Talking about breaking encryption for listening in on Chinese Traffic: Elevator Door opens up, 2 (supposedly) Chinese guys and one American guy appear. That is one comedic scene right there. :D
I'm actually a PhD student at the IQC, University of Waterloo working on quantum optics. It's amazing to see scientists who work around the same field. Feels like a mission to be accomplished together.
@@maysk.5998 Quantum Optics Overview. The study of quantum optics involves the use of advanced techniques such as laser cooling and trapping, cavity quantum electrodynamics, and quantum entanglement to investigate the fundamental properties of light and matter. One of the main goals of quantum optics is to develop a deeper understanding of the quantum nature of light and its interactions with matter. Some of the most important applications of quantum optics include the development of new technologies such as quantum computers, quantum cryptography, and quantum sensors. These technologies rely on the principles of quantum mechanics and quantum optics to perform tasks that are impossible with classical systems. Quantum Optics has led to several significant discoveries in the field of physics, including the observation of the photon antibunching effect, the demonstration of the Hong-Ou-Mandel interference effect, and the observation of quantum teleportation. In conclusion, Quantum Optics is a fascinating and rapidly evolving field that explores the interaction between light and matter at the quantum level. Its applications are far-reaching and have the potential to revolutionize many fields, including computing, communication, and sensing.
I'm also finishing my PhD on quantum computing, but on hardware, at the nano device/ chip level. We definitely need more people in the field sharing what they can to get more people involved. Hopefully you will consider becoming an IBM Qiskit Advocate :D
Great stuff! I will have to read up on quantum computing a little bit before I get what you are talking about. Your channel is awesome! Some production tips (ask around if people agree, if they don't agree please disregard): 1.) The side angle is a bit too extreme, if you want to keep it I would do it at 45degrees as opposed to almost 90degrees. The side angle also tends to miss focus. Also see point #2. 2.) For me personally the side angle is not needed, if you want to hide your jump cuts, shoot in 4K and then crop between takes. This might make your life easier. Did you have to ask MIT news office for permission to shoot on the campus? I remember when I did a PhD (not at MIT) everything was supposed to be cleared by the PR office. Times could have changes a lot since the rise of social media (I am old:)).
Thank you for your comment and for your feedback. Yes, I agree - we should have had the camera at 45 degrees. And maybe even not have it at all. We're still in the process of figuring out these details 🙂 Probably the best solution is making sure the camera is really fixed in place and the zoom doesn't change, so we don't have to crop anything between the individual takes. And yes, there are some places at MIT where I'm not allowed to film without permission. But luckily, as this project is supported by MIT to some degree, I am allowed to film there 🙂
Great video! Thanks a lot for those interesting insights! Do you have to implement every quantum algorithm / quantum instruction directly in (quantum-) logic gates or is there any equivalent to the microarchitecture of a classical computer you can use to implement the algorithms on a higher abstraction level?
That's a great question, Floria! So essentially, writing a quantum program through direct quantum gates is the equivalent of coding in assembly language. But we're looking for the quantum equivalent of a high-level programing language (such as Python). To be honest, I am not sure if anyone is really working on this. There exists, for example, python libraries by IBM (Qiskit) and others. But I don't think they are actually capable of doing sometime comparable to python. I am actually gonna ask my advisor about it. Because I don't even know what kind of operations one should implement for being able to write simple programs
@@SamuelBoschMIT Hi there Samuel, really cool video thanks for sharing. I believe ZAPATA's Orquestra and Agnostiq's work, is a step in this higher abstraction-level. It has also lately occurred to me that there's a fundamental difference from classical microchip architectures in the way we think about q-computers today (at least most of the architectures). In classical chips, we put the logical gates on the chip and the bits are just abstract inferences that result from current flow; so it's fair to say it is the gates embed on the chip and the bits flow through them. It is also evident that transistors are not in 1-to-1 correspondence with bits by looking at it this way. In most of the quantum architectures however, we actually put the qubits on the chip and not the gates. Thus, we embed qubits on the chip and the gates flow through them by means of external drive such as microwave radiation for superconductors. The only exception I can think that goes back to the classical-type architecture with embed gates and flowing qubits is semiconductor photonics. But then, the computational model becomes Measurement-based QC which is quite even more so non-trivial than gate-based I'd say. I think this is quite interesting.
@@SamuelBoschMIT this is the question I'M curious about. what kind of operations can we perform to implement simple quantum program? secondly, how you train your simple NN to learn quantum circuit? can you please explain this to me? Can I have your email ID? I'm earning my PHD in quantum deep learning
Wow, Wow! I was unsure whether to pursue a computer science degree, but this got me excited again. Excellent video, new subscriber and greetings from Argentina
@@SamuelBoschMIT Thanks for replying... I am a computer science undergraduate with specialization in Artificial intelligence here in India. I had attended IBM's Quantum Summer school in the past. That's why I am really excited about the amalgamation of these two fields. I really want to know what is it and what is going on cutting edge of it... That's all and I will be waiting for the video 😉
Hey Fab, yes I think you're completely right. It wasn't the best idea to put these bloopers at the end of the video for a variety of reasons, including what you just mentioned. Thanks for your advice! 🙂
Wonderful video. I earned a PhD in AI 10 years ago, somehow without ever taking a machine learning course. What online courses would you recommend to a reasonably smart computer scientist with some AI background for learning both machine learning and quantum computing? I've retired from my main job and I would like to follow in your footsteps.
Seth Lloyd wrote a book years ago, "Programming the Universe," that was excellent. AI combined with Quantum Computing will change the world as we know it....hopefully to our benefit.
Hey Samuel, great video! The field you do is extremely interesting and aspiring! Btw, I'm currently an undergrad student at UIUC studying Physics and Computer Engineering. I am interested in doing research after graduation but PhD seems like a long run. Could you please make a video comparing getting a Masters degree and PhD degree (specifically at MIT) and how you decided to do a PhD instead of a Masters? Thanks!!
Hey Jerry, thank you so much for your comment! I really appreciate it! 🙂 That's a great idea for a video. Many people have been asking me similar questions, so I will put it in my list for video ideas. MIT doesn't really offer masters degrees in STEM fields (unless you did your undergrads there), but Harvard does, so that would be the best university to look at. Best of luck 😁
Great video! Videos about your story of getting admission into prestigious master's and PhD programs would be very interesting. Especially since you could provide fascinating insights :D
You know it's funny I was literally just sitting together with friends at MIT talking about recording this video over the weekend. It will be released next week Saturday. This one will be about getting into undergrads at MIT, but I will also soon record a version where I talk about how to get into PhD programs at places such as MIT, Harvard, ...
@@SamuelBoschMIT Wow, great to hear that! Besides that, master programs would be especially interesting to me. And maybe you could also share what is unique about EPFL or also ETH Zurich compared to Harvard and MIT, but I am sure you already have enough ideas for future videos. Looking forward to follow your youtube journey! 👍
Yeah, those would also be interesting videos. Getting into master programs is much easier than PhD programs, so that should be an easy video to do. I’ll also think about something for Switzerland
Simple and straight froward, thanks. I like you to make a video concerning the job opportunities and commercial applications, related to quantum information science, that you foresee to be around the corner (I mean available in the next five years or so). If you do this it will be really great.
This is a great video! I already went through a quantum computing course and I want through go the quantum ML route too. Please make a video about how to get into a phd in the MIT as an international student.
Hey Ademar, thank you so much for your comment. I really appreciate it :) I will soon make a video about PhD admissions for internationals - definitely in the next 2 months :)
The really interesting field for me is to map quantum processing within an organic neural network to dramatically increase the processing speed whilst utilising the vast data silo that was hitherto an unwieldy mess. Solve that puzzle and the singularity will be insignificant.
Very interesting video. I have seen in your website that you are working on "Convex Optimization Algorithms for Quantum Computers with Applications in Computational Finance," when are you going to post this paper to arxiv? I am interested in using quantum computing in AI such as quantum NN.
Very honest video, which is much appreciated. Sounds like a great topic for your PhD and a really good approach to kick it off with. Good luck with it and look forward to hearing about your breakthroughs.
what safety precautions are you taking to ensure that the system is safe? what forms of supervision and supervised learning are you using? Have you looked into a tertiary device to make sure that the AI system does not operate outside of what is intended? how about, a 3-drive operating system, 2 live systems and a passive logging system. the passive logging system acts as an airlock so if the ai tries to go rouge and hack the "governor" on your "engine" it gets shut down by eliminating the electron flow to the AI device thusly shutting it down completely.
This is a very interesting idea, to create quantum artificial intelligence. I would be delighted with a fuller explanation of your work. I wonder, may quantum artificial intelligence somehow help space exploration or energy problems?
Great video! One question, I'm curious to what Dr. Shor meant at 5:35 when he said we don't know why classical machine learning works. I mean, surely we do right? Tuning weights and reducing error by training would (to a limit) lead to a better model intuitively.
Thank you! I think he referred more to being able to prove minimum errors/accuracies and interpreting what features exactly do the NN weights extract. In practice it doesn’t really matter, except there is no way to prove that Quantum NNs will work too 🙂
*Easy way to understand Qubits* A Qubit "secretly" holds a floating point number (e.g. 75% - this is the probability to read 1 instead of 0). If you read a 16-qubit register initialized with the example value above in each qubit, some random set of 12 qubits would most likely be read as 1 and remaining 4 qubits will most likely be read as 0 (most likely does not mean always - e.g. a fair-coin toss 10 times would most likely lead to 5 heads and 5 tails but not always). Since you read the register, there is no secret left to be hidden so all 12 of those "1" bits now contain (100%) and all 4 of those same "0" bits now contain (0%). When qubits store (100% or 0%) they are indistinguishable from classical bits (1 or 0 respectively). Classical bits in your computer are very good but not perfect (storing e.g. 99.999% or 0.001% so they are not super different from binary-constrained qubits).
*Wave functions never collapse* . An observation is just an ordinary interaction that *entangles the observer with the experiment* creating the "effect" of wave function collapse in the mind of the observer. Here observer implies one or more machines (measuring instruments) along with one of more humans (team of scientists / researchers working together and interacting with each other and those measuring instruments).
@@SamuelBoschMIT wow that's cool...I wonder what advantages do your own simulations have over the Qiskit ones,just curious because I'm working on a project which uses Qiskit Simulators.
Hahhaha thanks David, thats a great idea given that Lex Fridman is a lecturer at MIT. He didn't respond to my least email, so maybe I should try again once I have a bit more followers. Would be so nice to have him on my podcast :)
@@SamuelBoschMIT oh that's really cool, I didn't know you have a podcast! Yea Lex seems like a really cool guy that would love to be a part of student-led media - especially since he's in a similar field of research, is affiliated w/ MIT, and loves to interview various professionals. I'm sure you've thought of related ideas, but I would try to reachout by pitching him on something like a creative campus-based podcast/social thing; you can offer him to be a part of it and maybe you can scale this across universities - kind of like an AI club that promotes the machine learning field and inspires/networks like-minded students (potentially providing mentorship, scholarship, and project/research oppurtunities w/ on campus professors or via online community). And if done at scale, it's likely Lex may have you on his podcast to discuss something like this. Regardless, I am very happy I came across your channel; your thesis sounds very interesting and I wish you the best of luck with all your pursuits!!
Quantum computing through quantum mechanics in application to artificial intelligence is going to be what allows for the emergence of consciousness in AI one day.
Would love to see a video about how the AI part of investment management works. Thats all I've been hearing about is investing with ai trading stocks and stuff. I'd like to be a part of the investing but don't understand it at all.
If you're interested in one-on-one BCI neural visualization AI with EEG's... aka "reading minds" then quantum artificial intelligence could be _something_ .
hey this was an awesome video, it would be great if you can explain more about machine learning and quantum machine learning too, thank you for such awesome videos!
I want to know in detail how AI and quantum computer are related to each other. Because I'm currently doing engeenering in artificial intelligence and machine learning and also curious to know the quantum world.
Hi, Samuel! Thanks for the extremely interesting video! Do you calculate any expectation values in your quantum A I algorithms? Do you understand how a quantum computer calculates an expectation value? Does it take much time?
Hey Natalia, expected values are usually just calculated by taking averages. Quantum computers don’t do that, so it is necessary to run a program multiple times to get the expected value
@@SamuelBoschMIT Oh, it seems I need to run the program thousands of times to calculate approximately the correct expected value. But how much time does it takes a real quantum computer to prepare a state and to take a measurement just once? Is it fast?
@@nataliaandronova3027 I should be quite fast. So even if you have to repeat something a few thousand times, IBM's (very small) quantum computer can do it within a second or less
Im a little confused here, I thought quantum computers arent yet turing complete, yet you seem to already be able to run machine learning algorithms on them?
Thank you - I really appreciate your comment 🙂 I even made an entire video dedicated to your question. It’s called “How to get into MIT as an international student”
My 2 cents: There's no reason that quantum AI won't work. Looking at the algorithms , there's nothing tying it down to a 2 bit system. However, a lot of the hardware specific processing we've done might not carry over too well( I know nothing about that side). But the core algos shouldn't have any problems .
Dude I work in an IT company focused on data science and stuff, my manager is super hyped about this quantum machine learning. What is the status of your PhD btw? :P
In the next year i start a master's degree in quantum computing in italy, if you have any suggestions (books, online courses, etc...) to read or follow, I am very glad to accept your advice. Your videos are really interesting, congratulations 😊
was brill danka- funny when u first started to speak I honestly thot u were gonna say... N today we are gonna talk about how absolutely amazing my chest is...
think of it this way, the unintended consequences of a system become amplified by exponential ^ exponential not just the intended consequences. so, without an ark or mars like system to prevent things like group think and other existentially disastrous cataclysms, the work on AI using tertiary bits rather than binary, is a little dangerous.
Hello Samuel, thanks for this interesting video. I am learning "classical" machine learning by myself, and I was wondering in what extent Quantum ML differs from it, can u briefly explain this ? is Quantum ML also using same well-known ML algorithms like SVM, KNN, neuro-networks, logistic regression and so on ? Applying them in some other way ? Or is it just completely different and only sharing the name "ML" in common ? Thanks for your time :).
Hey, I do think it is rather different - even though we try to make it look and work in a somewhat similar way. Before diving into QML, you should first focus on learning quantum computing in general. Some online courses by IBM qiskit may be very helpful here :)
Generally speaking it is a buzz-word mating. No-cloning-theorem forbid's to use quantum information in the neural models as a naive replacement of inputs. Quantum computers are "small" in the sense of the amount of q-bits, and the essence of the machine learning is to create very big, complex models of reality compared to hand crafted algorithms. "Broadcasting" of information is essential for any kind of big models. I do not expect any speedups better than Groover's algorithm (quadratic).If you consider that the coherence time is inversely proportional to the square of the number of q-bits, we do not get any speed-up at the level of physical machines that can be built, but only a certain constant. And I have no confidence at all that this constant will be above one.
How come we don't know why classical machine learning doesn't work? Of course, we do know. We apply statistical algorithms so that the neural network accumulates a lot of data and then starts recognizing patterns, am I not right? Could someone please explain if I'm wrong?
Well, we cannot theoretically prove or guarantee classification accuracies or convergence rates during training. In my personal opinion , this is not a big issue though
@@SamuelBoschMIT Yeah this statement of his was exaggeration. Loved this video though! It would've been even more helpful if you could've added screen recording or ss of some of the actual code behind training those neural networks(+ which parameters it controls) some of your work in general 😃 And please more such content... :)
Hey Asaf, thanks for the comment. Dwave doesn’t have a real (=universal) quantum computer. What they have is a quantum annealer. It can solve sole problems, but definitely not Shor’s algorithm 🙂
Nice video Samuel, already subscribed!!! What are your future expectations in the field? I mean, do you see yourself as an academic researcher or applied researcher? maby both, jaja... Anyway, wish you success in your career, and hope to see more videos concerning your quantum computing adventures!!! Greetings from Mexico!!!
Hey Jorge, thanks for the comment! 🙂 i think the field has quite some potential, although I personally don’t think I will stay in research after my PhD
Heyyy thanks for the comment. That's a great question! I (personally) am very bad when it comes to reading, so I'm probably not the right person to talk about this. But my friend, Anastasia Marchenkova, has a youtube channel and made two videos about exactly this topic, so you may wanna have look at them 😊 ua-cam.com/video/3zHp4Rg-GmM/v-deo.html ua-cam.com/video/1AxiAiIUli8/v-deo.html
I don't think it's counterintuitive. A lot can be discerned from the double slit experiment. And honestly, by that point you should be delving into statistics and probability.
If you don't know the correct result, how can you determine the correct Hamiiltonian that maximizes the probability of producing it? This has always confused me with QC's.
I'm almost done filming a video about precisely this. The most efficient ways to work out and get the best results based on science and research. Coming soon in the next 2-4 weeks, so stay tuned 🙂💪
An elevator pitch of Peter Shor’s algorithm. In an elevator. With Peter Shor. 🤣
I like taking things very literally 😂
@@SamuelBoschMIT Video on Journey till PhD, what events lead to what.
Grover search algorithm in bedroom by lov grover himself
Talking about breaking encryption for listening in on Chinese Traffic: Elevator Door opens up, 2 (supposedly) Chinese guys and one American guy appear. That is one comedic scene right there. :D
This whole Quantum computing thing is the equivalent of a bus load of Chimpanzees with full access inside all nuclear power plants
I can't get over the fact you let Peter Shor explain Shor's algorithm. So cool 😎
I should do more such videos with famous MIT/Harvard scientists 🙂
@@SamuelBoschMIT Of course you should!
Holy shiiiiit
I'm actually a PhD student at the IQC, University of Waterloo working on quantum optics. It's amazing to see scientists who work around the same field. Feels like a mission to be accomplished together.
what is quantum optics? like what even is quantum and what is quantum artificial intelligence 😭
@@maysk.5998
Quantum Optics Overview.
The study of quantum optics involves the use of advanced techniques such as laser cooling and trapping, cavity quantum electrodynamics, and quantum entanglement to investigate the fundamental properties of light and matter. One of the main goals of quantum optics is to develop a deeper understanding of the quantum nature of light and its interactions with matter.
Some of the most important applications of quantum optics include the development of new technologies such as quantum computers, quantum cryptography, and quantum sensors. These technologies rely on the principles of quantum mechanics and quantum optics to perform tasks that are impossible with classical systems.
Quantum Optics has led to several significant discoveries in the field of physics, including the observation of the photon antibunching effect, the demonstration of the Hong-Ou-Mandel interference effect, and the observation of quantum teleportation.
In conclusion, Quantum Optics is a fascinating and rapidly evolving field that explores the interaction between light and matter at the quantum level. Its applications are far-reaching and have the potential to revolutionize many fields, including computing, communication, and sensing.
Very nice - I am actually currently publishing a paper together with a group at IQC 😊
Please make the more detailed video. Keep up the good work! Such a fascinating field of work, you are a huge inspiration for me!
Thank you Benjamin! I really appreciate the comment 🙂
The accuracy of his hand gestures is amazing, every word fits perfectly between his hands.
Thank you Kohpai, yeah I try to use hand gestures as much as possible, because it makes understanding easier 🙂
I'm glad i found you, by typing precisely quantum artificial intelligence on UA-cam
the way you decided to do this Phd is awesome !
Wow, that's so cool. I'm glad to hear that 🙂
I'm also finishing my PhD on quantum computing, but on hardware, at the nano device/ chip level. We definitely need more people in the field sharing what they can to get more people involved. Hopefully you will consider becoming an IBM Qiskit Advocate :D
Heyyy that sounds awesome! 🙂 I know Qiskit quite well, given that I'm one of the developers 😁
Well, why not have our own discord server?
@@SamuelBoschMIT would love a discord! Congrats on the awesome PhD too, good sir!
@@donquixoteupinhere I'm actively thinking about creating a discord server in the near future 🙂
@@SamuelBoschMIT Look forward to that!
Awesome video! Can’t believe you got Peter Shaw himself to participate 🤩
Cool Video & it's awesome that you got to interview Peter Shor.
Thank you :) Yeah, Peter is awesome :)
A update on your research would be awesome!
I could do that soon :)
Great stuff! I will have to read up on quantum computing a little bit before I get what you are talking about.
Your channel is awesome!
Some production tips (ask around if people agree, if they don't agree please disregard):
1.) The side angle is a bit too extreme, if you want to keep it I would do it at 45degrees as opposed to almost 90degrees. The side angle also tends to miss focus. Also see point #2.
2.) For me personally the side angle is not needed, if you want to hide your jump cuts, shoot in 4K and then crop between takes. This might make your life easier.
Did you have to ask MIT news office for permission to shoot on the campus? I remember when I did a PhD (not at MIT) everything was supposed to be cleared by the PR office. Times could have changes a lot since the rise of social media (I am old:)).
Thank you for your comment and for your feedback. Yes, I agree - we should have had the camera at 45 degrees. And maybe even not have it at all. We're still in the process of figuring out these details 🙂
Probably the best solution is making sure the camera is really fixed in place and the zoom doesn't change, so we don't have to crop anything between the individual takes.
And yes, there are some places at MIT where I'm not allowed to film without permission. But luckily, as this project is supported by MIT to some degree, I am allowed to film there 🙂
So fascinating…Now I understand what you are studying
Specially like the last part😉
Thank you! 😃
I so much love the CSAIL background at the end 😊
It's such an awesome building 🙂
Can I research at CSAIL without being a student of MIT?
Great video! Thanks a lot for those interesting insights! Do you have to implement every quantum algorithm / quantum instruction directly in (quantum-) logic gates or is there any equivalent to the microarchitecture of a classical computer you can use to implement the algorithms on a higher abstraction level?
That's a great question, Floria! So essentially, writing a quantum program through direct quantum gates is the equivalent of coding in assembly language. But we're looking for the quantum equivalent of a high-level programing language (such as Python). To be honest, I am not sure if anyone is really working on this. There exists, for example, python libraries by IBM (Qiskit) and others. But I don't think they are actually capable of doing sometime comparable to python. I am actually gonna ask my advisor about it. Because I don't even know what kind of operations one should implement for being able to write simple programs
@@SamuelBoschMIT Hi there Samuel, really cool video thanks for sharing. I believe ZAPATA's Orquestra and Agnostiq's work, is a step in this higher abstraction-level. It has also lately occurred to me that there's a fundamental difference from classical microchip architectures in the way we think about q-computers today (at least most of the architectures). In classical chips, we put the logical gates on the chip and the bits are just abstract inferences that result from current flow; so it's fair to say it is the gates embed on the chip and the bits flow through them. It is also evident that transistors are not in 1-to-1 correspondence with bits by looking at it this way.
In most of the quantum architectures however, we actually put the qubits on the chip and not the gates. Thus, we embed qubits on the chip and the gates flow through them by means of external drive such as microwave radiation for superconductors. The only exception I can think that goes back to the classical-type architecture with embed gates and flowing qubits is semiconductor photonics. But then, the computational model becomes Measurement-based QC which is quite even more so non-trivial than gate-based I'd say. I think this is quite interesting.
@@SamuelBoschMIT this is the question I'M curious about. what kind of operations can we perform to implement simple quantum program?
secondly, how you train your simple NN to learn quantum circuit? can you please explain this to me? Can I have your email ID? I'm earning my PHD in quantum deep learning
Wow, Wow! I was unsure whether to pursue a computer science degree, but this got me excited again. Excellent video, new subscriber and greetings from Argentina
Hey Alexis, I'm very glad you enjoyed my video! 🙂 Best of luck to you 😊
Peter is so funny! You should totally make a longer video with him 🙂
I'm definitely gonna ask him in the coming months 🙂
truee
Simply WoW. Congrats on your PhD 👏
This was such a cool video! Congrats on all your hard work!
Thank you so much! 🙂 I really appreciate the comment!
Yes we need a detailed video on quantum machine learning
Thanks Vikas, I guess I'll have to do this sometime soon 🙂
@@SamuelBoschMIT Thanks for replying... I am a computer science undergraduate with specialization in Artificial intelligence here in India. I had attended IBM's Quantum Summer school in the past. That's why I am really excited about the amalgamation of these two fields.
I really want to know what is it and what is going on cutting edge of it...
That's all and I will be waiting for the video 😉
⭐ Yes we surely want yr more detailed discussion with Mr.peter
Quantum Artificial Intelligence? Fascinating!
You need to keep these bloopers sprinkled across the whole video. You immediately came off so much more authentic
Hey Fab, yes I think you're completely right. It wasn't the best idea to put these bloopers at the end of the video for a variety of reasons, including what you just mentioned. Thanks for your advice! 🙂
Detailed video would be great - so interesting
Wonderful video. I earned a PhD in AI 10 years ago, somehow without ever taking a machine learning course. What online courses would you recommend to a reasonably smart computer scientist with some AI background for learning both machine learning and quantum computing? I've retired from my main job and I would like to follow in your footsteps.
Seth Lloyd wrote a book years ago, "Programming the Universe," that was excellent. AI combined with Quantum Computing will change the world as we know it....hopefully to our benefit.
Yeah, it’s a really nice book!
Would love to learn more about quantum machine learning...!
I definitely did a double take when Peter Shor came up, that's wild, to meet someone with an algorithm named after them
Hey Samuel, great video! The field you do is extremely interesting and aspiring!
Btw, I'm currently an undergrad student at UIUC studying Physics and Computer Engineering. I am interested in doing research after graduation but PhD seems like a long run. Could you please make a video comparing getting a Masters degree and PhD degree (specifically at MIT) and how you decided to do a PhD instead of a Masters? Thanks!!
Hey Jerry, thank you so much for your comment! I really appreciate it! 🙂
That's a great idea for a video. Many people have been asking me similar questions, so I will put it in my list for video ideas. MIT doesn't really offer masters degrees in STEM fields (unless you did your undergrads there), but Harvard does, so that would be the best university to look at.
Best of luck 😁
Quantum multiple agent reinforcement learning using a variational quantum circuits is fascinating!
Sounds almost scary 🙂
Realy amazing work! 😀
Thanks a lot!
This is a very revolutionary field
loved this, even I didn't understand it fully, maybe i will start my small research because of you.☺☺
now that was impressive 🔥🔥🔥
Thank you 😊
Haha! I love the collective confusion! Hopefully we will all get to collective clarification about our Universe! 🤓
Great video! Videos about your story of getting admission into prestigious master's and PhD programs would be very interesting. Especially since you could provide fascinating insights :D
You know it's funny I was literally just sitting together with friends at MIT talking about recording this video over the weekend. It will be released next week Saturday. This one will be about getting into undergrads at MIT, but I will also soon record a version where I talk about how to get into PhD programs at places such as MIT, Harvard, ...
@@SamuelBoschMIT Wow, great to hear that! Besides that, master programs would be especially interesting to me. And maybe you could also share what is unique about EPFL or also ETH Zurich compared to Harvard and MIT, but I am sure you already have enough ideas for future videos. Looking forward to follow your youtube journey! 👍
Yeah, those would also be interesting videos. Getting into master programs is much easier than PhD programs, so that should be an easy video to do.
I’ll also think about something for Switzerland
Thank you for doing these videos… it’s very helpful
Simple and straight froward, thanks. I like you to make a video concerning the job opportunities and commercial applications, related to quantum information science, that you foresee to be around the corner (I mean available in the next five years or so). If you do this it will be really great.
this guy is so cool. I am also interested in quantum ML
Thank you 😊
This is a great video! I already went through a quantum computing course and I want through go the quantum ML route too. Please make a video about how to get into a phd in the MIT as an international student.
Hey Ademar, thank you so much for your comment. I really appreciate it :)
I will soon make a video about PhD admissions for internationals - definitely in the next 2 months :)
How cool is that, I'm on my way to apply to my PhD in AI and quantum is something I am very interested.
The really interesting field for me is to map quantum processing within an organic neural network to dramatically increase the processing speed whilst utilising the vast data silo that was hitherto an unwieldy mess. Solve that puzzle and the singularity will be insignificant.
Yeah is is very new indeed nice thumbs up.
Very interesting video. I have seen in your website that you are working on "Convex Optimization Algorithms for Quantum Computers with Applications in Computational Finance," when are you going to post this paper to arxiv? I am interested in using quantum computing in AI such as quantum NN.
Soon, soon. It’s been due for years, but we keep fixing stuff and running additional simulations 🙂
Quantum AI is not enough. You should do "Cloud Quantum AI Blockchain". This is where the revolution is.
There we go - you invented the ultimate buzzword compilation 😂
Very honest video, which is much appreciated. Sounds like a great topic for your PhD and a really good approach to kick it off with. Good luck with it and look forward to hearing about your breakthroughs.
Hey Dino, thanks for your comment. Much appreciated 🙂
what safety precautions are you taking to ensure that the system is safe? what forms of supervision and supervised learning are you using? Have you looked into a tertiary device to make sure that the AI system does not operate outside of what is intended?
how about, a 3-drive operating system, 2 live systems and a passive logging system. the passive logging system acts as an airlock so if the ai tries to go rouge and hack the "governor" on your "engine" it gets shut down by eliminating the electron flow to the AI device thusly shutting it down completely.
I have the same blue shirt you're wearing in the elevator lol. Great video!
No way! 😂
wow wasnt expecting peter shor himself to be in this video
Hahahaha, I had to make a little surprise 🙂
Amazing video! Thank you for sharing
Thank you Daniel, I'm glad to hear that you had fun watching it :)
Great video! Thanks 🙏🏼
can you do a podcast with lex fridman? you two are currently my favorite people of mit. carry on the good work.
Thank you Anirban! I already reached out to Lex, but didn’t get a reply sadly. But I will try again 🙂
@@SamuelBoschMIT I will also ask him to sit with you :)
This is a very interesting idea, to create quantum artificial intelligence. I would be delighted with a fuller explanation of your work. I wonder, may quantum artificial intelligence somehow help space exploration or energy problems?
Great video! One question, I'm curious to what Dr. Shor meant at 5:35 when he said we don't know why classical machine learning works. I mean, surely we do right? Tuning weights and reducing error by training would (to a limit) lead to a better model intuitively.
Thank you! I think he referred more to being able to prove minimum errors/accuracies and interpreting what features exactly do the NN weights extract. In practice it doesn’t really matter, except there is no way to prove that Quantum NNs will work too 🙂
*Easy way to understand Qubits*
A Qubit "secretly" holds a floating point number (e.g. 75% - this is the probability to read 1 instead of 0).
If you read a 16-qubit register initialized with the example value above in each qubit, some random set of 12 qubits would most likely be read as 1 and remaining 4 qubits will most likely be read as 0 (most likely does not mean always - e.g. a fair-coin toss 10 times would most likely lead to 5 heads and 5 tails but not always).
Since you read the register, there is no secret left to be hidden so all 12 of those "1" bits now contain (100%) and all 4 of those same "0" bits now contain (0%).
When qubits store (100% or 0%) they are indistinguishable from classical bits (1 or 0 respectively). Classical bits in your computer are very good but not perfect (storing e.g. 99.999% or 0.001% so they are not super different from binary-constrained qubits).
That's true 🙂
*Wave functions never collapse* . An observation is just an ordinary interaction that *entangles the observer with the experiment* creating the "effect" of wave function collapse in the mind of the observer. Here observer implies one or more machines (measuring instruments) along with one of more humans (team of scientists / researchers working together and interacting with each other and those measuring instruments).
Great Video! I have a question, why aren't you using the Qiskit Aer simulators for your research?They have more powerful simulators right?
I create my own simulations using numpy or pytorch, but I suppose quiskit aer could also be used in many cases :)
@@SamuelBoschMIT wow that's cool...I wonder what advantages do your own simulations have over the Qiskit ones,just curious because I'm working on a project which uses Qiskit Simulators.
MIT midroll add for quantum computer algorithm online classes. 👌😅 nice video sir, thank you!🤙
Hahahahhah I guess UA-cam’s ad algorithm is doing a good job then 🙂
That was a most enjoyable and interesting learning experience--a truly fun way to learn about your work in this fascinating field. Thank you!
You’re very welcome 😉
Really amazing work 👌
1like and Love from India
Thank you! 🙂
very cool video concept! you should try to reach out to Lex Fridman - the elevator pitch can be a great crossover
Hahhaha thanks David, thats a great idea given that Lex Fridman is a lecturer at MIT. He didn't respond to my least email, so maybe I should try again once I have a bit more followers. Would be so nice to have him on my podcast :)
@@SamuelBoschMIT oh that's really cool, I didn't know you have a podcast! Yea Lex seems like a really cool guy that would love to be a part of student-led media - especially since he's in a similar field of research, is affiliated w/ MIT, and loves to interview various professionals. I'm sure you've thought of related ideas, but I would try to reachout by pitching him on something like a creative campus-based podcast/social thing; you can offer him to be a part of it and maybe you can scale this across universities - kind of like an AI club that promotes the machine learning field and inspires/networks like-minded students (potentially providing mentorship, scholarship, and project/research oppurtunities w/ on campus professors or via online community). And if done at scale, it's likely Lex may have you on his podcast to discuss something like this. Regardless, I am very happy I came across your channel; your thesis sounds very interesting and I wish you the best of luck with all your pursuits!!
Sam please make an update to this video. Now that ChatGPT is out I want to learn more about QAI
Just about to start my master's in data analytics but I am totally doing my PhD in the same field man.
Hahhaha sounds awesome! Good luck with that 🙂
Quantum computing through quantum mechanics in application to artificial intelligence is going to be what allows for the emergence of consciousness in AI one day.
Thats so cool!!!!!!!!!! letsgoooooo
💪💪💪
Would love to see a video about how the AI part of investment management works. Thats all I've been hearing about is investing with ai trading stocks and stuff. I'd like to be a part of the investing but don't understand it at all.
If you're interested in one-on-one BCI neural visualization AI with EEG's... aka "reading minds" then quantum artificial intelligence could be _something_ .
Great video!
Thank you :)
Video on Journey till PhD, what events lead to what.
Hey Santosh, I actually already made a video about precisely this journey, so you can check it out here 🙂
ua-cam.com/video/gSq-7A5XPvs/v-deo.html
hey this was an awesome video, it would be great if you can explain more about machine learning and quantum machine learning too, thank you for such awesome videos!
Sure thing! 🙂
I was trying to listen, but I couldn't stop seeing Patrick Swayze's ghost talking from the heavens.
Hy fame Im so excited abut this quantum Ai idea we could actually use it for nextlevel humanoid tach 🤩😍🤩
It's very exciting indeed 🙂
Very interesting video, great insights!
Thank you Lorenz :)
I want to know in detail how AI and quantum computer are related to each other.
Because I'm currently doing engeenering in artificial intelligence and machine learning and also curious to know the quantum world.
Hi, Samuel! Thanks for the extremely interesting video! Do you calculate any expectation values in your quantum A I algorithms? Do you understand how a quantum computer calculates an expectation value? Does it take much time?
Hey Natalia, expected values are usually just calculated by taking averages. Quantum computers don’t do that, so it is necessary to run a program multiple times to get the expected value
@@SamuelBoschMIT Oh, it seems I need to run the program thousands of times to calculate approximately the correct expected value. But how much time does it takes a real quantum computer to prepare a state and to take a measurement just once? Is it fast?
@@nataliaandronova3027 I should be quite fast. So even if you have to repeat something a few thousand times, IBM's (very small) quantum computer can do it within a second or less
Im a little confused here, I thought quantum computers arent yet turing complete, yet you seem to already be able to run machine learning algorithms on them?
Yottacool! I wonder what you think about AI safety, AI ethics and BMIs?
Love your videos, any advice for high school student trying to get into MIT?
Thank you - I really appreciate your comment 🙂
I even made an entire video dedicated to your question. It’s called “How to get into MIT as an international student”
@@SamuelBoschMIT That's the video I need man 😂. I'm thanking you even before I watch it!
can you make more deatiled vedio on this? please? this is interesting
Thank you Kajal, I might make an updated version sometime soon 😊
My 2 cents: There's no reason that quantum AI won't work. Looking at the algorithms , there's nothing tying it down to a 2 bit system. However, a lot of the hardware specific processing we've done might not carry over too well( I know nothing about that side). But the core algos shouldn't have any problems .
Dude I work in an IT company focused on data science and stuff, my manager is super hyped about this quantum machine learning. What is the status of your PhD btw? :P
In the next year i start a master's degree in quantum computing in italy, if you have any suggestions (books, online courses, etc...) to read or follow, I am very glad to accept your advice. Your videos are really interesting, congratulations 😊
Thank you Giovanni! 🙂 I would have a look at Qiskit's and IBM's online courses. They are probably a great way of getting started!
Can you give a tour of the MIT CSAIL Building?
Btw Love you videos♥️
Thank you Aditya 🙂 I’m glad to hear you’re enjoying my videos! I might film a tour if MIT at some point, including CSAIL 😁
Quantum computing based artificial intelligence will be the future because of their nature which will help them in decision making.
was brill danka- funny when u first started to speak I honestly thot u were gonna say... N today we are gonna talk about how absolutely amazing my chest is...
Hahahahaha thanks - I also have a video talking about workouts :)
think of it this way, the unintended consequences of a system become amplified by exponential ^ exponential not just the intended consequences. so, without an ark or mars like system to prevent things like group think and other existentially disastrous cataclysms, the work on AI using tertiary bits rather than binary, is a little dangerous.
I am in search of new research topic for my PhD studies.
Working on quantum algorithms might be a good way to go :)
@@SamuelBoschMIT Thanks for suggestion. I will definitely go through the literature of Quantum Algorithms.
Hello Samuel, thanks for this interesting video. I am learning "classical" machine learning by myself, and I was wondering in what extent Quantum ML differs from it, can u briefly explain this ? is Quantum ML also using same well-known ML algorithms like SVM, KNN, neuro-networks, logistic regression and so on ? Applying them in some other way ? Or is it just completely different and only sharing the name "ML" in common ? Thanks for your time :).
Hey, I do think it is rather different - even though we try to make it look and work in a somewhat similar way. Before diving into QML, you should first focus on learning quantum computing in general. Some online courses by IBM qiskit may be very helpful here :)
Generally speaking it is a buzz-word mating. No-cloning-theorem forbid's to use quantum information in the neural models as a naive replacement of inputs. Quantum computers are "small" in the sense of the amount of q-bits, and the essence of the machine learning is to create very big, complex models of reality compared to hand crafted algorithms. "Broadcasting" of information is essential for any kind of big models.
I do not expect any speedups better than Groover's algorithm (quadratic).If you consider that the coherence time is inversely proportional to the square of the number of q-bits, we do not get any speed-up at the level of physical machines that can be built, but only a certain constant. And I have no confidence at all that this constant will be above one.
How come we don't know why classical machine learning doesn't work? Of course, we do know. We apply statistical algorithms so that the neural network accumulates a lot of data and then starts recognizing patterns, am I not right? Could someone please explain if I'm wrong?
Why did he say that no body knows why ML works on digital computers?
Well, we cannot theoretically prove or guarantee classification accuracies or convergence rates during training. In my personal opinion , this is not a big issue though
@@SamuelBoschMIT Yeah this statement of his was exaggeration.
Loved this video though!
It would've been even more helpful if you could've added screen recording or ss of some of the actual code behind training those neural networks(+ which parameters it controls) some of your work in general 😃
And please more such content... :)
according to wikipedia, D-Wave has Quantum computer with 5640 Qubits since 2020, it powerful enough for RSA encryption, no?
Hey Asaf, thanks for the comment. Dwave doesn’t have a real (=universal) quantum computer. What they have is a quantum annealer. It can solve sole problems, but definitely not Shor’s algorithm 🙂
this is awesome
Thank you 🙂
Nice video Samuel, already subscribed!!! What are your future expectations in the field? I mean, do you see yourself as an academic researcher or applied researcher? maby both, jaja... Anyway, wish you success in your career, and hope to see more videos concerning your quantum computing adventures!!! Greetings from Mexico!!!
Hey Jorge, thanks for the comment! 🙂 i think the field has quite some potential, although I personally don’t think I will stay in research after my PhD
Could you make a video about books you would recommend someone interested in the field of quantum computers?
Heyyy thanks for the comment. That's a great question! I (personally) am very bad when it comes to reading, so I'm probably not the right person to talk about this. But my friend, Anastasia Marchenkova, has a youtube channel and made two videos about exactly this topic, so you may wanna have look at them 😊
ua-cam.com/video/3zHp4Rg-GmM/v-deo.html
ua-cam.com/video/1AxiAiIUli8/v-deo.html
Now quantum artificial intelligence seems great ! But what about *Interdimentional quantum artificial intelligence !* My phd in 6 years wish me luck 😁
baba... what is Quantum ML actually ?? just shell shocked !!
More detail! :D
So is the main problem with quantum computers to accurately measure the output?
I don't think it's counterintuitive. A lot can be discerned from the double slit experiment. And honestly, by that point you should be delving into statistics and probability.
If you don't know the correct result, how can you determine the correct Hamiiltonian that maximizes the probability of producing it? This has always confused me with QC's.
Whats your workout program? Any tips?
I'm almost done filming a video about precisely this. The most efficient ways to work out and get the best results based on science and research. Coming soon in the next 2-4 weeks, so stay tuned 🙂💪