Комментарии:
So with chapter no. 2 (Shor) what you are saying is that a future iteration of an AI model will in fact be way over the point we are too afraid to admit it might be?
Oh hey Rocco, didn't see ya there. How's it going? Gee whiz, I sure am happy to see you!
Please stop saying "Shor's Algorithm breaks the internet". It's simply false, and always has been. At the point Shor discovered/invented Shor's Algorithm, there were already at least two "quantum safe" encryption algorithms, and several more have been invented/discovered since then.
While Shor's Algorithm is surely one of the great intellectual achievements of the last 40, if not 50, years, it doen't "break the internet".
I was expecting the "Arithmetic 3-Progression" lower ceiling to be included here as well - as it is in your "Math: 2023's Biggest Breakthroughs" video.
ОтветитьSo can I ask this person to explain where and how he came up with this new algorithm then I will tell you how much more there is to gain and how we improve
ОтветитьI believe that the reason AI has stagnated is because it is still too closely related to mathematics to create truly emergent behavior. The universe we exist in has many layers of emergent behaviors which lead to our existence, and the specific existence we find ourselves in is subject to the laws of the universe. However the universe we have created for robots is essentially one where its main purpose is to create a mind, starting at our equivalent level of atomic physics. We try to use logic gates in large enough combinations to discover a mind. Instead we must first let individual components of a mind develop much like organelles in cells and then allow cells to combine to create a more general system. For example reasoning cells which control how well a system can logically deduce facts about the universe.
I believe that this could be done through an intensive training program where logicians are tasked with judging the work of a reasoning bot until it hones in on truly valid logic and eventually pushes the envelope. Allow the bot to speak in the language of discrete mathematics and it will come to understand the significance of its existence.
The idea of harmful emergent behaviors is quite humorous. You might be tempted to think they mean Terminator/Skynet level harmful, but what they almost always mean is something they don't like, such as an AI proving that some popular political point is clearly wrong. If you hear the word "harmful" think politically harmful, not actually harmful.
ОтветитьCongratz to the minds behind this breakthroughs
Ответитьq visaje (what a time to be alive)
ОтветитьStop teaching the computers
ОтветитьShor seems not happy with someone improving his work, lol
Ответитьshors algorithm uses one dimention, the newer one uses multiple, so why don't we go higher? into higer dimentions like the fourth and fifth, computers don't understand that we cant understand them vissually
ОтветитьPlease explain to me: How did a girl and three fish lead to "finding Nemo"?
ОтветитьCan anyone explain to me how hyperdimensional computing is different from previous large neural networks? The video described using high dimensional vectors to represent concepts, but I didn't see anything that was different about that vs the way we embed words/images in past neural networks.
ОтветитьAi doing ai shit 🤓
Ответитьabbas rahimi sounds just like borat
ОтветитьThanks
Ответитьhow come the blockchain is never mentioned ? it has entire new sectors of computer science
ОтветитьYour choice of topics is questionable at best.
Ответитьno way prompt engineering made it to top achievements of 2023
ОтветитьPhew, better info. Cannot deal with...the rest of info...
ОтветитьNice video, thanks :)
Ответить"we can reduce the carbon footprint"
omg, what is science down to.... research is about breakthrough, not eco-propaganda. "reducing carbon footprint" is not a scientific achievement
Maybe in 2024 we will learn how to exit VIM
ОтветитьAnd so things feel quite unpredictable right now, but I don't think the'll always be unpredictable!
- Said the woman hopefully
The movie is a movie about a man who is a man!
- Low complexity AI
I can’t stand anymore the excuse of reducing the carbon footprint of LLM: 1) the absolute number of energy consumption is minuscule compared to the rest of consumed energy in the world; 2) it has mostly happened in computing farms powered by solar energy , so it’s actual carbon footprint it’s almost 0 because 3) they not were installed exclusively to train AI systems rather than the rest of cloud services; 4) with the energy consumption one must also take into place the exponentially large number of algorithms that big AI models are replacing. And 5) CO2 is the least of the problems , focusing only on CO2 is a hoax originated by BP , as combustion ( fossil or not) also creates thousands of toxic and carcinogenic chemicals that we directly breath , and create chronic subclinical inflammation that is the root of most of the nowadays illnesses . So let’s stop at least combustion cars , and then it will make sense to revisit the ridiculously low carbon footprint of training AI systems.
ОтветитьI like this AI breakthrough; what has been weird to me about traditional AI is that it doesn't think in abstract ways
Ответитьwhat is needed is a model building program that takes existing data and randomly inputs that data then analyzing the results in runs . A sort of bootstrapping . The model would have "related to" and "how" related to links. Just guessing tho ! Once a correct predicting model is found use it on other data to discover new outcomes
ОтветитьWhat a year and what a time to be alive!
ОтветитьBackground diagram looks oddly similar to Kabbalah
ОтветитьBreakthrough biggest jump in computer history, Massproduction of Laser-Chips ONN (Optical Neural Network) in a few years. 2000 Times faster and 100 times less energy consumation. Prototyp allready build and testet from the FU Berlin and the MIT.
“Building on this know-how, we adapted and optimized the VCSEL for use in an ONN chip,” says Professor Reitzenstein from the Universtiy of Berlin. The aim is to achieve the highest possible light output and to bring the arrangements of 25 VCSELs on a chip segment to exactly the same light wavelength.
According to the scientists, initial tests with the laser chip ONN were very successful. The result: The new chips are 100 times more energy efficient and can provide 20 times more computing power per area than the best electronic digital processors. And that is far from the end of the story. “By increasing the clock frequency of the lasers, which is relatively easy to implement, these values could probably be increased by a factor of 100,” says Reitzenstein.
The extremely powerful but also extremely economical chips could also be used in smartphones and other mobile devices in the future.
This could happen faster than expected. The knowledge gained must first be implemented industrially. But because the components of a VCSEL system could be manufactured using manufacturing techniques already in use today, "we expect that it can be scaled up for commercial use in a few years," the scientists write.
So it seems the chip industrie is near a newstart. Maybe we will see that nvidia in 5-10 years (1400 Billion worth now) not longer be important, maybe only a licence producer of ONN-chips. Becouse the University in Berlin and the MIT have developed the first working ONN-Chip.
20 Times faster then the fastest current chip and the Professor from the Berlin University (which have solved the biggest problem ONN-chips had (Optical Neural Network chips)
had have, the small Laser) that it is very easy to make the laserfrequency higher so that the 25 Times faster chips (fast then the best normal current chip) can get easily up to 100 Times faster (20 x 100 = 2000 times faster then the fastest current chips) and that with much less consumation of energy up to 100 times less energy consumation.. It would be the biggest jump in computing power in human history. 2000 Times faster (I Read 25 times faster too so 25 x 100, so maybe 2500 times). Similiar fast normal chips we would normally see first after (with the current speed of increasing , moores law) around later then the year 2040 (means NVIDIA when they would go one in the speed in developing normal electronic digitalprozcessors they would reach the speed of ONN-Chips first after 2040 but would have a much higher energy consumation). It was said so in a few years we could have Smartphones with ONN-Chips that can run programms for that nowaday supercomputers are needed.
So it will be a newstart of the industrie. Maybe many manufactors (e.g Intel, AMD, Infenon, TSMC, Nvidia, Qualdcom, Samsung) which are producing the same ONN-Chips in licence. Maybe, Intel, AMD, Infineon, TSMC, NIVIDIA.
Means different manufactors that produce the same ONN-Chips in license means same chips but different brandnames, that could mean, that the profit is going down.
But, AMD, Nvidia and Apple dont have the experience in producing their chips themself, they design it and let it produce from TSMC.
And the race, better games need better CPU grapicards would find a end. Maybe with ONN-Chips the game industrie will need 10-20 years until the have games which use the whole computing power. E.g fastest graphic card at the moment 80 Terraflops 2000 times faster means = 160 Petaflops means maybe around 2027 (maybe earlier maybe later) a Laptop with a ONN-Chip of the size of the chip of a geforce 4080 would have a speed of 160 Petaflops (with 100 times less energy consumation) thats around 14,5% of the 600 Million Dollar expensive 1,1 Exaflop fast Frontier supercomputer.
Having read papers about it, emergent behaviours from large language models can (also) be caused by metrics (tests checking the model capabilities) which are not linear but binary. So actually some emerging behaviours are not really emerging, but they are noticed only after “a while” because the metrics are binary.
Although, as a matter of fact, it’s still not accepted as an universal answer to this behaviour.
Nero is real and painful at times
ОтветитьGetting knowledge from hyperdimensional vector calculus strikes me as voodoo. It's not logic as we understand it. It's like the kid who solves a complex math problem -- but didn't show his work. So we get this vector calculus to "compute stuff." What have we done? it's shooting in the dark.
ОтветитьThat's a bit illogical. There are fallacies going back in science which distort the logic. So relax and re-arrange the logic and you will be fine.
ОтветитьI will study computer science A-level and therefore can only contribute with a somewhat ‘witty’ remark that may arouse offence in regards to the binary number system; I writing this message because I can even though that sounds a
‘Bit’
Off 🙁
I know i suck at reading papers, but i wish newly published papers are like this informative and easy to understand 🥺
Ответить"AI can't reason" - shows article from 2018 Really? Lol
ОтветитьEmergent Behavior in AI is so fascinating. How an AI can just develop something new even though it was never trained in it specifically is amazing. Obviously harmful emergent behaviors like harming humans would be a bad thing, but imagining that one day a massive model might have consciousness emerge by accident with no one on Earth knowing it and seeing it coming is wild.
ОтветитьI hate being a computer scientist right now. Everything people talk about rn is AI. It is so boring
Ответитьyoud think these hyper smart and educated individuals could figure out how to setup a decent microphone
Ответитьthe fact that you overtook the accent of the first professor with subtitles, while combining it with an utterly irrelevant approach in a year of AI, while simultaneously explaining the most basic of things while not giving any details about the actual research (in both this and the math video) makes me severely suspicious of the quality of journalism at play here
ОтветитьYeah what the real rub is with this Quantum computing that no one is stating is that Quantum computing is absolutely essential for AI to be able to think and create and be able to run thoughts of possible outcomes to scenarios, it's the superpositional aspect of it that enables this, we're all hooked into the universe directly into the machine itself at a Quantum level!
ОтветитьI think by focusing and further delving into the intensive and extensive studies of crystals, elements, chemical science and compounds plus the scouting of other components that pertains to it as combination we can in our own effort and research studies can come up with the materials needed to propel ourselves to the better future and have that capability like that of UFO space ships. UFO's have the capability to instantly by speed appear and quickly disappear more than the fastest aircrafts of this generation and UFO's have that speed more than Mach 5 to Mach 15 or faster than that, a Mach is faster than Supersonic speed. UFO's have that technology to travel to space and enter that gravitational layers of a planet without burning by its material components alone plus it's cloaking energy, UFO's according to those with expert studies of that can travel to time using it's binary equivalent of it's shape and have different shapes and components from a particular space habitat, UFO's have tech capability to detect space energy shortcuts and uses anti- matter technology to travel to space and uses antigravity technology to lift up from any planet towards space countering the gravity pull of a planet to propell upwards, and have the capability to condense it's crews scientific mass inside like the dust particles re- microthings adjustment inside accommodating many of it's crews inside like tricking our eyes by the size of it's shape wherein we can guess for example that the crew each UFO spacecraft are like 5 or 6 in its medium size UFO almost the size of a fighter jet but actually by advanced science the space inside is large enough to accommodate many of them by science manipulation could be by reverse nanotechnology or structural shrinking method of atomic structures or by energy particle shrinkage and lastly they have the capability as an other dimension portal for object transfer to another planet by superposition and they sourced their energies from stars energy like that of Pulsar energy scoop only with rod thing extended for energy harvesting using electromagnetism similar to nuclear fusion technology. Now, let's first focus in its components. It's about Quasicrystals. Quasi which pertains to resembling to or with some property of this or that element type like how compounds capability to break into other forms as well as combine to form elements, study chemical science of elements and compounds because elements cannot be numerically nor property be broken into smaller parts but compounds it's other related branch can do that be broken into something or be combined into something. But the crystals plays a major part of its component because crystals in its different forms and varieties and colours each have it's own uses and functionality unique from each other because a planet can be formed from gaseous elements as well as each element comprise of each colour that when gas factors form from insoluble to soluble become dust to hardened materials of it's elemental type and colours as prism are lights with many colours that when solidified from transformation of insoluble to soluble with dust particles combination become precious stones and crystals when deposited to cold temperature areas and there are plenty of crystal varieties and according to expert studies and previously hidden but discovered facts by experts there were secret crystal made cities inside the icy snows deep down in Antarctica as well as somewhere hidden in the North Pole probably by guess only in the inside of the hidden icy snow mantle deep down the surface ice which under that mantle to whatever numerical layers lies a vast cold ocean waters brought about by the sinkholes. Remember the baby Superman crib was made of unique crystal, and they habitate in a crystal of castle technology to be shielded from it's opposite energies and cyber attacks and they communicate using hologram through elongated crystals. Study the element combination, each and every crystal functionality, dig deeper in compound studies and it's combinations, particle intensive studies, chemistry science and physics science and also study light particles particularly Photons because that's another tech the extra terrestials used for space travel to speed them forward like laser beams something like that, another is that light particles can be used as superior non physical data storage aside from other science knowledge. Don't waste your time in over partying this holiday season specially if you're one of the pillar nerds of tech wizard of your company. Spare some time for studying, you might be stepping into a future formula within your inherent gift or talent.
ОтветитьAmazing and Blazing 😍💪
Ответить