discussion for summarization for podcasts so if I was to go grab the link for my latest meta connect video here plug it in as a YouTube link click insert you can see it quickly gave me a summary the YouTube video by Matt wolf titled metac connect blew my mind here's everything they shared is a summary of The Meta connect conference etc etc but I can generate a deep dive conversation here and in just a moment it will give me an audio version but I can also create a study guide based on the video I could create a timeline based on the video create an FAQ based on the video like here's a meta connect event timeline that it just generated for me meta Quest 3 meta AI updates llama 3.
2 AI voice mode AI clone feature AI translation pretty much everything I just got done talking about a moment ago cast of characters Mark Zuckerberg Don Allen Stevenson Roberto Nixon Cleo Abram Kane Callaway Rowan Chung Riley Brown lonus ekenstam Daniel Mack all these people that I mentioned that I actually got to meet at this event it actually put a cast of characters together of all the people that I mentioned here's the FAQ what is meta Quest 3s and how is it different from meta Quest 3 and it answers that question what AI advancements were announced for meta's chat apps Etc it created a whole FAQ based on my video it created a study guide and now it created an audio podcast based on my video all right so we just got done diving into Matt wp's breakdown of meta connect 2024 and I got to say this wasn't your typical like oh here's the new phone here's the new whatever it's like meta looked at the tech landscape and decided you know what we're going all in on AI on everything it's kind of crazy cuz it sounds fairly natural like it doesn't sound like an automated AI voice that you're going to quickly tune out it sounds like kind of a real discussion between two people talking about my YouTube video kind of interesting this to me is one of the most useful things Google has done with the AI technology that they have it's more useful to me than than using the Gemini Advanced chat I'd much rather put in information that I really want to Deep dive on and then chat with just that information and get a podcast about just that information this is really cool Steven Johnson here who actually works over at Google suggested this way for students to actually use the technology here he says one record audio from class on your phone two keep your laptop closed just jot down some short phrases to describe the most important points upload an audio and a PDF scan of notes to notebook LM ask notebook to expand your notes with details from the recording so you can take handwritten notes scan them throw them into notebook LM and it will use that for context as well bonus at the end of the week create an audio overview from all of your class summaries to review the most important Concepts in podcast format and once you've got your audio overview you can even change the playback speed listen at 2x speed you can download it and send it to other people it's just really really useful I'm probably going to make a whole separate video just talking about notebook LM my only fear is that people will start getting podcast versions of my YouTube videos instead of actually watching my YouTube videos which sort of disincentivizes me to make the YouTube videos I think it's really cool but I'm sort of conflicted about the implications of it if I'm being totally honest all right that was the majority of like the major news that happened but there's a handful of other things that I want to share with you that I thought were interesting so here's kind of a rapid fire of some of the other stuff that was talked about this week since we were talking about Google Snapchat is actually going to use Google Gemini to power its chatbot and a generative AI features snap entered into an expanded partnership with Google Cloud to power generative AI experiences within Snapchat's my AI chatbot it's going to leverage the multimodal capabilities of Google's Gemini AI to enable the chatbot to understand different types of information like text audio images and videos they also recently added Google Lens like features well it turns out that that technology is also being powered by by well Google Microsoft claims that it has a new AI safety tool that can pretty much eliminate hallucinations essentially when information is given back from a response with a chatbot it will actually kind of double check and make sure that there's actually a source for that information the new feature is called correction and it gives their AI systems the capability to automatically detect and rewrite incorrect content and AI outputs it's currently available in preview as part of the Azure AI Studio AMD the chip company that's a competitor at Nvidia just rolled out their first small language model called AMD 135m there's not a whole lot of information here about what this model is actually designed for due to the size of the model I'm guessing it's used for on device inference with your AI maybe for mobile phones I'm not sure they don't really go into details about what the sort of main use case of this model is if you use sunno to make your AI generated music They just added a new cropping feature for pro and Premier users so you can adjust the start and end of the song I'll link up this tweet from sunno in the descriptions so if you want this little five-step tutorial you can find it in the link below cloudflare is rolling out a new AI audit tool to help content creators block Bots if they want if you're not familiar with Cloud flare it's sort of a tool that lives between like your domain name and your hosting company if you run a website so when somebody goes to the domain name the data kind of comes from the hosting company routes through cloudflare and then shows up on your browser at the domain name that you plugged in well if you're a user of cloud flare they're going to give you some features that are going to allow you to block AI scraping if some of the big companies are out there trying to scrape your website dualingo the company that helps you learn other languages is launching an AI powered adventure miname and a video call feature so imagine you're learning a new language you're trying to learn Japanese on dualingo and you want to practice having a conversation with somebody you can actually have a conversation with an AI bot with the video call with Lily here it's designed to simulate natural dialogue and provide personalized interactive practice environment there's also this Adventures feature it's basically a game where you walk around like a simulated environment and you interact with characters in this game and you interact with them in the language that you're trying to learn so it's designed to kind of simulate a more immersive environment so imagine you're play playing a game that's like a Zelda like top down or like a stardew valley type game where you're going around and having conversations with different characters in this world but you're doing it in the language that you're trying to learn just trying to make that learning experience a lot more fun quite honestly I can see my kids absolutely loving something like this they're both trying to learn Spanish I might put this in their hands and let them play it a little bit and see how it goes this week the FTC announced that they're cracking down on deceptive AI claims and schemes this article is specifically singling out a handful of companies like do not pay a company claiming to sell AI lawyer services and Ascend Ecom Empire Holdings writer FBA machines basically multiple companies claiming they could use AI to help consumers make money through online storefronts they feel like a lot of these companies are misleading with their claims do not pay claims to be the world's first robot lawyer but the product failed to live up to its lofty claims that the service could substitute for the expertise of an actual human lawyer the site also claims that it offered a service that would check a small Business website for hundreds of federal and state law violations based solely on the consumer's email address it would detect legal violations that if unaddressed would potentially cost a small business $125,000 in legal fees but according to the complaint this service was also not very effective there's a whole bunch of other cases like this Ascend Ecom e-commerce Empire Builders this writer product FBA machine all of these tools claim that AI can help you build a the company that will make you money and none of them really come through on their promises and the ftc's saying eh no more of that and finally this is pretty cool Google Deep Mind has a program called Alpha chip that is transforming computer chip design it's basically an AI model designed to help create new chips that are better at training AI models so it creates this like Loop of AI is helping design a better chip that's going to make AI better and smarter these new chips are going to be used for AI that can then be used to make chips that are even better and smarter and faster and more efficient and it's just going to create this Loop of chips getting better and better and faster and smarter and more efficient and more cost effective Etc ideally creating this exponential curve of compute capability to train smarter and smarter and smarter models and that's called Alpha chip it's out of Google Deep Mind everything I mentioned in today's video will be linked up in the description below all of the tweets all of the Articles it should all be down there really cool stuff really exciting week I am absolutely exhausted I just spent the last month on the road you've probably noticed a lot of my videos aren't in my home studio I'm finally back home again I did vid Summit I was at Disneyland with the family I went and spoke at HubSpot inbound with Nathan L I was just at meta connect I'm exhausted there was a ton of stuff happening this month but I am excited to finally be back back in the studio back into a routine ramping up my video production going to try to get back in that habit of making three plus videos a week talking about all the coolest AI news sharing some cool tutorials sharing my favorite tools and how I use them so much exciting fun stuff that I'm going to be putting out on this channel if you like that kind of stuff give this video a thumbs up and maybe consider subscribing to this channel it will really help me out it will also make sure you see more stuff like this inside of your YouTube feed one last thing before I wrap this up I'm going to be helping judge a hackathon coming up in LA on October 12th and 13th in Santa Monica along with some other amazing judges and this hackathon is really cool because it's an AI hackathon whether you have developer experience or no developer experience but you just use AI to help you code you can participate in this hackathon if that's something that interests you and you're going to be in the LA area in mid October make sure you apply for the hackathon you can find it over at hack. cerebral Beach it should be a really good time and finally if you haven't already make sure you check out Future tools where I curate all the coolest AI tools that I come across keep the AI news page up to date on pretty much a daily basis a little bit slower when I've been traveling but it's up to date now and I have a free AI newsletter that will deliver the coolest tools and the most important news directly to your email inbox you can find it all for free over at future tools.