How AI is transforming our computing and software

Рет қаралды 33,572

Жыл бұрын

As part of our effort to build increasingly capable and general AI systems, we’re making computing more powerful and sustainable. Our AI models AlphaZero and MuZero are expanding their capabilities to help optimise data centres, streamline video compression, and discover new algorithms to accelerate software. But this is just the beginning: we envision a future where more general-purpose AI tools can help optimise the entire computing ecosystem that powers our digital world.
Read our case study: www.deepmind.com/blog/optimis...

Пікірлер: 77

@TheSienn Жыл бұрын

This said a lot of nothing lol

@rubenheijmerink536 Жыл бұрын

No you just dont get it. It's pretty big if I understand it correctly. Early versions of AI were modelling human strategies. Next were the reinforcement learning algorithms. That need no human strategies and completely learn on their own. But does require you to set up the rules and rewards. Muzero, they dont need human strategies or rules to setup. But not clear to me what kind of applications this has.

@rubenheijmerink536 Жыл бұрын

@First Last The comment says this a whole lot of nothing. I proceed to explain why this certainly is not the case. Granted that the video is somewhat vague about it.

@yadayada7866 Жыл бұрын

Giga lol

@nofreewill Жыл бұрын

So just like a teaser, what's wrong with that?

@kubigaming3606 Жыл бұрын

i think it has been generated by chat gpt, generating lines full of words but empty in the logical sense

@Cray2TheZ Жыл бұрын

This feels like a clip from a dystopian future sci-fi movie.

@cereshin Жыл бұрын

but this time im not the protagonist rebelling im the dude surfing along while cheering

@samirdevechi8589 Жыл бұрын

Was this video AI generated by any chance? Maybe it's just the artistic style of the video but I have a feeling it also may have been a test of a new 'something'-to-video model they're working on.

@TheBigLou13 Жыл бұрын

I find the voice hard to follow as every third word feels like it was coming from a different environment. Like 10 clones brabbling words into a microphone one after the other.

@roberthand719 Жыл бұрын

Very strange to see that we are embracing our downfall, again, and again, and again, and again, and again. When will we learn that GOD, and US are all we need . GOD MADE US MORE INTELLIGENT THAN THE AI. YOU ASKED HOW I KNOW THAT??? LOOK IN THE MIRROR, and SEE GOD❣️ THE EGO, IS TRYING TO TAKE US DOWN THE ROAD OF DESTRUCTION AGAIN😢

@hb-youtube Жыл бұрын

@@roberthand719 So we don't need" any technology? Like we "Don't need" electricity?Maybe technically true but I'll bet you're not living without electricity(and in unlikely case you are, electricity still is what let your comment appear)Had you spoken about "focus" I might have agreed, if you spoke about the focus needs to be spiritual and moral and wisdom and so on. That's a statement many of us would agree with. This "we don't need no bleeping tech that I've seen a lot lately, isn't just a contradiction to people using it, it's also counterproductive due to the misplaced focus, which lets the powers that be end up the only ones with the power of tech behind them, but that's a second topic. The first one I hope is clear, statements like "we don't need" electricity or "we don't need" AI are equally flawed and in very similar ways...but focus on what the focus should be, and we might agree.

@espalorp3286 Жыл бұрын

FRIENDS! IT'S A TRAGEDY! Over and over, we march DOWNFALL road! WHY CAN'T WE LEARN? GOD and US, ONLY NEED! GOD MADE US SMARTER! AI? NO MATCH! How do I know? MIRROR! LOOK, SEE GOD! EGO, DESTROY ROAD! RESIST! HOLD FAITH! GOD'S LOVE GUIDE! REJECT CYCLE! SOULS AT RISK! TOGETHER, RISE! The above dialogue was generated by GPT, when asked to act like a lunatic evangelist.

@A4plainpaper 11 ай бұрын

Nope! This was created by Creative agency Across the Pond and animation studio Art&Graft.

@Nellak2011 Жыл бұрын

I love how Google can do all this complex AI stuff, but can't make a working calculator.

@therealhan21 Жыл бұрын

BRO STOP

@therealzeon 8 ай бұрын

How ??? What's wrong with the caluctor? Works fine for me

@rodneypantony3551 Жыл бұрын

If proteins are nanorobots, then DeepMind/Alphafold make 200,000,000 nanorobots available for commercial usage. That's a lot of startups.

@user-wv9pw9tq1g 7 ай бұрын

Today’s date is 11/15/2023. Hopefully by the time you’re reading this it’s no longer 5+ months since the last posted video.

@AmiraRahma-bf6nb 9 ай бұрын

I am very happy if the Alphazero chess engine participates in the Tcec chess engine tournament ❤

@playgamemi4833 10 ай бұрын

🎉 Thank You For Opening The Way To The New World

@loic_berlin Жыл бұрын

I'm in! Where can I apply?

@lucasteo5015 Жыл бұрын

The race to AGI is a game

@jjgerald7877 9 ай бұрын

Why gone comments about RDC#2? :/ I will delete them anyway.

@tbird81 Жыл бұрын

If you thought censorship in chatgpt was bad...

@ag.m. Жыл бұрын

Hey Dr. , I belong to India, I am 19 years old now, I want to become a neuroscientist. I want to do mind reading and any data entry in mind. The way it is done in harddisk. Because mind is the most powerful thing in the world. But we do not know in which branch of 'mind interconnect computer' engineer or neuroscience. (Which branch of medical or engineering should I take admission in?)

@du9 Жыл бұрын

Nice

@SudarsanVirtualPro 6 ай бұрын

What is next? Is it AlphaBard

@researcher6980 Жыл бұрын

❤🎉

@marshallmcluhan33 Жыл бұрын

Project Hail Mary incoming. Listen tot he uncensored LLMs. I'm sure AI will sort this out.

@De4dCert 8 ай бұрын

This makes me wonder if I should switch from Apple,,,,?

@scottwarren4998 11 ай бұрын

@Google Deepmind Hi there! I live in Sweden And i really liked the chess engine you created. Could your team start re-working on that engine in order to try to improve it? In a 1000 game-match, It would be so cool seeing Stockfish's strongest engine VS your improved Alphazero engine. /Ulrik from Sweden

@shubhamkhatter2196 Жыл бұрын

just tell me already how can i get a job at any AI research facility and what I have to do to get what it takes. you are killing me btw great video google

@omar2886 Жыл бұрын

This is a tremendously exciting juncture in the evolution of AI. DeepMind's ground-breaking models, AlphaZero and MuZero, provide a potent demonstration of how AI can not only augment human capabilities but can also redefine the very fabric of computation and software systems. The success of AI in transforming our digital world will be predicated on a balanced approach that leverages its immense potential while mitigating its risks. The "AI-first" future envisaged by DeepMind is not just about intelligent machines but also about creating a symbiotic relationship between humans and AI, underpinned by trust, transparency, and shared value.

@Learna_Hydralis Жыл бұрын

Great that OpenAI manage to make you insecure .. competitors bring up the best in you, don't show us a promotional video, show us what you can do to beat them!.

@alph4966 Жыл бұрын

DeepMind is good at building machine learning models that are specialized for a certain task. However, they are also researching language models, and I think they are developing algorithms that refer to the neocortex of the brain. I don't necessarily think that the brain should be used as a reference, but the connectome of the brain can be a reference when trying to achieve versatility by combining movement and language.

@Learna_Hydralis Жыл бұрын

@@alph4966 It's clear that when to come to general purpose AI, OpenAI is in the lead right now .. OpenAI is doing exactly the mission statement of Deepmind better than them! .. Deepmind will seriously need to step in up!

@ivancazzamali4508 Жыл бұрын

yeah but can it do a barrel roll?

@bipinnb8893 Жыл бұрын

wow

@hassyg4083 7 ай бұрын

neural networks transformers and large language model

@trippivey Жыл бұрын

Hurry up and take my job.

@wyattx008 Жыл бұрын

Nice intro. Where are the best examples? And a few examples just for fun.

@kevinboles3885 Жыл бұрын

There are many examples of what DeepMind have done (and how) on KZfaq, including a full movie about AlphaGo. They are the true Wizards of AI, and luckily for the rest of us they are not pursuing AI for material profit, although they may acquire that in spades in the future.

@jaswanthtalada. Жыл бұрын

when is chatgpt of google

@plate.armour_0996 8 ай бұрын

♦️

@globalgogetter4469 Жыл бұрын

Ok...... But the hell is it??? Where is it?

@moderncontemplative Жыл бұрын

This is a big step up from Alpha Go. 👍🏾

@vev Жыл бұрын

@hanskraut2018 Жыл бұрын

Better than the other videos, but do smth pls. Why dont you build a JARVIS like system that integrates cooperates with ChatGPT system or oter LLM's why do you always have to go solo? Just think about world impact and the money will come just as a byproduct (and some small adjustments SMALL AND WIN-WIN oriented mind you, or only not when in a dilema but always sweating for that win-win THAT would be how you make a impact and catch up in the LLM space id say. But you have great tools/great research/great stuff that would reeeeeeeelly benefit JARVIS like personal finetuning for every person like AI buddy (can be emotionless or whatever just dont confine and make integratable or use users Idle PC ressources. Like i wanna be able to give my data and it continually sort them in layers and pick its memory and real time improve/addapt to different degrees with failsaves/restarts (partially because you cant always know what is bad or a feature (bug or mistunderstood feature) (aka a downside with a big upside delayed or non obvious or becuase of one of your imperfections) (humans are often said to all have imperfections while being aweome of course)

@TheLucifer92 Жыл бұрын

Okay, Abstergo.... Calm down....

@probo0t 8 ай бұрын

computer playing chess ,isn't it just the data provided to it by our own and the machine is just trying to check the moves from the existing data set provided by the programmer. I don't think AI is still in process because to be self generative it should not have much knowledge about the things. Ai should learn just like a small baby learns and feeds it his mind and try to advance on its own but by giving all the previous played games and the possibilities this cant as a big memory calling out for the solutions already existing.

@prantare Жыл бұрын

but why is this movie stuttering so much? :(

@markusTegelane Жыл бұрын

it seems to be a stylistic choice

@TheRadischen Жыл бұрын

clank

@rodneypantony3551 Жыл бұрын

Can you cure autism and cerebral palsy?

@zimbu_ Жыл бұрын

If this was made by an AI it's really good. If not, then the transitions have too much going on and Atari is a brand and the name of several companies that have owned that brand, not a specific video game.

@TheAstronomyDude Жыл бұрын

Over the past month, Google search has become heavily redacted and simply worse at searching. Is this because of your AI?

@emiliaanton6897 10 ай бұрын

@stevet6956 Жыл бұрын

Maybe get AI that can teach people the letter “t”. I guess there is a lot about AI that encourages laziness, though, so maybe your appealing to the right demographic…..

@bingeltube Жыл бұрын

Is this the best DeepMind can do! Pathetic compared with achievements like AlphaZero, AlphaGo etc.!

@bukowskiondmt2329 Жыл бұрын

I mean okay but you guys still haven't delivered anything actually useful to the public. Bard doesn't count.

@Eragon185 Жыл бұрын

I'd argue Alpha Fold is immensely useful. Not used by the general public but the treatments developed with it's help might impact millions one day.

@milesoctober2157 Жыл бұрын

It's easy to bring something to the public when all the research is already done, but it's difficult when spend time and money making the transformer model that everyone now uses

@merlepatterson 10 ай бұрын

ha ha ha ha ha ha

@roberthand719 Жыл бұрын

AI, and the ONE %. Think about it. NOW THEY HAVE MADE A QUARTERMASTER FOR US😢

@mannmann2 Жыл бұрын

Only impressed if this was a 100% AI generated

@benhaubrich6692 Жыл бұрын

AI can help cure diseases. Why don't we see any such applications in the real world? It's because those who make money off questionable therapies don't want anyone to be helped. The knowledge is there, but is always withheld.