The AI News You Most Likely Missed This Week!

foreign[Music]didn't see you there sorry that wasreally dumb there's actually quite a bitto talk about this week in the world ofAI but there was one big event that kindof overshadowed all of it so if youweren't paying attention you probablymissed most of this stuff but we mightas well start by addressing the Apple inthe room this week was what the Nerdscall dub dub or WWDC the worldwidedeveloper conference from Apple now Ialready made an entire video about whathappened at WWDC so I'm not going to goto Too Much depth about that in thisvideo that video looks like this and youshould definitely watch it especially ifyou want details about all of thevarious AI things that they talked aboutas well as details about the New VisionPro which is rolling out next year butlong story short it's a very coollooking mixed reality headset but it's 3500 we won't get access to it till nextyearand it only has two hours of batterylifeforeignso if you were to watch the WWDC eventon an Apple Vision Pro you would haveran out of battery before the event waseven over shortly after the event metareleased an internal memo about theirthoughts on Apple Vision Pro some of thehighlights of what Zuck said about itwas the good news is that there's nokind of magical solutions that they haveto any of the constraints on laws andphysics that our teams haven't alreadyexplored and thought of they went with ahigher resolution display and betweenthat and all the technology they put inthere it costs seven times more and nowrequires so much energy that you need abattery and a wire attached to it to useit he went on to say we innovate to makesure that our products are as accessibleand affordable to everyone as possibleand that is a core part of what we do hetalked about how his vision of themetaverse is fundamentally social and bycontrast every demo that they showed wasa person sitting on a couch bythemselves he claims that they have areal philosophical difference in termsof how they're approaching thisbasically saying that he is optimisticof about the metaquest future afterseeing what apple is putting out also atthe Apple event on Monday there was abunch of announcements about additionsbeing added to iOS 17. many of them wereAI editions scratch that I mean many ofthem were machine learning additionsbecause Apple didn't say the words AIonce during the entire presentation theyacted like it was a dirty word thatshould never be uttered one of themachine learning add-ons was thatvoicemails would be transcribed in realtime so you can read who is actuallyleaving you a message and if you want toanswer it you can this is a cool newfeature to iPhone something that Androidhas already had but being an iPhone usermyself this is something cool to seethey made it so you can turn your ownimages into stickers and use them asreactions to text messages you get theyalso added automatic transcriptions toaudio messages in text messages which Ithink is a great addition because Ireally really dislike getting audiomessages they're also using AI slashlarge language models to improveautocorrect on the typing fee feature sobasically the more you use the keyboardthe more it learns the way you type andgets better and better at autocorrecting what you're trying to saywhich has led to this funny memecirculating Twitter so what did you dowith generative AI Microsoft we built aco-pilot for the web Google we createdBard Apple we fixed autocorrect they'realso using generative AI to add extraslow motion scenes inside of backgroundson the iPad they also improved PDFs onthe iPad where you could take a pictureand it will automatically find formfields that you can fill out and makethem fillable iPad is the best devicefor PDFs thanks to new machine learningmodels iPad OS can identify the fieldsin a PDF so you can use autofill toquickly fill them out with informationlike names addresses and emails fromyour contacts this even works ondocuments you scan with the camera goinginto this event a lot of people arespeculating that there might be somesort of update to Siri because Siri isone of the first AIS that a lot ofpeople used and when they finally got tomentioning Siri here was the bigannouncement it instead of saying HeySiri Siri can now be activated simply bysaying Siri with that my foe just turnedon Siri on me and that was pretty muchall the AI related news to come out ofthe WWDC event again I did make anentire video breakdown of WWDC I highlyrecommending watching that for a deeperbreakdown of everything they covered andnow let's get into all of the stuff thatalso happened this week that was kind ofovershadowed because of the Apple eventso this was actually news that droppedlast Friday but Friday I was travelingback from the augmented World Expo and Ialso record these videos on Thursdays sothis didn't make last Friday's video butlast Friday Google announced some newresearch called style drop this allowsusers to generate new images that followa specific style of their choice givenonly a single style reference now thisdoesn't seem to be available foreverybody to use yet this is sort ofresearch that they've put out but Iimagine the code will be availablefairly soon but here's an example ofwhat it can do here is this input imagefor reference of a watercolor of someflowers and then any prompt you give itwill generate the image in that samestyle so a baby penguin it's got thatsame watercolor style a banana a bench aboat a cow a moose a robot it allmatches this watercolor style herehere's another watercolor style of a catwith this initial reference image nowall of these images are generated inthis similar style here's a differentstyle here's the style of a starry nighthere's another one where it started withthis little phone graphic here here'sanother one looks like a little clip artkind of image here's one that looks likea sticker and here's some more andanother example so many cool examples ofhow this can be used here's another oneanother one even more it's only a matterof time before this feature is added totools like automatic 1111 and Leonardoand dream studio and all the varioustools that run stable diffusion underthe hood I can see this being especiallyuseful for things like logo designs andicons on a website here's something elseI came across this week which is coolbut also kind of creepy it's a tool thatallows you to replace the face on anyvideo so here's an example that I cameacross you can see here's the initialvideo and then the face just shifted tosomebody else's face here and here'sanother one of a stock footage of thisguy kind of dancing and then boom theface shifts to somebody else's face andhere's an example with the intro fromthe office where somebody took all ofthe characters from the office and thenreplaced the faces on all thosecharacters with their own face this isactually pretty hilarious especially thescene coming up right there that's scarynow this is actually something calledRoop r-o-o-p you can find it on GitHuband it is open source and it's availableto download you can actually install iton your local computer this is one thatI'll likely make a future video tutorialon because it's one that I'm gonna playaround with but there are instructionsright on the GitHub page I'll make sureit's linked Below in the description yougive it a face that you want to insertinto the image you give it a Targetvideo and then you just let it run andit will swap out the face onto thatvideo Once again this stuff is kind offun to play with but also getting alittle bit scarier we are getting closerand closer to that point where you canno longer trust video footage we alreadycan't trust images or audio anymore andwe're about this close to really reallynot being able to trust video very muchanymore and I know everybody thatwatches my videos is probably only goingto use it for good but I'm reallyworried about all the people that haveplans to use this for not so goodreasons speaking of which the news cameout this week that Google and Facebookare being urged by the European Union tolabel AI generated content soessentially if anything was AI generatedthey want laws in place where whoeverposted that content needs to mark thatit was AI generated so anybody who viewsit knows it's AI generated personally Idon't think that's a bad idea however Ijust don't understand how it's going tobe enforceable I feel like this is goingto be a constant cat and mouse gamewhere people get better and better andbetter at making fake images fake audiofake videos tools get better and betterat spotting them and labeling them asfake while the people that are creatingthem just figure out ways to get betterand better and better at circumventingthese tools that try to figure out ifit's fake or not so while I agree thatit's smart to have things in place wherewe're labeling everything is AI I justdon't know how this can possibly beenforced for the long term also in veryrelated news this week it's alreadyhappening now that the political seasonis upon us here in the United Stateswe're already seeing fake AI generatedimages circulating to sort of discreditthe other campaign these images havebeen circulating as part of a campaignfor Ron DeSantis against Donald Trumpwith images of trump kissing fauci andhugging fauci and they're being used totry to discredit and disparage Trump I'mnot going to go down the political routeI'm just pointing out that politiciansare starting to use fake images to tryto discredit other politicians and thisis exactly what Sam Altman was superworried about out when he did his Senatehearing last month I love AI I love thetech I love the tools but this is theworld of AI right here that really kindof bums me out sometimes now in othernews hugging face just released anotherlarge language model of their own is a40 billion parameter open source modelthat in their testing actuallyoutperformed all of the other opensource-ish models I say ish because LamaI don't think is technically open sourceyou can actually come to hugging facessite and test out the model yourself andhave chats with it but it's not quiteready for us to run it on our owncomputers yet because right now itrequires 90 gigabytes of GPU memory andI don't believe there's really anyconsumer gpus right now that are goingto give you 90 gigabytes of GPU memorybut if you want to come play around withthis model and see what it's capable ofagain they do have a demo where you canplay with it right here explain it likeon five it's explaining the Everettinterpretation it's saying that if youhave a coin and you flip it and it landson heads then there is another Universewhere the coin lands on Tails these twouniverses are parallel to each othermeaning that they exist simultaneouslybut independently in my very shortamount of testing it was actually apretty impressive model that gave prettygood responses I'm personally stillusing chat GPT and GPT inside of theopen AI playground for most of my largelanguage model chatting needs but I amexcited to see what the open sourceCommunity does with some of these modelsthat are being put out and madeavailable for others to iterate off ofalso this week this image was leakedfrom Instagram I don't believe reallyanybody knows the authenticity of it sotake this with a grain of salt but therewas this image released that says chatwith an AI bring AI to your chats for amore fun and engaging experience answersquestions gives advice chat with 30 AIpersonalities and find which one youlike best Inspire your creativity findthe best way to express yourself and gethelp writing messages this is really allwe have to go on right now I don'ttotally know if this is a chat botthat's going to be inside of yourInstagram DMS or if this is going to bea chat that's going to help you writesocial media posts better if it's goingto do both really all we're going off ofis this one screenshot that got leakedon Twitter so we'll have to wait and seewhat Instagram rolls out just somethingto keep an eye on all right let's getinto something really really fun thisvideo dropped the other day from acompany called Daz 3D now if you're notfamiliar with Daz 3D they're a companythat helps you create character modelsit's a free tool that you can downloadand create characters from and then takethose models into tools like blenderunreal Unity Etc well this week thatcompany announced that they're actuallyadding AI into the mix and here's thedemo video that they released they typedmake me a middle aged Asian guy and thenit generated that image now make himolder and shorter it shortened him gavehim gray hair made him older let's givehim a uniform that makes him look likehe's from a sci-fi film they gave himsome sci-fi film uniforms let themswitch them out a little bit what if wemade the character a woman withmid-length hair changed it out tosomething that looks like a Barbie dolltake out hair let's give her a buzz cutlike the original character all rightnow she's got a buzz cut no go back tothe older man all right goes back tothat older man that's awesome can youshow me four stages of him turning intoan alien generates four different stagesof that guy turning into an alien andyou put number four on an alien planetokay puts him on an alien planet greatnow export it to Unreal Engine and thereyou go the character is in Unreal Engineand because it's got so much data and somany other models that it's been trainedon you could pretty much type anythingyou can imagine and have it generate acharacter that looks like that nowcombine this tool with a tool likeconvey and we're getting that muchcloser to anybody being able to createtheir own video game now you may not beimmediately familiar with convey butremember this demo video of somebodycoming up and having a chat with acharacter in a Ramen Shop and theconversation taking place in real timethis was a presentation that was at thecomputex event and Jensen Huang showedoff this technology of somebody having areal-time combo conversation well thecompany that made this conversationhappen that gave this character here abackstory gave him a personality thatcreated the conversational style forthis character that company was conveyand this is a tool that anybody can useright now to create characters that willhave unique one-on-one conversationswith you in real time inside of a videogame now what's also cool is that Irecently sat down with the founderpernendu of convey and we had aconversation he gave me a private demoshowed me behind the scenes showed mehow all the tool works and how anybodycan set it up in their own gameenvironment and I'm excited to actuallyrelease this conversation in this demoin a future video the point beingimagine taking a tool like the Daz 3Dcharacter creator where you can promptany character you can imagine and getthe look of the character you want andthen create the personality andstoryline and conversational Style witha tool like convey and you can takethose two and pull them into UnrealEngine or unity and now you've got agame character with its own personalitywith its own way of thinking with itsown conversational style and you're wellon your way to creating very realisticNPCs I know it's a buzzword but it'sreally democratizing things like gamedevelopment also this week Runwayreleased Gen 2 to the public in the pastit was in a private beta where only ahandful of beta testers were able toplay around with it inside of Discordwell now it's available for everybody touse over at runwayml.com you cangenerate up to 90 seconds of video forfree before having to sign up for a paidplan and anybody can demo this now justgo to runwayml.com click on try now andclick on introducing Gen 2 and you cangenerate any video you can imagine Nowit only generates four second videosright now and you do have a limitedamount of free credits that you can usebefore you need to upgrade this isprobably the most coherent text-to-videomodel that's available right now let'sdo a realistic race car drifting thingaround a track go ahead and generatethis and see what we get it's actuallypretty quick as well and it took lessthan a minute to generate that andhere's what we got out of it now the caris kind of going in the wrong directionit's sort of drifting in Reverse but notbad I mean it's pretty coherent you cantell it's a car you could tell it's on arace track it's got the smoke coming offthe wheels it's a fairly decent videoit's just kind of going in the wrongdirection oftentimes with all of thisgenerative AI it takes prompting it andgenerating it a few times before youfinally get what you're looking forexactly if you follow me over on Twitterover at Mr e flow I actually did sharesome other examples of what I createdusing gen 2. here's one there if Iscroll down a little bit here's someother ones that I generated so if youwant to see what I made and how I madethose make sure you're also following meover on Twitter at Mr eflow here'ssomething else that I came across onTwitter this week from this companyFlair AI it's a company that you cantake your physical products and sort ofblend them with AI to get real coolproduct shots and their new tool letsyou put it in people's hands it lets youput them in different environments itlets you even choose the skin tone andthe way the person looks so check outthis super quick demo that they put upgenerate human models on flare AI tosell your products our AI humans arehighly diverse and customizable weprovide thousands of faces to choosefrom in a variety of dynamic poses goneare the days of coordinating expensiveproduct photo shoots that can costthousands of dollars simply choose apose template position your product andthen hit generate the opportunities areendless you've got you covered whetheryou're targeting different demographicsor aiming to highlight specific physicalattributes like skin tone hair color orage it's about more than justconvenience and affordability our goalis to use AI to also promote inclusivityat scale that everyone can seethemselves represented in your marketingcampaigns so that looks pretty coolespecially if you have a physicalproduct let's say you've got ane-commerce store and you need someimages for Shopify fi you could justtake a picture of your product and thenput it into all sorts of different posesnow this demo it showed it withdifferent people holding it but Flairalso lets you put it with differentbackgrounds and different environmentsand all sorts of cool stuff so you cancheck that out over at flair.ai alsothis week the company Baidu out in Chinareleased their own AI Cloud codingassistant watching their demo it lookslike you type in some code and it sortof Auto fills the code it's like youstart typing and then it kind of triesto figure out what you're trying to typewith the code and then automaticallycompletes the code for you it lookssimilar to what tools like Amazon's codeWhisperer and github's copilot do butnow I do is in the mix as well there'sbeen a ton of tools out there that willwrite articles for you and create blogposts and do your copywriting for youand all that sort of stuff there's somany now that I've actually stoppedaccepting new tools like this on theFuture tools website because I getprobably 10 to 12 of them submitted aday there's just so many clones of thissame concept now however now there'sthis jetpack AI assistant and what makesthis one different is this one wasactually created by automatic which isthe company that made WordPress so thisis a WordPress add-on by the creators ofWordpress that help you create blogposts and do the Creative articlewriting that you would do in yourWordPress websites so here's an exampleof how it works they press the plusbutton up here they click on the AIassistant button and then they just askit write a blog post about Japaneseculture and food and then it just goeson and writes the blog post for you soif you've used tools like Jasper this iskind of like that but it's just builtstraight into WordPress there is thisarticle over on the jetpack blog thatbreaks down all of the features insideof it it's pretty much what you wouldexpect you could change the tone of thearticle make sure to check for spellingand grammar it's got the conversationalUI so you would chat with it like youwould have chat bot and I will make surethis article is linked up below thevideo so you could read all about it ifyou you use WordPress and you want toinstall this one for yourself you get 20free requests and then after that it's10 bucks per month to use and if you arehosting your website on wordpress.comthe AI assistant is available for freeof charge on all wordpress.com sites fora limited time Bard got an update thisweek and Google announced that it'sgetting better at logic and reasoningnot much to cover there and finallylet's wrap up with this news thatstability.ai just launched a new featurecalled uncrop now I'm sure you'refamiliar with photoshop's generativefill well uncropped kind of does thesame thing you can take an originalimage expand the borders and it willautomatically use AI to generate what isin that outer area of the image sohere's a square image they make ittaller and it adds to the image and thisone right now is actually available touse for free so if you don't pay for aPhotoshop subscription or your freetrial ran out you can actually use thisright now by going over to clipdrop.coslash on crop and you just drag and dropan image in so I'll go ahead and drop inthis image that I generate AI I'm goingto go ahead and expand it on both widthand height see what it generates for meand click next and here's some exampleimages of how it added to that originalimage I don't know what's going on withmy fingers in that one that one doesn'tlook too bad actually and this one itseems like it mangled my fingers andthen added a whole bunch more fingersaround me but that one actually prettysolid image now yes this is an AIgenerated image so I know the nextquestion people will ask me is will itwork with real life images as opposed toAI images well let's find out here's arandom image that I have of Elon Musk onmy computer let's go ahead and drag thisin here let's make it a lot taller andwider so it adds to both the width andthe height let's click next sometimesthere's a little bit of a queue ofpeople waiting in front of me but frommy experience it actually moves prettydang fast and here's what it generatedElon holding some sort of book and allof these it seems to think he's holdingsomething so he's holding a book I'm notsure what that is maybe also a book thisone looks like he's holding a pillowwith a bed in the background uh I don'tknow what he's holding there maybethat's a monitor with mountains inbackground but anyway yes it works withreal images too it will try to add on tothose real images so despite the BigApple event this week there was actuallya ton of other news a lot of big stuffhappening it did get overshadowed byApple but there's a lot of new toolsthat you can use out there a lot of coolresearch that's happening and this isjust the space that never sleeps there'salways new cool stuff that's happeningall of the time I just love it every dayI wake up hoping hey hopefully there'smore AI news that I can go and talkabout today because I just love all thisTech I nerd out about it every day I'mso excited to be able to share it withyou and point you in the direction ofstuff that maybe you haven't seen yetand I'm really appreciative that youcheck out this YouTube channel and ifyou want to even further stay in theloop head over to futuretools.io I doupdate this news page every single daywith even more news than what I share inthese videos and all of the cool toolsthat I come across I am adding tofuturetools.io on a daily basis and ifit's a little bit too much for you jointhe free newsletter and every FridayI'll send you the tldr for the week ofjust a handful of tools a handful ofnews articles tools a few YouTube videosand a cool way to make money with AIgoes out every Friday you just gotta goto futuretools.io click this button andI'll hook you up thanks again for tuningin really appreciate you see you guys inthe next video byeforeign[Music]