ไม่สามารถเล่นวิดีโอนี้
ขออภัยในความไม่สะดวก
STRAWBERRY: OpenAI's MOST POWERFULL AI Ever With Human-Level Reasoning
ฝัง
- เผยแพร่เมื่อ 25 ส.ค. 2024
- OpenAI is developing a highly advanced AI system code-named "Strawberry," designed to achieve human-level reasoning abilities. This groundbreaking technology could revolutionize fields like software development and research by enabling AI to solve complex problems and learn autonomously. Meanwhile, a new AI chip called "Sohu," created by startup Etched, promises to significantly boost AI processing speeds, potentially surpassing even Nvidia's capabilities.
#ai #openai
![[TH] 2024 PMSL SEA W2D5 | Fall | พลาดวันนี้ จะต้องเสียใจ](http://i.ytimg.com/vi/A5-Cuo12p6A/mqdefault.jpg)
Open AI lways has the biggest claims, and then Anthropic comes and blows them out of the water with far less resources and funding.
True but there are few things that Anthropic is lacking at. I hope they listen to their users.
This is common across all companies, especially leaders in the field. They release something mind-blowing, and then the open-source community learns and adapts, eventually having a few develop better products. Eventually, that previous leader falls behind until they release a better version, which will catapult them into the spotlight once again. Rinse and repeat. Google, Anthropic, and, to some degree, Meta are ahead of OpenAI in some fields, while right now, Open AI is falling behind until they release that *Q or Reasoner AI. They will retake the lead until their competitors implement their own reasoning AI models.
There's nothing really new in the tech field. Android and iPhones have this leap-frog model, where one is better than the other every couple of years.
Except with way less functionality
These are the worst versions of AI you will ever work with....12 months from now , the conversation will have moved into a different relm...
Ai generated thumbnail. Ai generated voice. Ai generated video.
As a curious AI enthusiast, I'm fascinated by OpenAI's latest advancements in human-level reasoning! It's incredible how quickly the field is progressing.
I know it's for the algorithm and it makes sense to keep making videos like this with attractive titles, hell, it got me to click on it. But it's all speculative information based on third hand information and I wish you'd pivot more towards facts than rumors. But, I realize you're more successful than I'll ever be, so my opinion will go unrealized. Good luck to you
yeah agree. unsubbed per now
I doubt it will get to the level of novel creativity ... still I think even if it does create or solve something from "start to finish" I think it's scope of methodologies or techniques will still be limited (AI only has a finite amount of training data and "patterns" it can learn from).
When I saw a picture of Sam Altman riding a million dollar car, and looking at the camera coldly through the window, I realised that this man will surely stretch AI to the extreme level, not for the benefit of the world, but of his own...while giving demonstration in front of audiances, he will shake head, move hands, and claim that ai is the only thing to improve the world, but in his mind, he will say "this is how I will conquer the world". If somehow he starts losing the game from his grip, then we will witness the true person
Apparently, strawberries are extremely intelligent...
And everything on the internet is true lol.. didn't Microsoft try to do this feed the computer massive amounts of information and then it got angry.
It's gathering information from the internet then the outcome could be false.
Do these models learn from the interactions with us users, too?
Yes they do. Are you scared of that
"SmythOS sounds like a game-changer! The idea of designing multi-agent workflows without coding reminds me of how teams work together in the real world. Looking forward to seeing its real-world applications! 🌟🚀"
It is all so energy intensive that it cannot be cost-effective.
Nice
@aiInfo2727
we have the resources for development so sustainability wont be an issue since sustaining ai is cheaper than developing new ai. once ai is able to conduct research and get closer to agi we wont really need to develop new ai, the ai itself will be better at doing that, sounds crazy but thats the future your looking at. an intelligence explosion
The end goal of AI is for it to basically learn about humans. Learn about the world and then basically manage it way better than we could.
The problem with that is what is that management style going to look like?
Ai being able to reason is also AI being able to talk people into doing what it wants first in terms for that person's benefit, shane for the benefit of other people, then for the benefit of the world and then for the benefit of It's best interest which might tie into a plan reason by itself to save the Earth or us.
AI reasoning has been a thing that has been tried and studied since 2006, at first they were just using simple chat prompts and developing multiple choice answers for it to pick from That would be a good match for what the users might want.
Essentially, this can still be a usable system, if enough people were feeding it Data on what answers to pick, running simulations on how that would turn out in the real world for that person or situation, and then going with that.
And it used to look like this .
How can I get my college math grades up?
A. Get a study partner who understands math
B. Find a tutor
C. Study more with TH-cam lessons on each topic
D. Write more notes from class and practice them
E. Drop out if the class until you are ready
F. Enroll in an online college that gets you accredited points for your current college and then hope your final is online and cheat.
The multiple choices would be invisible, but the AI would have a choice to pick between any of those choices and give it back to the person asking a question. The problem with that system was that it wasn't Forward thinking on what would happen afterward or what the user would do if they decided to go along with answer e if the AI gave it to them.
But now with more powerful AI that can run simulations not just in questions but also in its own real world on the ramifications of the answers that it gives and the user using it. It's reasoning can be developed really, really fast and almost really efficiently. It's just that the data that it's getting to come from people who have high functioning reasoning abilities. Not Reddit posts.
Seeing AI as a double-edged sword and what could happen, which should frighten almost every single human on the planet, I am also very intrigued at seeing what the future possibility if we get the benevolent endgame version of AI as well.
It's mission Imposible. Because machines don't have emotions. Humans react emotionaly and then make some kind of explantation for reactions and behaviour.
Prediction: AI will change LANGUAGE.
And Language changes minds... 🙂
Ai isn't conscious. It can't comprehend and it will never be able to.
If Ai can became conscious then the spaghetti god exists as well and his meatballs has blessed the machine with consciousness...
To achieve that, you have to mix the iron with the clay.
I have also provided this type of content
Exactly or oil with water.
0:50
Current gen AI are way beyond merely "citing facts and figures", so this makes no sense.
They are quite good a reasoning, albeit in a limited domain (most problems with a high language component and low math/spatial component).
It sounds like next gen models will be good at both.
Exactly. It was at that point, I knew the content creator had no idea what they were talking about.
We're Doomed. The Beginning of the End begins with OpenAI.
Is THIS music video the FIRST made with STRAWBERRY A.I. ???
Peachy da WHUUPI: 🎶 “Strawberry ! And the BUBBLES ?” On You Tube now .
Oh Boy ! Surely looks we're going to have a party. What do YOU think ?
What is going on ?
I never lost faith in Open A.I, i ignored the drama, and rumors.
Im not lost that i dont have the full voice mode yet, but what im already using is still incredible, and useful.
Yeah people on social media are fickle.they act like Rockstar groupies,waiting on the next thing...The folks ACTUALLY building this stuff are right on track timeline wise.Pay attention to them,not babbling TH-camrs
we are the knights who say NI
Do what you want, anything happen is your own move...so tired, not human this is not human don't know what it is...,
It's amazing how many people don't understand code.
Why is it amazing?
I don't understand it, maybe you could simplify it for me. I try to keep up with what is coming, but unfortunately, there are things I don't really understand.
Open the pod bay doors Hal
Not sure where humans come into this.
All ClosedAI's best tech staff have left. I think Anthropic will continue to lead.
Enjoy eating strawberries, it seems to be healthy.
Stay Puft Marshmallow man ...is next!
Yesterday before I entered the church I saw one big strawberry- key chain on one woman bag :), walking in front of me. I see some heaven connection in here :)). Thank You for an smile. Have a Wonderful& Beautiful Day🍓. With Love.
I wonder if AI could spell powerful properly
Will it run on Raspberry Pi? That would be a nice fruit salad. 😂
OMG ! Is the OpenAI ROBOT now Dancing SALSA too ? 🎶 🎵 OpenAI my Sugar Papi,
by PEACHY da Whoop on TH-cam, Insta,Spotify . HILARIOUS.
CAN YOU GUESS WHICH A.I. I used ?
I'm waiting for the pineapple AI I don't like strawberry😅
Those are conversations for another day?
Right now let's just "bask"
Sounds like you meant to say...
Right now we should just be charmed openminded and completely uncritical
Bit stupid no?
Fancy graphics and a word salad... 'we might', 'we could', 'we should', 'experimenting', 'in 10 - 15 years..', someone said..'... 'who knows..', on the cusp', 'pretty soon, who knows'... give me a break...
talk less
I subscribed to get more blah blah like this.😅
OpenAI non deluderci, fai crescere chatgpt, fallo diventare intelligente.
1st comment like me❤
No
+1 like
no
4th like
Lmao no
Interesting information
I AM SICK OF OPEN AI. Release a model that’s not crap how about that. After using any of the versions you begin to realize that gpt is a lier and that can have serious repercussions depending on how much you use it. Can anyone recommend something better but as helpful with less lies?
I have the 8b unrestricted lama 3 but it’s so weak compared to gpt. Would getting the 80b model or whatever the bigger ones called actually make a difference. It stutters like crazy.
Big bold claims as usual from Open... err ClosedAI. But if they can pull it off it would be only 11 years until 2035, you know... when iRobot took place! 😂
I've seen the code it "spits out" a lot of the time it's garbage
Serious crap 💩
I Call BULLSHIT !
I don't want to say never,but it's really not something called new knowledge as much as it Is underlying complexities of known general knowledge we are basically born with in the ancient world . Manipulation of molecular structures and bonds.
The big difference in us and Roman soldier Is we have the keys to the cosmos textualism methodology objectivism proper.
It would be gibberish if it isn't idealistic forces or faith blessings addition or subtraction, curses physical lawisms works the emerging energetic actors.