To try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/AnastasiInTech The first 200 of you will get 20% off Brilliant’s annual premium subscription

@internetnomadism

10 ай бұрын

Love 💕 your Vlog, thank you for all these great insights. It’s so astounding how fast AI is improving.

@allstargoldnuggets1309

10 ай бұрын

I mi

@allstargoldnuggets1309

10 ай бұрын

Look at te ancient Mayans computers c Ips

@allstargoldnuggets1309

10 ай бұрын

AI🥷🏿

@Stopinvadingmyhardware

10 ай бұрын

Photonics can be stacked. Which traditional transistors can’t be stacked as effectively. By being stacked, I mean there’s no true limitation on how many are stacked.

Yes, with photonics. The future is definitely bright!

@martiddy

10 ай бұрын

Badumtss!

@abhialt

10 ай бұрын

Don’t make light of it

@Sulayman.786

10 ай бұрын

Bright spark!

@Cmm4626

10 ай бұрын

Bro 😂

@wackalaca

10 ай бұрын

😂true that

For me the most mind-blowing part is that it’s possible to multiply the chip throughput by performing parallel computations at different wavelengths! This is just AWESOME!💗 P.S. Amazing to watch you from startup channel to PRO!👍👍👍

@AnastasiInTech

10 ай бұрын

Thank you for the constant support! :)

@revimfadli4666

10 ай бұрын

Imagine scaling it further using quantum photonics. Then using NN robustness to mitigate the unreliability

@SpiritualGangs

10 ай бұрын

​@@AnastasiInTech cool

Super interesting and so well presented. It has always struck me as amazing that when a bottle neck appears, people find ways to remove the delays, it is almost as though we live in a game where the designer provides obstacles to progress, but also work arounds that are are hidden and difficult to find as though part of some great challenge to test the brightest living minds with each step taking us all closer to some great enlightenment. We are blessed to live in these extraordinary times. Thank you for sharing!

@danieln7777

10 ай бұрын

I really like the analogy you made there

@fredfrond6148

6 ай бұрын

You didn’t know?🤔😂😂

What a time to be alive

UGH im soooo glad this is the right path i chose, i saw photonics chips as the future 2 years ago when i started getting deeper into my physics bachelor, and now in my master im focusing on optics and solid state matter physics. my dream is to build a company based on any of the ideas, the easiest would be a company that designs chip circuit plans to sell to fabs or companys to get them produced

@zs9652

10 ай бұрын

If you are starting a company one day I would love to check it out. That seems like something I would want to help support one day.

Bright idea. There might be a way to micro prism each colour in the spectrum for data carrying. Maybe infrared, ultraviolet, and others could be utilized? Facinating new data transmission discovery. Cool video well explained! 👍

You seem more self-assured in this video; and I'm happy to see it, because you really know your stuff, and are great at explaining it. Much respect, and thank you!

@user-io8bm6gz5z

10 ай бұрын

simping lol

@pokwerpokwerpokwer

10 ай бұрын

@@user-io8bm6gz5z Nah. Just encouraging a fellow human being.

@user-io8bm6gz5z

10 ай бұрын

@@pokwerpokwerpokwer

It is always important in any form of innovation to step back, create something in a completely different way, and then test it to see if it works better. Awesome to see that things that would seem futuristic are already being developed

@oryxchannel

10 ай бұрын

can't wait to point a science AI at this video when we're going to study videos with AI and not just summarize them.

Wow, what a great episode! I love the detail and your explanation makes this fairly complex stuff easy to understand!

@AnastasiInTech

10 ай бұрын

Thanks John !!! :)

@williamgidrewicz4775

10 ай бұрын

Perhaps in the future, they use various types of masers in quantum computers.

A very interesting matter. This development might accelerate photonic quantum computing as well. The "infrastructure" looks very similar, only are used instead of wavelengths entangled photons.

I really appreciate the longer videos, hope to see more of those in the future!

I'd love to see brilliant deliver their ai modeling and photonic chips fundamentals classes in VR. Secondly, I wonder how long before they integrate photonic chips to quantum computing.

@rokpepeshogun

10 ай бұрын

What we get to know, might already be around for a while, papers and tech prototypes get not created once an article is made about it, it has been around for a while

@pixelfairy

10 ай бұрын

Why vr?

@louielouie9502

10 ай бұрын

@@pixelfairy The emersive experience is like non other. It's a dream come true for visual learners. It's the difference between talking about space and stars and actually going to space in a complete simulation. I was the student who looked out of the window in class wishing that somehow I can visit the lands ,cities and times my history teachers spoke about. Now Students can do just that in VR. It's still a new thing. Not everyone has a grasp of the potential to redefine how we learn.

@DJVARAO

10 ай бұрын

One of the two main quantum computing devices is actually a photonic device. Companies like Xanadu (Toronto) and Psiquantum (Palo Alto) are developing those chips.

Thanks for the education on photonics. I hope this can be scaled up quickly enough to meet the demand. I hope a mainstream company like Nvidia pick this up, so we stand a chance of having this technology available for everyone.

@questmarq7901

10 ай бұрын

Nvidia > available for everyone

@Leadvest

10 ай бұрын

Ironically optics manufacturing has a long history of being an opaque field. Just think about Antonie van Leeuwenhoek taking his lenses fabrication methods to his grave. The difficult part of what you're talking about isn't revolutionary computing, but restructuring society.

Could be a big thing. All depends on the size penalty long term. But for data centers a larger room will be a small sacrifice for lower energy costs. Also less heat could mean less failure and longer lifespan.

Trained here. The simplest formula to remember is this. Bandwidth multiplied by frequency equals information exchange potential. That’s a formula for fiber optic cable and equipment design. The internet has demonstrated that the available bandwidth of our light spectrum is almost unlimited by our normal numeric standards. As for putting it all in a micro chip. Stand back.

Photonic chips is the futur of computing, not only because of less energy used, less heat produced and faster speed, but also because of parallel light frequency used simultaneously. And I don't even talk about Photonic quantum computing at normal temperature.

After watching couple of her videos I have concluded, She is smart, Looks like a super Model, Most probably works a job that she Loves and Now building a successful YT channels. 😭 😭😭😭 she really figured everything out.

@samkostos4520

10 ай бұрын

She can speak multiple languages. She's amazing.

@thomasmaiden3356

10 ай бұрын

@@samkostos4520 ZoeTheRobot agrees.

Amazing! The future is bright!

very interesting as someone who has spent years building pc's and doing networking its good to learn how it all works photonics has a lot of potential as for the size of the chips they will shrink but given the light nature of the tec i could easily see distributed clusters of chips using fibre as an interconnect add multi mode transmission and you have the capability to prosses vast amounts of data even over a larger physical area. thanks for the vid look forward to more.

I really appreciate this, Anatasi. You didn't use any of the jargons as a crutch and actually explained things clearly at a level that we can undrestand. It got me thinking about the photonic chip as a potential replacement for the current chips we have right now.

Was looking for an intro to optical computing. This was perfect Anastasi. Thanks!

@AnastasiInTech

9 ай бұрын

Thank you !

I guess these will have to do until we invent spooky logic chips which use quantum entanglement to transfer information. Now THAT'S fast! Ciao Bella!

just signed up for the RL course of brilliant, thank you for the discount and great video 💖💖

Insightful, thank you Anastasi In Tech. ❤️👍

Wow, I understood this a lot better than I thought I would. Excellent teacher and your sponsor should know that I actually watched your pitch. /salute

this is epic thanks for the information

I love it. I think you are spot on the light chip. The main bottleneck for existing chips is heat dissipation. The light chip will be virtually no heat or limited and it will have the color rainbow to transmit data because of different frequencies leading to parallel computing.

Great video Ana 🎉

Never ending new stuff to learn about. Love it 😁

Thank you for fixing the mic issue. Now I can listen.

The more i watch the more i think Anastasi is AI. She is too perfect to be human.

That is a great video! Simple and easy to understand. Plus, your accent is AMAZING

Thanks for the info)

Thank you again for the way you explained exciting and scary at the same time p.s. your English is getting really fluid meaning really good 👍

Thanks again for the very interesting information!

wow... I want to know more about photonics. thank you for this.

Very exciting indeed! thank you @AnastasiInTech

Can't wait to buy one!

Thank you for all your hard work. This may be unhelpful, but, from a exploratory perspective, regarding fabrication, I'm excited for space mining to accelerate. Can't help but think there are materials, undiscovered, up there, that will supercharge at lot of these ideas that we as a species, are innovating towards.

Thank you Anastasi!

@AnastasiInTech

9 ай бұрын

Thank you ))

Nice information

Thank you.

This optical information technology is really interesting, you can absolutely go in depth the subject ❤❤❤🌹!

Brilliant!

This vid felt even more elegant then usual ❤

Thank You ... and .... looks nice 😊

Please explain the comparative advantages and disadvantages of photonic vs quantum chips/computing. Which uses less power? Which is fastest? Which can be smallest in terms of physical footprint? Can they both be used in a symbiotic conjunction with each other...and along with conventional micro-transister chips? It's so cool that Asimovs AI robots had photonic brains...the guy was speculating about this tech in the forties!

Awesome video. Photonics seems to have unlimited potential. You make those videos so incredibly interesting & entertaining.

@AnastasiInTech

10 ай бұрын

Thank you !

"in situ" means "on site" and to convey that meaning, the word "situ" cannot be used without "in". To put this another way, "in situ" must be considered a single word, which cannot be broken apart.

That’s very cool!

With only 4 minutes into the video it appears to be somewhat of a hype, but you asked, and I am overwhelmed with your appealing talent and beauty. My opinion will probably change as I watch you explain how different wave lengths of light can work in photonic chips.... ((((some time later)))) I watched it and I am still not convinced. But watching and listening to you is still a bit fascinating.

Never ever thought about the color band communication, def looking forward to that video. Maybe an arduino example :-)

Anastasi, thanks for this video, I have been watching photonic computing for many years. When do you think photonics will disrupt GPU companies like Nvidia?

@techpriest4787

10 ай бұрын

You do not disrupt Nvidia. Soviet Nvidia disrupts you.

@billkemp9315

10 ай бұрын

@@techpriest4787 "Soviet Nvidia"?

@techpriest4787

10 ай бұрын

@@billkemp9315 son, not knowing memes is a blessing and a curse. Meme History: Something cold war era origin. A phrase that suggests that only the Soviet Union is in power. It serves to point out the totalitarian nature of the Soviet Union even over its own people. So. You do not ask the questions. Soviet asks the questions.

@billkemp9315

10 ай бұрын

@@techpriest4787 OK I guess but how did the Soviet Union get into this topic? Why? Seems off-topic to me.

@techpriest4787

10 ай бұрын

@@billkemp9315 it is a joke also. Though a joke I actually should not make because I am not even one of those who think that Nvidia is totalitarian. At least not more than large corporations usually are. However. It was also referring to Nvidia's ability to innovate itself. But more so it was implying that Nvidia just like any other corporation would just buy out such photonic tech companies. Seems photonic tech can not function without being a hybrid with silicone transistors. Which may mean that they can not do their own thing and rival Nvidia/AMD.

Sci-if stories like to use the term “positronic brain” but perhaps they should say “photonic brain”!

@DonaldKronos

10 ай бұрын

I think the reasoning was that positronic sounded more futuristic, since we had lots of technology that used photons but not much that used positrons.

Wow !!Tech step up!!

Díky!

@AnastasiInTech

10 ай бұрын

Thank you :)

She's like a dream. Beautiful, successful, intelligent. I love her.

There's one hidden half-truth in academic papers that quote "accuracy" with the existing datasets. When you test it on data that's not in the training or validation dataset, often the error rate is over 50%. In many cases, the practical error deployed in the real world is over 60%, which means it's not really usable. We still don't have the right loss function and no practical way of verifying the model actually generalized like "believe".

Hi Anastasi is there a difference in between this technology and the quantum photonic chip you talked about in a previous video

Great to see "extreme UV" ASML joining the party in order to address the feature size issue. One saving grace with optical, however, is the very low power dissipation, and it's power dissipation which dictates maximum volumetric density of conventional electronics. Thus we can pack photonics far more densely than electronics, and that will mitigate to some extent the feature size issue.

So then using loops and maybe materials they are able to slow light down or let it through unguided to speed it up, and thereby use the delay which is infinitesimal, not only as zero and one but easily becomes useful in quantum computers. With a grey scale between zero and one based also on statistical reasoning and A.I. and fuzzy logic systems as part of expert systems. Excellent video. I just love your presentation.

Photonic computers have another interesting feature. If the clock frequency is very high, the hardware has to be small to avoid signal delays. A frequency of 1GHz, corresponding to a period of 1 nanosecond, allows a light signal to propagate only 30 cm.

@danielmurogonzalez1911

10 ай бұрын

Hi! Could you elaborate more about the implications this interesting feature? I am not to familiarized with this physics

@LindsleyDbrt

10 ай бұрын

@@danielmurogonzalez1911 The electronic signals on a chip cannot use very high frequencies, as the high density of the circuits would generate so much heat that it would be impossible to dissipate it properly, burning the chip. But photonic signals do not have this limitation because they do not generate heat in the same way, allowing the use of much higher frequencies. A signal switched at 1Ghz has a period of 1 nanosecond. As the speed of light is limited, in that time it travels only 30 cm. Theoretically, to synchronize all signals from a photonic computer, hardware distances greater than this could not be used, as data could be lost. If the frequencies were even higher, the distances would have to be proportionally smaller. If the frequency is 100 GHz, for example, the light travels only 3mm.

@unbiased3949

10 ай бұрын

@@LindsleyDbrt Your understanding is partially correct. The speed of light in a vacuum is approximately 300,000 kilometers per second, or approximately 30 centimeters per nanosecond. This means that light can travel a distance of approximately 30 centimeters in one nanosecond. The clock frequency, expressed in gigahertz (GHz), indicates the number of cycles per second. For example, a clock frequency of 1 GHz means that there are 1 billion cycles per second, which corresponds to a period of 1 nanosecond.

@LindsleyDbrt

10 ай бұрын

@@unbiased3949 I´m confused. Where is my error?

Amazing!

Saya ikut support untuk cenel anda di KZread saya warga negara Indonesia 🇮🇩 Sekarang berada di Penang Malaysia sebagai perantauan, walaupun saya tidak paham dengan bahasa Inggris tetapi saya suka menonton berita berita di KZread semoga anda sukses selalu dan selalu dalam lindungan Allah SWT Aamiin 🤲😊

well we had a good run boys

Miss could you please tell what are these filters and light detectors used in photonic processor, how they function what material they are of

Great review of what is coming. A review of the newly announced Tsingshau university opto-electric chip vs MIT opto-electronic chips. Would be appreciated.

so interesting ! I thought photonics can only do inference. Do you think Lightmatter would pick up this technology ?

@davidstar2362

10 ай бұрын

our,thoughts runs on light

@AnastasiInTech

10 ай бұрын

I think so! I will ask them ;)

Asynchonous or "no synchronous" is an important aspect to consider. It can say compress a century of high speed sample data to produce and "answer" in a bit or a byte or bytes. It allows so many assumed rules to be discarded or alleviated towards progress to quality of information in efficient time domains. To include "feedback loops" to both sides of the underlying way a program can re-program itself, while including ways to incorporate "vectors" in time and in space. Any mix/match use of these concepts at any level will drive efficiency, and produce answers to create better questions. It is better to know an answer before one need consider the questions. Gr8! Peace ☮💜

This is really significant and cool. Training is actually the bigger energy waste compared to inference. This kind of thing could help to democratize AI models better for individual researchers & consumers instead of only being affordable for mega corporations. Well, first the photonic chips have to reach more mass adoption for that, but it does look kind of inevitable.

@robertweekes5783

10 ай бұрын

And Nvidia just released mainframe GPU stacks that are twice as fast and 3X more efficient. The AI apocalypse is approaching fast 🤖💨

@JustSomeDinosaurPerson

10 ай бұрын

@@robertweekes5783 Good, let it come and bulldoze the world. I welcome our AI overlords haha

Brilliant, ❤ it

I've been waiting for a full photonic computer for years , I can't wait to own a home pc run on light , or wavelength

I would be interested to see a version of this with the Forward-Forward Algorithm.

Vamos dominar esse mercado.

7:12 Back propagation, is when you've let it make a prediction (which is given as a probability, a bunch of potential answers with a % that shows how sure it is that it's the right answer), then you run it back to all the wrong answers it gave (eg. 20% certainty that a camel is a giraffe), then tell it, "Use this image as reference for what is NOT a Camel", an archaic way of training, that is very computation intensive. As for every right answer, there is a hundred wrong answers. So it ends up taking longer to back propagate than it does to simply run the next one. Instead, it should use it's stockpile of correct images, to find similarities that exist among all of them and start recording those as definites. Eg. A school bus definitely has wheels, or a human definitely has eyes. You can already tell why this is better for 99% of applications, but runs into pitfalls if you only use 1 definite. For example, what if a school bus is on bricks, or a human has lost their eyes? But these are missing definites, which are on an individual basis, not as a whole. So if the person has no eyes, but still has a nose, mouth, ears, then chances are, it's still a human, even without eyes. It's actually how our brains work, which is why people generally recognize a thin, long-haired person from behind as a woman, even though it can also be a man. But that's where tier 2 probability comes into play, "how likely is it that a man has long hair, vs a woman", at which point, you have secondary recognition phases, like identifying a human first, then gender, by looking at features that are unique to either, like boobs on a woman, or an adam's apple on a man.

I think using polarity and wavelength of light creates the fastest computers allowed by physics. I love this idea. Since wave lengths are infinite, so is our multitasking. We won't be limited by CPU cores like we are now. I am curious how the bit rate will change. Will we still be limited to a 64bit bus? Will we be talking about terahertz processing levels? Once photonic computers reach the physical size of our laptops, it will render the laptop as obsolete as the abacus. And to think that the laser generation chip is already 1x1 cm, and only needs 6w of power (5v x 1.2a = phone charger). Since the majority of the energy is in production and output, rather than processing, I wonder how much heat such a computer would produce. Would we need fans? If we can get away from fans, we could get away from any mechanical parts, which means significantly longer life. Does anyone know how much power the LCD screen itself uses on a standard 17 inch thin screen? I'm curious the power requirements of a photonic computer in a laptop form.

Would it be helpful to polarize the input and 90 degree different back feed ? Full spectrum both ways ?

I'm fairly sure AT&T has been doing this same thing for decades with fiber optic cable transmission. 13 pair of fiber optics replaced huge copper cables installed during the cold war. The transmission rates were astronomically faster and then they went with adding colors to isolate individual transmissions on the same 13 pair. This was in the late 80's. I patrolled those initial installations and talked with a lot of the techs daily. The tech has been around for a long while but not shared apparently. I could listen to you read the dictionary or phone book regardless. I won't state the obvious...

Size can be an advantage for photonics. The frequency of light, multiple wavelengths, parallel processing, could allow computers to advance while fabrication on the small scale is reaching its limits.

It needs a laser light source and the red/ir, uv and green can be made from semiconductor but for yellow i'm not so sure as it needs a ionic gaz of may it can be achived by a doped crystal and optical pumping witch can be achived into a ship maybe... :) I was playing with a MYRIAD-X days ago (an Inference accelerator ship) so i'm exited by photonic hardwares !

These chips sound incredible. I would like to imagine a home computer with this technology inside, but, it sounds like from the fabrication perspective as a retail consumer this could be 5 - 10 years away, would that be fair? Also, investing in these companies sounds like a good idea - I want to be a user of this technology a.s.a.p. 😊

Not only can you multiplex across wavelength, you can also multiplex phases states at individual wavelengths.

10:00 98% accuracy is not pretty good, it's excellent! - 70% is pretty good, 80% is good, & 90% is really good.

Given the low power requirements of photonics, it seems like a really good candidate for multi-layer stacked silicon packages, which could massively increase density, negativing the drawbacks of the larger silicon processes required. Thoughts?

... good get ... makes me wonder whether Tesla can combine their Dojo technology with the photonics processing you describe here ... and how much of it would be able to be captured and replicated at the individual vehicle or bot level for localized training and inference to supplement the centralized networks ...

This is the way

This demonstrates to me how vast and creative computational processes really are. If you have a powerful mathematical Model like machine learning you can try to find out how to optimize the Hardware first. Some computational models deal with some energy effiency on some computational Hardware and the relationship between these may be Important for a greener approach in computing

Youre so perfect , and your content is so interrsting Anastasia❤💐😊

Indeed, photonic chips have their advantages, such as in lidar, and in recording photos and video.

Amazing... What are yr news sources? Just amazing

Awesome

If infrared and ultraviolet can be used in addition to red, yellow, blue, and sensitivity developed to include orange, green, and purple, the possibility is astounding. We should have the ability to read even finer increments in color wavelength as we're very familiar with interferometry, and spectrometry in astronomy and other applications, so, the range of light wavelengths we could use should be quite expansive.

Far more than three wavelengths can be used simultaneously, perhaps, someday there will be nanometer wavelength breakdown of the entire light spectrum, yielding billions of individual computational channels.

Amazing video. Now get your priorities straight here and solve our issues with climate change.

Looks that ASML will continue being the key role player for the foreseeable future.

2:46 Low Losses High BW = OF yeih 😄 specially the monomode one

I remember seeing a report of back engineering a ufo where controls use photonic wiring

As I am sure, you well understand that many of us are fascinated by your considerable technical expertise along with your personality as well as other factors. Thank you for your professional explanations of complex technology - so we average people can understand!

Amazing, and a great channel, love from Pakistan