New GPT upgrade, 3D model generator, New AI voice assistants, Liquid AI models, Pika 1.5

Поделиться
HTML-код
  • Опубликовано: 1 ноя 2024

Комментарии • 125

  • @theAIsearch
    @theAIsearch  27 дней назад +13

    Thanks to our sponsor Katalist, a powerful AI tool that creates storyboards in seconds: www.katalist.ai/?fpr=ref

    • @AIChameleonMusic
      @AIChameleonMusic 27 дней назад

      b4 a major election this could be used for "uncanny valley propaganda" as Elon IS ALLOWING TRUMP TO DO WITH GROK and META just wants to be more responsible number 1 than that. Secondly atm its just not practical, it cost a ton of money and HOURS to generate a prompt, When it is more efficient and cost effective to run, They can think about releases.

  • @budekins542
    @budekins542 27 дней назад +21

    Looks like SORA have met their match. .The day MovieGen is released with the photo to video generator will be the greatest day in the history of A.I😂

    • @theAIsearch
      @theAIsearch  26 дней назад +1

      for now, both are just vaporware. i'll believe it when it actually comes out 😉

    • @budekins542
      @budekins542 26 дней назад

      @@theAIsearch True!

  • @AshraldRails
    @AshraldRails 26 дней назад +14

    how many R's in "strawberry"
    Liquid 40b: There is 1 R in "strawberry"

  • @manonamission2000
    @manonamission2000 27 дней назад +29

    Flux: "dating apps, we got you"

    • @theAIsearch
      @theAIsearch  27 дней назад +14

      dating apps will soon be filled w these

    • @manonamission2000
      @manonamission2000 27 дней назад +3

      @@theAIsearch some are owned by publicly traded firms, due for an audit

    • @TheBann90
      @TheBann90 27 дней назад

      I cant help but giggle for days at some of these possibilities we have with AI 😂😅

    • @ianwatson5767
      @ianwatson5767 27 дней назад +2

      ​@@theAIsearchthat's a good thing, at least an AI will respond to messages you send them on dating apps. 😂

    • @ViralManiaOficial
      @ViralManiaOficial 26 дней назад +1

      It's no joke, in my profile I have a pic I made with a Lora of myself, I made it so close to reality that is almost indistinguishable, but with that I have a picture of myself with suit 😅 I was also able to verify it as it if it was really me, I mean it really looks like me with suit

  • @adra420
    @adra420 27 дней назад +4

    20:25 No soft talking, straight up spitting truth roasted the F out of it, this is why I love this channel.

  • @NeuralDev
    @NeuralDev 27 дней назад +5

    I believe Sora was so buggy that they decided to not release it and during this time so many companies released great AI video that the early avantage they had with Sora is totally gone

  • @AIChameleonMusic
    @AIChameleonMusic 27 дней назад +14

    Meta didnt release their "moviegen" yet. It is not yet cost effective they stated as much lets try and keep that in mind

  • @hdfsgervda
    @hdfsgervda 27 дней назад +9

    Quality of Meta Movie Gen looks way better than OpenAI Sora. But neither actually matters since they'll never release them. I'm expecting Black Forest Labs to 'win' like StabilityAI's Stable Diffusion did for image generation (same team members but better business model)

    • @VraserX
      @VraserX 27 дней назад

      OpenAI ist actually training a new Sora model. I‘m sure it will be released next year with GPT 5 and will be state of the art.

    • @theAIsearch
      @theAIsearch  26 дней назад

      exactly. both movie gen and sora are just 'vaporware' for now

  • @TheEleventhDoctor
    @TheEleventhDoctor 27 дней назад +3

    It’s sucks for people in the UK who don’t get to experience Meta AI. :(

    • @theAIsearch
      @theAIsearch  26 дней назад

      you're not missing out on much. meta AI (llama) isn't great, and movie gen isn't out yet.

  • @vi6ddarkking
    @vi6ddarkking 27 дней назад +3

    I am really curious to see if the liquid models pan out.
    The lower memory and computation requirements would be a gamechanger for local AIs.
    Also I am rather curious to see when the ControlNet tools for the 3D models generators get going.
    Considering the Openpose Blender Scene we use It'd be quite a nice fit to use it to generate the image and then already have the skeleton ready for rigging.

  • @ArcanePath360
    @ArcanePath360 27 дней назад +4

    I'm getting the feeling that using cameras and actors to make movies at some point will be like what happened to silent movies. Since anyone will be able to do anything. We have completely devalued the entire movie making world. This is not a good thing.

    • @alonzoperez2470
      @alonzoperez2470 27 дней назад +4

      Why isn't it a good thing? It's a good thing. We are heading to a cyberpunk reality 😉

    • @VraserX
      @VraserX 27 дней назад +2

      It’s absolutely a good thing. Everyone can make movies and be creative without the need of millions of dollars. Imagine how many brilliant writer are out there who will soon be able to create blockbuster movies and TV Shows.

  • @GTFOVIS
    @GTFOVIS 26 дней назад

    Thanks for your videos! Best AI observation channel!

  • @MarioTGP
    @MarioTGP 27 дней назад +2

    of course the tribe has to add neighbors if your prompt only says "man" the fatigue is real

  • @hiken95
    @hiken95 26 дней назад +1

    24:22 "sorry can we change the subject"

  • @rickchaves
    @rickchaves 27 дней назад +1

    Yours are the best and most informative videos about AI out there. Good work. One question, do you have a video showing the difference between ChatGPTs versions? I didn't find any that are up to date with the latest releases. Thank you!

    • @theAIsearch
      @theAIsearch  26 дней назад

      Thanks. Are you asking about o1 vs 4o?

  • @yeaaaonly
    @yeaaaonly 27 дней назад +3

    we need the meta video now!!!!

    • @theAIsearch
      @theAIsearch  27 дней назад +7

      hope they actually release it

    • @AIChameleonMusic
      @AIChameleonMusic 27 дней назад

      @@theAIsearch they have no intention on doing that atm, unfortunately it is not cost effective to do so. IF AND WHEN it becomes more cost/time effective that may change. Just to set expectations.

    • @manonamission2000
      @manonamission2000 27 дней назад +1

      @@yeaaaonly has OpenAI released Sora yet? nope. will they ever? tech leadership moved to Anthropic, so...

    • @yeaaaonly
      @yeaaaonly 27 дней назад

      meta video looks very promising, i hope they make it open source

    • @manonamission2000
      @manonamission2000 27 дней назад +1

      @@yeaaaonly after the U.S. election 😅

  • @tarcus6074
    @tarcus6074 26 дней назад +1

    It feels like AI winter.

  • @fernandoz6329
    @fernandoz6329 26 дней назад

    Good video as always!!! The phrase '...they are probably cherry-picking the examples...' you bet. Given the complexity of some prompts, it seems they are using an AI which generates the prompts but also the ai somewhat 'knows' what work best with the video generator.

  • @razak8528
    @razak8528 26 дней назад

    Next step is to create the holodeck from Star Trek. Then we’ll be set.

  • @alonzoperez2470
    @alonzoperez2470 27 дней назад

    Copilot live voice mode is cool as hell. It matches OpenAI live video

  • @NostalgicMem0ries
    @NostalgicMem0ries 27 дней назад +2

    will meta stuff be available in europe? ;\ also copilot is down atm, looks like i have advanced mode but it says "im working on making time for you so please check back soon", maybe it has line to wait :D

    • @theAIsearch
      @theAIsearch  26 дней назад

      yeah they said it won't be released 'anytime soon'

  • @manonamission2000
    @manonamission2000 27 дней назад +4

    folks have caught onto Kling... videos take ages to generate now

  • @dreamphoenix
    @dreamphoenix 27 дней назад

    Thank you.

  • @Alttabdie
    @Alttabdie 26 дней назад

    Cant wait for the day that you reveal you have been using ai to voice your videos.

    • @theAIsearch
      @theAIsearch  26 дней назад

      i'm actually an AI from the future

  • @issiewizzie
    @issiewizzie 27 дней назад

    Within a day, the Microsoft copilot seems to have changed. Mine can whisper and play music and also generate sound effects.

  • @High-Tech-Geek
    @High-Tech-Geek 25 дней назад

    I got Copilot to sing. It was actually very good.
    But after talking for a few minutes, it would simply stop responding and I'd have to close it and restart.

  • @jantube358
    @jantube358 27 дней назад

    6:54 I knew this will come but this came quicker than I expected. We should try prompts like "security cam footage, low resolution, black and white, 480p". That's really scary.

    • @theAIsearch
      @theAIsearch  26 дней назад +1

      yep, 'evidence' can easily be forged with this

    • @Razumen
      @Razumen 25 дней назад

      @@theAIsearch Unless it actually looks like the place it's supposedly filmed at (it won't) it can't.

  • @High-Tech-Geek
    @High-Tech-Geek 25 дней назад

    Apparently Copilot also has time limits. After about 30 minutes of chatting, it popped up and said "You have 3 minutes left today". Gah!

    • @theAIsearch
      @theAIsearch  25 дней назад

      good to know! thanks for sharing

  • @Yaddlezap
    @Yaddlezap 27 дней назад

    @3:20 He was all pasta inside. Of course.

  • @TheBann90
    @TheBann90 27 дней назад

    I wonder if we can make silicon chips that takes even better advantage of these liquid neural networks. Or if the gains from that is negliable.

  • @NekoDezzu
    @NekoDezzu 27 дней назад +1

    cat girls, fox girls is approaching !

  • @abrahamsimonramirez2933
    @abrahamsimonramirez2933 26 дней назад +1

    now you can literally pump dogecoin, stonks 📈

  • @MatkaFrFr
    @MatkaFrFr 26 дней назад

    Hi AI Search! Can you do a video about "Voicemod"?
    I really want to know if its a virus or not.

  • @GTe-gv3bn
    @GTe-gv3bn 27 дней назад +1

    24:39

  • @arshdeepbhangu8884
    @arshdeepbhangu8884 26 дней назад

    6:55 😮

  • @LaurinkoSattumaa
    @LaurinkoSattumaa 26 дней назад

    Y no electric sheeps in em videogenerator showcase vids?!

  • @AxelWXChannel
    @AxelWXChannel 25 дней назад

    hi everyone, our channel has many songs we made wth AI, listen to them and tell us what u think so we can make better songs. we have many songs we haven't release yet.

  • @adamholter1884
    @adamholter1884 27 дней назад

    I don't think liquid foundation models are actually based on liquid neural network architecture. It's just an adapted version of Transformer I think

  • @ImNotQualifiedToSayThisBut
    @ImNotQualifiedToSayThisBut 27 дней назад

    So, OpenAI released their voice mode to free users before I got it even though I've been a plus subscriber for over a year now?

  • @edengate1
    @edengate1 26 дней назад

    How to utilize the prompt to have that realistic selfie from real people on Flux?

    • @theAIsearch
      @theAIsearch  26 дней назад

      try selfie.jpg. see this for more tips ruclips.net/video/fJ-U6eORJjc/видео.html

  • @ilyass-alami
    @ilyass-alami 27 дней назад +1

    Is copilot voice free and unlimited ???? I hope so

    • @theAIsearch
      @theAIsearch  26 дней назад +1

      it's free & i haven't hit any limits yet

    • @ilyass-alami
      @ilyass-alami 26 дней назад +1

      @@theAIsearch
      Do you use vpn ?? You are in USA ?

  • @DemetriusZhomir
    @DemetriusZhomir 27 дней назад

    Meta's video generation model is probably not so cost-effective, and it's too powerful to be safe for public use... So yeah, another SORA.
    Meta and OpenAI don't want to get sued a lot for the harmful outputs of the models they provide.

    • @theAIsearch
      @theAIsearch  26 дней назад

      yeah they said it wont be released 'anytime soon'

  • @MichaelBaynana
    @MichaelBaynana 25 дней назад

    anyone knows a good free chat gpt with voice fuction for smartphone since copilots voice function doesnt work anymore

  • @Dr.R.
    @Dr.R. 27 дней назад +3

    Are you serious?! Advanced voice for free users has to be a bad joke. I am paying and still waiting for it 😭

    • @Reddglawer
      @Reddglawer 27 дней назад

      I am also in one of the regions its not available in. I used a vpn app to connect to a new york server and I got access.
      Free VPN apps work too

  • @ditheraith
    @ditheraith 26 дней назад

    19:09 "Free" just sign away your soul

  • @ManoOne-Music-Production
    @ManoOne-Music-Production 27 дней назад

    Plus user here. Advanced voice mode icon appeared last week, but then went back to the old one, before I could try it. Has this happened to anyone else?

    • @theAIsearch
      @theAIsearch  26 дней назад +1

      try reinstalling the app

    • @ManoOne-Music-Production
      @ManoOne-Music-Production 26 дней назад

      @@theAIsearch I tried that several times since it happened. It’s really strange. Also no Canvas, although that seems to be an iPad or desktop thing? Thanks for replying!

  • @MrGFYne1337357
    @MrGFYne1337357 27 дней назад

    Did anyone notice how he said "fiery". He's not real.

  • @quebono100
    @quebono100 27 дней назад

    2:25 thats douge not shiba inu

    • @AmiciCherno
      @AmiciCherno 27 дней назад

      Doge is a shiba inu

    • @quebono100
      @quebono100 26 дней назад +1

      @@AmiciCherno Oh I didnt know this

  • @hqcart1
    @hqcart1 27 дней назад +2

    by the way, all the effects you described in the begging is NOT AI, it can be generated by normal algorithms.

  • @gregblank247
    @gregblank247 27 дней назад

    Canvas doesnt perform fast with coding though. It scans or rewrites the entire code every time and is super inconvenient! You may not notice for small set of code, but make a small edit to 200-300 lines of code and it takes a long time going through everything every time. Very disappointed in that! Text works super fast.

    • @theAIsearch
      @theAIsearch  26 дней назад +1

      thanks for sharing. i also dont use canvas for code - it's too inefficient. i prefer a IDE plugin like codeium or cursor

    • @gregblank247
      @gregblank247 26 дней назад

      @@theAIsearch This is supposed to be one of the best... fittenai

  • @Shootingfoul
    @Shootingfoul 27 дней назад

    how censored is it?

    • @theAIsearch
      @theAIsearch  26 дней назад

      copilot?

    • @SwornInvictus
      @SwornInvictus 26 дней назад

      If its not open source, it's almost definitely going to be.

  • @Jottocop
    @Jottocop 27 дней назад

    Can someone tell me how can meta spend so much for this expensive ai research if they then make it open source?
    Do they do it for pure generosity 🤣?

    • @gregblank247
      @gregblank247 27 дней назад

      They typically do this to establish a standard that people get hooked into using for later

    • @Jottocop
      @Jottocop 27 дней назад

      @@gregblank247 thanks, so we should expect something even more revolutionary from meta that will be needed to pay?

  • @mitrooper
    @mitrooper 26 дней назад

    For a AI channel you are sure negative about the tech. 7:03

  • @megafoxatron3rd521
    @megafoxatron3rd521 27 дней назад

    awesome i am comment 32

  • @MrNobodyX3
    @MrNobodyX3 27 дней назад

    30:16 WTF is this bullshit graphic that is not how any of this works

  • @Strammeiche
    @Strammeiche 27 дней назад

    *sad EU noises*

  • @yeaaaonly
    @yeaaaonly 27 дней назад +1

    first comment gooooo

  • @Kaoabbzu
    @Kaoabbzu 19 дней назад

    Dajjal

  • @Jack122official
    @Jack122official 27 дней назад

    ?

  • @ROBOTRIX_eu
    @ROBOTRIX_eu 27 дней назад