The OpenAI Team Finally Reveals The BEST OpenAI o1 Use Cases

แชร์
ฝัง
  • เผยแพร่เมื่อ 2 ธ.ค. 2024

ความคิดเห็น • 126

  • @fai8t
    @fai8t หลายเดือนก่อน +38

    thanks for making it brief not 40+ mins

  • @good_vibes_20
    @good_vibes_20 หลายเดือนก่อน +125

    AI helps people who like to think, think faster and better.

    • @Copa20777
      @Copa20777 หลายเดือนก่อน

      Yay🎉

    • @mace400
      @mace400 หลายเดือนก่อน +1

      Wow great reasoning!

    • @itzhexen0
      @itzhexen0 หลายเดือนก่อน +3

      Open up 12 tabs. That's like 12 more brains.

    • @mace400
      @mace400 หลายเดือนก่อน

      @@itzhexen0 it’s not though

    • @itzhexen0
      @itzhexen0 หลายเดือนก่อน

      @@mace400 yeah it kind of isn't.

  • @GoodBaleadaMusic
    @GoodBaleadaMusic หลายเดือนก่อน +26

    It's going to help humans emotionally. Because right now we are all just mindsets trapped with no outlet to confirm our thoughts other than other mindsets trapped. These language models provide a non-human mindset for the first time in human history. It allows an emotionless objective interaction where each word doesn't change how the other mindset treats you for the rest of your life.

    • @spol
      @spol หลายเดือนก่อน +3

      Yeah love using it for settling arguments or at least moving the convo forward

    • @MMABeijing
      @MMABeijing หลายเดือนก่อน +1

      How is ai mindset non human? You can say it is easier to talk to, but the reasoning is what you would get from an objective human

    • @erinphilippe5993
      @erinphilippe5993 หลายเดือนก่อน +3

      Holodeck for the mind

    • @Zbezt
      @Zbezt หลายเดือนก่อน +2

      Exactly i lost friends discussing things that are being comfirmed more and more through out everyfacet of human abstraction XD in short they were too emotional to actively engage in intellectual matters which if they kept up to studies would realize i was right the whole time and they lost a loyal person atleast gpt can handle being analyzed to a T

    • @armadasinterceptor2955
      @armadasinterceptor2955 หลายเดือนก่อน +1

      Well said, as it has really helped me delve into my more creative thoughts, without outright rejecting(without considering).

  • @Cory-v4w
    @Cory-v4w หลายเดือนก่อน +2

    Human insight is short-sighted in the long run. The object of my affection is my imagination unfold.

  • @ThoughtfulAl
    @ThoughtfulAl หลายเดือนก่อน

    This should help those of us who are thoughtful.

  • @QueenLover-j5i
    @QueenLover-j5i หลายเดือนก่อน +3

    Insights By "YouSum Live"
    00:00:00 OpenAI's O1 series enhances productivity across fields
    00:00:29 Best use cases: strategy, coding, and research
    00:00:41 O1 aids in breaking down complex problems
    00:00:48 Effective ideation partner for business operations
    00:01:01 O1 mini excels in coding tasks
    00:06:25 O1 improves coding speed and detail
    00:09:01 Research automation streamlines laborious tasks
    00:09:37 O1 provides insights for optimal dog food
    00:11:00 O1 enhances mathematical problem-solving capabilities
    00:14:15 BCG uses O1 for streamlined problem-solving
    Insights By "YouSum Live"

  • @ytubeanon
    @ytubeanon หลายเดือนก่อน +2

    6:27 this is a clumsy / primitive way of coding, he should use Aider or Cline, config to use o1-mini (through Openrouter), prompt it and let it generate all the files, folders, connections, includes, etc...

    • @Sylabis1
      @Sylabis1 หลายเดือนก่อน +1

      @@ytubeanon have you exactly tested 01 mini over 01 preview? It’s hopeless at coding in comparison.

  • @patruff
    @patruff หลายเดือนก่อน +21

    I love the coding example where it doesn't build anything functional and he needs to copy and paste 12 files. Next level!

    • @jdholbrook33
      @jdholbrook33 หลายเดือนก่อน +1

      I hear ya. I tried to use it with Cline to eliminate all the copy paste but you can't use o1-mini via API.

    • @3rdeyesociety
      @3rdeyesociety หลายเดือนก่อน

      @@jdholbrook33 cursor

    • @jareddalessandro9813
      @jareddalessandro9813 หลายเดือนก่อน +9

      Bro seriously? Is copying/pasting too hard now?

    • @archvaldor
      @archvaldor หลายเดือนก่อน

      @@jareddalessandro9813 Yeah I mean for those of us who don't keep up with coding any amount of c&p is vastly preferable than having to do it from scratch.

    • @patruff
      @patruff หลายเดือนก่อน

      @@jareddalessandro9813 it's actually pretty bad when you can just tab your way through things like with cursor

  • @octopusfinds
    @octopusfinds หลายเดือนก่อน +1

    where is the original video? I want to see the structural analysis part

  • @Create-The-Imaginable
    @Create-The-Imaginable หลายเดือนก่อน +8

    Hmmm! What's up with all of these private videos from OpenAI all of a sudden?

    • @MA_SK_ID
      @MA_SK_ID หลายเดือนก่อน +4

      @create-the-imaginable From my experience, PR and marketing are scrambling to improve Open Ai’s image with the public. To the lay person, OAI technology is getting lost in the noise of all the advances across the industry this fall. All the while the public definitely is remembering drama reported in the media. If the average person thinks all the models work the same they might choose a model from a company that is reported in the media as being less controversial. Hence all the recent engagement and knowledge translation videos.

    • @zenyatta5064
      @zenyatta5064 หลายเดือนก่อน +1

      They changed name to closedai

    • @OhsoLosoo
      @OhsoLosoo หลายเดือนก่อน +2

      Locking in stakeholders investments & ensuring that while they went for-profit they are not switching to selling a “rug-pull” (like NFTs etc.) that the tech space has been littered with for the past 10 years.
      Canvas was a good upgrade & O1 is phenomenal if you know how to use it. I believe the only reason why the public got it at all is because of the hefty competition. Otherwise I believe they woulda locked these features behind enterprise.

  • @nuclearthrone2497
    @nuclearthrone2497 หลายเดือนก่อน

    Matthew posted a video, if you use the same prompts in 4o, the results are literally very similar

  • @itsjustme373
    @itsjustme373 หลายเดือนก่อน +3

    There are more than a few ways to get o1 to access the the internet.

    • @VisionaryPathway
      @VisionaryPathway หลายเดือนก่อน

      Yup Pplx API is a great one (simple)

  • @CindyWright-c9v
    @CindyWright-c9v หลายเดือนก่อน

    I feel like I can finally tackle this!

  • @johnbollenbacher6715
    @johnbollenbacher6715 หลายเดือนก่อน +4

    They could’ve found a better speaker to narrate the video

  • @MasamuneX
    @MasamuneX หลายเดือนก่อน +1

    ill keep it real if there isn't at least 20 steps in the CoT just use o1-mini or 4o

  • @yordanosabebaw767
    @yordanosabebaw767 หลายเดือนก่อน +5

    This channel creates AI-generated videos every 8 hours ................many people comment as if it's a real TH-camr.......wake up people

  • @rokljhui864
    @rokljhui864 หลายเดือนก่อน +2

    The AI used it's extensive knowledge to recommend feeding your dog Raw Meat, without saying the words 'Raw' or 'Meat'

  • @vio_tio12
    @vio_tio12 หลายเดือนก่อน +6

    did he actually run the code or just got it to output without even verifying whether it actually works lol?

    • @mace400
      @mace400 หลายเดือนก่อน +5

      Yeah the fact that they had this dodo doing the most basic examples, and a terrible job showcasing the technology, is not good lol

    • @bertobertoberto3
      @bertobertoberto3 หลายเดือนก่อน +2

      What the heck?? That’s not the type of video it’s just an overall demonstration

  • @Turnpike617
    @Turnpike617 หลายเดือนก่อน +2

    What model do you use for a financial advice?

    • @DJWESG1
      @DJWESG1 หลายเดือนก่อน +2

      Acorn database

  • @rangerCG
    @rangerCG หลายเดือนก่อน +1

    I'm not a big fan of o1. It adheres to OpenAI policies so much that it won't do anything, so it's not fun for me to use anymore, and I don't like an AI contemplating within itself about me when I can't see it (it offers summaries but even the summaries can be hidden in future updates).

  • @taomaster2486
    @taomaster2486 หลายเดือนก่อน +3

    People not using it are they.....

  • @jimmytaylor1279
    @jimmytaylor1279 หลายเดือนก่อน +3

    the model is useless without the internet or the ability to upload files. Cant get any useful work done with it, especially for coding. I don't want to past in all the code I'm trying to get help with so it has context.

    • @JA-gz6cj
      @JA-gz6cj หลายเดือนก่อน +1

      Why doesn't it have these abilities?

    • @jimmytaylor1279
      @jimmytaylor1279 หลายเดือนก่อน +1

      @JA-gz6cj nor sure why this one doesn't, but 4o with canvas doesn't because they want people to feel more secure in writing code so it can't get stolen. Not sure why o1 doesnt.

    • @JasonDoege-js8io
      @JasonDoege-js8io หลายเดือนก่อน +1

      @@JA-gz6cj because its just an mml, its not intelligent about how it does things its just a really good guesser

  • @V8Li
    @V8Li หลายเดือนก่อน +1

    I got too many people laughing in my face since Covid and AI as a programmer, I’m not talking to relatives and friends that are nurses, farmers and cops. I’m done with coding for others unless I do it for obscene money, it’s just too much hate.

  • @kwest84
    @kwest84 หลายเดือนก่อน +8

    KInd of funny hearing him say "its thinking and reasoning process" when the Apple study just showed that no LLM today can reason at all. Not even o1.

    • @DJWESG1
      @DJWESG1 หลายเดือนก่อน +3

      "it's" means the unique way in which it does it.

    • @DaronKabe
      @DaronKabe หลายเดือนก่อน +6

      It can think and reason better than you, that’s for sure

    • @Zbezt
      @Zbezt หลายเดือนก่อน +3

      I made it perform the apple problem with the same result but a simple reframing of reason made corrected the problem in short thats a skill issue on the test originator who doesnt understand the logic train itself

    • @archvaldor
      @archvaldor หลายเดือนก่อน +1

      @@DaronKabe With broad knowledge topics yes. It sucks at more specific niche information.

    • @newfrontiers5673
      @newfrontiers5673 หลายเดือนก่อน

      Ya and Apple has no interest in pushing their own agenda because their foundational model sucks and they are WAY behind....

  • @Sylabis1
    @Sylabis1 หลายเดือนก่อน +4

    o1 mini sucks at coding compared to o1

    • @17cloverbiz
      @17cloverbiz หลายเดือนก่อน +1

      I agree , o1 mini give you too much information that didn’t ask for . O1 preview stays on topic

    • @Sylabis1
      @Sylabis1 หลายเดือนก่อน +1

      @@17cloverbiz yeah that is part of the issue. It’s goes off on a tangent in its own direction and starts replacing critical components. The premise that’s its best for ‘coding’ over 01 preview is fuking ridiculous. It’s borderline unusable for any mid complex task. 01 preview is significantly better than any other model at producing high level code. Especially when it comes to statistical analysis. For some reason it’s useless using Cursor though, and falls back to high school coder level

  • @bertobertoberto3
    @bertobertoberto3 หลายเดือนก่อน

    Nice video

  • @sanjuburkule
    @sanjuburkule หลายเดือนก่อน

    This model says 'you are right', no matter what you ask. Its a bad model to make strategy decisions.
    How is Dubai? It will say 'Dubai office is a excellent strategy'
    If you ask, Berlin, it will say 'Berlin is a great choice'.
    Its not the best strategist. Its okay, or less than okay strategy helper.

  • @hiddenintheshadows530
    @hiddenintheshadows530 หลายเดือนก่อน

    "IT" does not think, only guessing at the next word based off the training data. the software is putting together words and isn't always correct.

  • @ggghhhbbnjjjbb2330
    @ggghhhbbnjjjbb2330 หลายเดือนก่อน +4

    4.5 minutes in and all I'm hearing is a long list of pompous marketing buzzwords, and nothing of substance. That's as much as I can handle.

  • @jackbauer322
    @jackbauer322 หลายเดือนก่อน

    gpt-4o does EXACTLY the same , none of the use cases presented here offers a significant overhead ... useless

  • @andreaskrbyravn855
    @andreaskrbyravn855 หลายเดือนก่อน +2

    So release it already then

  • @Мага_Крутой
    @Мага_Крутой หลายเดือนก่อน +6

    Man I am tired where is AGI???

    • @shadygamererfan377
      @shadygamererfan377 หลายเดือนก่อน +2

      Agi isn't gonna be here until 10 to 15 years from now if it comes into being, it may not be a llm model or a generative ai but a world ai by meta AI team!

    • @JA-gz6cj
      @JA-gz6cj หลายเดือนก่อน +1

      Right here buddy

    • @JasonDoege-js8io
      @JasonDoege-js8io หลายเดือนก่อน +1

      pi ai remembers and learns from conversations... i just taught it how to be an agi, test it!

  • @EFL0123456789012345
    @EFL0123456789012345 หลายเดือนก่อน

    I can buy a new microphone to Joe, in case OpenAi cannot pay. really... My God ... where is you media ppl?

  • @interstellarobesity
    @interstellarobesity หลายเดือนก่อน +2

    Please just look at your subtitles once before posting and fix them, its not that hard dude

  • @Michael-Gill
    @Michael-Gill หลายเดือนก่อน +1

    Edit.

  • @ArcherScience-u9w
    @ArcherScience-u9w หลายเดือนก่อน

    I’m very disappointed in the poor progress that’s being made in the area of data analytics

  • @angloland4539
    @angloland4539 หลายเดือนก่อน

    ☺️🍓❤️

  • @jennifergraham5615
    @jennifergraham5615 หลายเดือนก่อน

    Okay

  • @zyxyuv1650
    @zyxyuv1650 หลายเดือนก่อน +5

    o1 is a horrible. It's the most inefficient, highest priced model. lt has latency 50X higher than most other models. It costs between 4 and 100X higher pricing than other models. It's the slowest model ever made. And with a crappy 128K context. This is a compromised architecture to build on going forward, it cannot continue and OpenAI is going to have to backtrack because this latency is unacceptable. o1 already looks like a joke compared to Nemotron's performance.

    • @VisionaryPathway
      @VisionaryPathway หลายเดือนก่อน +13

      Fascinating take. Because the shift from ~200 lines of complicated Python code being the ceiling for zero bugs (with GPT-4)…
      Shifted to ~1,800 - 2,000 lines of bug-free code with o1 overnight. In one-shot prompting. It’s quite unreal for me, especially after 600-800 hours of Software Engineering & Programming experience (strictly with AI) this year.
      If OpenAI wanted $2,000 per month for me to access o1 instead of GPT-4 or any other models on the market… I’d pay that in a heartbeat. Because I’m currently able to extract around $250k worth of code per month (assuming full focus, 60h weekly on SWE) from current o1 models.

    • @zyxyuv1650
      @zyxyuv1650 หลายเดือนก่อน +2

      @@VisionaryPathway Please evaluate Nemotron first so you can see how far behind the 'poor mans o1' is. My point is that models like Nemotron will soon match the quality of o1 while costing I.e. 10X less and being 10X faster. Therefore OpenAI will backtrack on this design. You're talking about one moment in time, but we won't be waiting 30 seconds for generation to start in the future. OpenAI is closed. Nemotron is open source. Consider what an open source model like Llama/Nemotron could do if it was spending as much time and compute as o1.

    • @williamparrish2436
      @williamparrish2436 หลายเดือนก่อน +10

      @@VisionaryPathway Some people aren't very good with querying AI. And so they think it sucks. People who know how to ask the right questions get valuable answers. o1 is powerful in the right hands.

    • @xLBxSayNoMo
      @xLBxSayNoMo หลายเดือนก่อน

      ​@@williamparrish2436 right, I love the longer it thinks because I know it's answer will be even better and/or my prompt was a really good prompt to get it thinking. Mini is still a better Coder than preview but I use preview to review code or add on top of code from mini

    • @yoyo-jc5qg
      @yoyo-jc5qg หลายเดือนก่อน

      @@zyxyuv1650 openai's models have been getting cheaper and faster by the month and soon gpt-5 will have 01's abilities