The Future of Computing: Optical Computers, Bio Digital Twins & The Evolution of AI - Kazu Gomi

What happens when you give the world’s leading scientists the freedom to research whatever they want? You get NTT Research, a global powerhouse operating at the edge of what’s possible in physics, medicine, and AI.

In this episode of the SCL Podcast, filmed inside the MD Acoustics anechoic chamber in Gilbert, Arizona, we sit down with Kazu Gomi, CEO of NTT Research. Kazu pulls back the curtain on how a company with over 300,000 employees is tackling the massive energy crisis facing modern data centers and why the future of computing might not be silicon, but light.

IN THIS EPISODE, WE DISCUSS:

• The Energy Crisis of AI: Why GPUs require so much power and how NTT is aiming to reduce energy requirements by 1/10th or even 1/100th through optical computing.
• The Bio Digital Twin: A revolutionary look at how digital "twins" of the human heart are being used to predict disease and automate medical treatments like "autonomous driving" for drug infusions.
• Can AI Feel? New research into the "granularity" of AI emotions and how modern models are beginning to understand feelings with the sophistication of an adult human.
• Attribute-Based Cryptography: Moving beyond "all or nothing" security to create files where different keys unlock specific pages or sections.
• Leadership & The "Strong Core": Kazu shares his journey from electrical engineer to CEO and his advice for young professionals on perfecting their "major" to reach the next level.

CHAPTERS:

• 00:00 – Intro: The Anechoic Chamber 
• 06:27 – The Scale of NTT: 300,000 Employees & Underwater Cables 
• 10:47 – A Unique Approach to R&D 
• 13:04 – Optical Computing vs. The Power Crisis 
• 22:50 – Heart Health & Bio Digital Twins 
• 29:22 – Next-Gen Cryptography 
• 33:18 – AI’s Understanding of Human Feelings 
• 41:03 – Career Advice: Finding Your Strong Core 

Transcript

SPEAKER_02 0:06

Welcome to the SCL Podcast, filmed in the MD Acoustics Anachoic Chamber at their testing facility in beautiful Gilbert, Arizona.

SPEAKER_01 0:17

Today's guest on Soundcave Labs is Kazuhiro Gomi, also known as Kazu. Kazu is the president and CEO of NTT Research, where he leads global research initiatives focused on advancing technologies such as artificial intelligence, photonics, cryptography, and next generation communications. Under his leadership, NTT Research is working to bridge fundamental research with real-world innovation, helping shape the future of computing, networking, and digital infrastructure. Kazu brings decades of experience in telecommunications and global technology leadership. Prior to founding NTT Research in 2019, he served nearly a decade as the president and CEO of NTT America. This followed his earlier role as Chief Operating Officer and Chief Technology Officer, where he led development of global IP networks during a transformative period for the Internet. A 40-year veteran of NTT, Kazu began his career contributing to strategic research initiatives at the company's cyberspace laboratories, a foundation that continues to drive his passion for technical discovery even today.

SPEAKER_03 1:35

Thank you so much for coming on. We're excited to have you here.

SPEAKER_00 1:38

Thank you very much for having me, Mike. Good to talk with you.

SPEAKER_03 1:42

Yeah, so you know, let's just jump right into it. We we're really excited to have you here. Um you know, your title is CEO of NTT Research, right? And so maybe you can kind of give us a little bit of background about you know who you are, uh kind of where you came from and and how you got to where you're at.

SPEAKER_00 2:02

Okay, sure. Um uh thanks, thank, thank you very much again for having me. Uh it's it's very nice to uh uh talk to you uh at this podcast. Um so uh I'm uh I I was born and raised in Tokyo, Japan. I wonder if I'm the first Japanese national to be invited to this podcast.

SPEAKER_03 2:23

You series? You are. We're excited to have you from 100%. I made it. Yes.

SPEAKER_00 2:27

I made it, yeah. So I'm I'm excited about that as well. Uh so I work for this company, NTT Research, and I'm the CEO of it. Um and I probably should uh spend a little bit of time to explain who NTT Research is and then what we do. Uh that's probably the main dish uh for this uh podcast. But uh before going to that, uh just to introduce myself as a person. Uh so uh like I said, that I'm I I was born and raised in Japan, and this company, NTT Research, is rooted from uh from Japan. Uh so natural kind of kind of fit into it. Uh but uh I've been uh I I started working for NTT when I was in Japan uh right after college. Uh so I've been actually working for this group, NTT, uh close close to 40 years, believe it or not.

SPEAKER_04 3:21

That's amazing.

SPEAKER_00 3:22

Uh it's it's kind of uh well, used to be the case, I guess, but uh I'm sure it's super rare. Uh so I'm kind of like a legacy from that point of view as well. Uh but uh I came to uh I started my career in Japan. Um and uh about 20 years ago I moved to United States. Uh and uh first 15 years um of uh my tenure in the United States, uh I stayed in New York City, and I was running the uh NTT's North America business. Okay. Uh so I was a CEO of NTT uh company called NTT America, uh which run uh global networking uh and uh data centers, uh IT uh managed services. Those are the areas that uh this, those are the business that this company uh has been uh conducting. Uh we did a lot of MA uh while we were uh while I was there. Uh and it has been growing ever uh since then. Uh but uh um my background is uh engineering uh and uh uh more yeah definitely more uh I I call myself uh tech tech oriented, a little more geeky uh side of the house. Yeah. And uh so uh for the last five, six years that I've been running this group called NTT Research, which is at uh uh research and uh development organization, RD organization, in short. Uh so that's where uh I am uh data center, right?

SPEAKER_03 5:06

Right where we live, where where we live. So really what's really exciting is obviously it's ex expanded and kind of NTT research and how you guys are kind of at the forefront. There's so many questions I have to ask, but your guys are at the forefront really of what's going on kind of in the guts of the data centers and everything. So yeah, maybe you can dive into that a little bit for us.

SPEAKER_00 5:28

Yeah, sure. Yeah, so um from um so back in the day, telephone was a big serve, a big, big deal, but uh you know, fast forward, uh not many people pay too much attention to the telephone, especially the landline telephone. Who who uses that now? Uh so things have changed to mobile phone, and of course, that uh um all that uh uh internet uh and then uh uh cloud services and the AI, everything is happening uh in the data centers. So NTT has shifted its uh uh uh business focus from telephone services to something else uh that includes the data centers. And uh as Mike uh pointed out, uh NTT owns uh data centers throughout the world now, and uh one of the top three uh data center providers in the world right now.

SPEAKER_03 6:24

Uh and uh I think many of the listeners of Jeff, but just for the listener again, like NTT, just so that the people understand here in the United States, the scale of how big you guys are. Uh, you know, I read, and and you could correct me if I'm wrong, over 300,000 worldwide employees. NTT's research, again, you can tell us how many employees and all about that, but like it's it's big. It's it's really it's really big.

SPEAKER_00 6:49

Yeah, uh you're right about uh size of the company, uh over 300,000 employees around the world. Uh I forgot exactly how many countries that we operate, uh somewhere around uh 60 uh or 70, uh give or take. And uh once again, data centers are one of the key uh pillars of our services, uh, but we provide all other uh telecommunication IT-related services uh uh there.

SPEAKER_03 7:21

And and and uh sorry, and to say, like when you say IT, you're also talking like underwater cabling across the ocean.

SPEAKER_00 7:29

Right. Right.

SPEAKER_03 7:30

So you guys own a majority of that as well, a lot of that. Maybe not all of it, obviously.

SPEAKER_00 7:34

Uh yeah, majority maybe a little bit of uh to exaggeration exaggerations, but we do own uh uh a lot of uh routes uh throughout uh especially uh between the United States and Japan. And uh the uh throughout the Asia, South Pacific, uh South uh South Asia to India. Uh that's our kind of core uh strength uh for the uh cables.

SPEAKER_04 8:00

Nice.

SPEAKER_00 8:01

And um back to your question about RD, maybe I guess uh pivot a little bit and uh shift that the conversation towards that. Um so uh Entity is perhaps one of the few companies uh still, well, maybe I should use the word still, uh, have uh pretty large uh in internal RD uh capabilities. The the reason I said still was that uh back in the days, maybe up until the 80s and 90s, uh many of American companies uh also had uh pretty strong uh in-house RD. Uh again, AT ⁇ T uh was one of the good examples of that. The uh ATT used to have uh organization called Bell Lab. And they're really iconic RD organization, they had uh many uh many inventions and uh many um Nobel Prize uh laureates uh were born from that organizations and so on. But uh perhaps except uh a few IBM, Google, Microsoft, that the many of those large companies, uh whatever the reason it is, that uh uh kind of uh divested those RD capabilities. Uh but um NTT, whatever the reason, right or wrong, we stick with it. And that's where we believe uh we should do. Uh so so that that that's where we are. Out of this uh 300,000, uh so it's a it's a small piece, but still just that two uh over 2,000 is uh pretty large uh size team.

SPEAKER_03 9:44

And and you have a campus, right?

SPEAKER_00 9:46

Like it's a full-fledged yeah, uh most of the 2000 over 2,000, uh most of the 2500 RD staff, as you can imagine, are based in Japan. Uh and we have four campuses in Japan. Uh but uh uh the team I manage directly is uh uh in uh Bay Area, California. Uh that's where I'm speaking from right now. But uh that we have a we have a well compared to 2,000 uh member team, this is much smaller, it's about 100, but uh we have uh operation there.

SPEAKER_03 10:25

Nice. So let's just dive in. Like talk to us a little bit about what's really exciting that you've been working on and all that fun stuff.

SPEAKER_00 10:33

Yeah. So um we do um so we we do RD. And then when you talk about RD, there are a couple of different ways to do it. Um and uh I believe that our approach is a little bit unique. Uh the reason I say unique is that uh many RD typically, you know, you want uh this type of product, this type of services, you want to release in say two years, three years. So back top, backtrack from there. And then let's just uh start doing uh this topic and assuming you can make a good business out of it two to three years. Our approach is a little bit different. Uh we are focusing on basic fundamental research. Uh so when we start, uh meaning when you pick the topic uh of the research, uh we don't know exactly what this is going to be uh used for in a business. Uh so in other words, the our approach is that we hire um what top, we believe top uh uh world world leading uh researchers, uh, including up-and-coming young researchers, and uh let them choose the topic that they are interested in and that they are curious about, excited. And we believe our belief is that uh if those people choose these topics, uh well, we are hoping that uh probably not two, three years, but a little bit longer period of time it may take, but uh it will make a big uh impact, impactful uh scientific discoveries, inventions uh these people should be able to uh achieve. Uh so that's how we operate. So once again, our approach is basic uh and fundamental research. And uh more specifically, uh, you may ask the quit want to know what exactly the topics that these guys are uh focusing on. Uh we have uh a few uh focus areas, uh even though we let them let them choose the topic, but we we have some sort of uh wide range uh kind of like a swim lane. And uh one is that uh uh I would like to introduce four areas that we are uh focusing on. Uh one is that uh uh optical uh devices. Uh more specifically, we're trying to create uh the optical computers. Um maybe we can dive into further uh if time allows. And then that's one. Uh and the second one is uh cryptography. Um so uh encryption encryption decryption technology. And uh third one is what we call the bio-digital twin. Uh once again, uh this keyword uh means a lot, and uh uh we can dive into further, but uh digital twin technology applied to human human body. And uh last one, uh as you may uh wonder, uh what about AI? And uh so we are also working on the AI, but uh uh again, our approach is from basic fundamental uh uh tangent. Uh so not so much on the how to use AI to basically you know uh bring more efficiencies in this uh tasks or anything like that. What uh we are doing is that uh trying to understand how the AI machine learning mechanism is uh done. Uh so more fundamental stuff of that the AI uh machine learning.

SPEAKER_03 14:13

Interesting. So when I think of data centers, right, I think of a lot of power consumption like we talked about. We think of, you know, and obviously, you know, we we do a deal with acoustics up here in an Anna Coke chamber. We deal with, you know, my company, we do a segmented of RD where we pay as a company, but you know, we're we're just a super small company, which is why I was excited to have you on here because I know you know data centers and power usage is a big deal. So I gotta think you're talking optical. I mean, we let's just dive into that and let's let's talk about a little bit about how you guys are looking at power consumption and kind of the the four elements that you just talked about. Can you can you dive in a little bit on the optical? Sure, of course.

SPEAKER_00 14:53

Yeah, yeah, sure, of course. Uh so data centers. Um, so one of the critical things is the power, and especially in this AI age, that uh those AI uh well GPU, which is the core of the core of to run that uh uh deep neural network, uh large language model, whatever you call it, um, that uh AI platform. And uh this GPU really, really requires a lot of power, uh electricities. Uh and then uh so everybody's really need given what's going on in an AI boom, uh, all the major AI companies building more and more AI platforms, and then to support that, we need data centers. And to support the data centers, you need a lot of electricity. So that's what's going on, and uh it's been highlighted in uh news uh newspaper articles every here, there, and everywhere. So I'm pretty sure people have seen it. Um and uh under the circumstance, what can we do uh for that? Immediately, something we have to do, and we are already doing, is that try to do everything we can do to run the data center more power efficiency, efficiently. Uh, one of the big things is that uh uh that those uh processing units they produce so much heat. So uh one of the big uh elements of the data center is a cooling. And then when it comes to cooling, there are actually quite a bit of a technology, um, new invention, innovation is going on, and then bring more uh energy efficient uh cooling systems and so on. So those are done, and of course, that the clean energy, we want to bring uh in clean energy as long as much as possible to avoid the uh the um global warmings and so on. So those things are uh ongoing and an incremental um improvement is uh uh are uh done, it's on the way. But fundamentally, what we have to do is this uh how to make those GPU much more energy efficient, and energy efficient. Um so from this, uh so that's a big uh big theme uh from the RD perspective. And uh already many of the chip companies, uh server company people have uh uh tackling uh this problem by implementing uh optical uh devices uh into the mix. Uh and uh this is um happening in uh in a kind of into a couple different phases. Uh first off, is that uh um how to well uh when you're diving into that uh AI data centers, uh one of the problem uh problem or one of the characteristics that you should be aware of is that uh within the data center, there are so much data transport happening between the processor and the memory. Uh super fast transmission is required, and a lot of data needs to be transmitted between the MME and the processors. And uh when you use uh, well, uh most of the device uh uh platform is built upon the silicon electronics uh basis, uh as you know. Uh but uh those uh electronic-based uh data transmission, it works, but uh this requires a lot of energies, especially uh when it comes to that uh uh high speed, uh high clock rate uh data transmission. Uh but uh if you replace that into uh optical system, uh you can reduce the energy uh requirement so much uh while maintaining that the clock speed uh or even that the higher, even higher clock speed is can be achieved uh with the optical system. So that's certainly step one. And uh a lot of people, a lot of companies start implementing those areas. And uh uh and then uh and that this data transmission can happen in many different ways. Uh when you assume the data, you know, data center, there is a rack where a lot of servers are uh stuck uh into this uh uh in a rack. So uh well, between the servers, we want to use more optical uh based system. And then within the servers, there are uh many boards. The how to connect the boards, uh, let's use uh optical. And within the board, uh there are many chips. So how to let's use optical system to connect the chips. And even further, there there is uh, you know, within a chip, there is uh element, many, many different in the chip. So connect those chips, uh elements in uh with the optical. So those steps uh uh uh I think by by and large, everybody's kind of looking towards that uh uh that steps. But from uh my team uh's point of view, when I say optical computer, we are actually trying to replace that the processor itself uh can be run by that the optical system instead of the silicon-based digital uh uh processing unit. Um it sounds kind of uh maybe weird, but uh it is actually the optical system uh or optics uh itself, you can do the uh computations, and this computation can be applied to that AI uh deep neural network. So that's that's where what what uh our team is uh looking at. And uh well, it's not happening over the night, it will take definitely a few years, but uh our dream is that uh we can replace that uh GPU and the memory with uh through the optical systems. Uh and uh by doing so, we should be reduced the energy requirements of those servers, of those compute platforms drastically. Uh it's a matter of uh like a uh one tenth or one hundredth uh that level of uh energy requirement reduction that we are shooting at.

SPEAKER_03 21:49

In in San Francisco, there, are you behind computers and running simulations and computation? Is that primarily what you're doing in there in Silicon Valley?

SPEAKER_00 21:58

Yeah, so yeah, yeah. A we have a uh those facilities, so physical and experimental uh tools and devices, uh, but also there are people uh working on more more theoretical things and a computer uh programming. So yeah, mixed mixture of those uh the people's uh working in uh in our lab.

SPEAKER_03 22:20

Okay. And so now kind of pivoting from the optical, by the way, that sounds fascinating. Obviously, if you're using optical and you can reduce it that much energy, that's like you said, so dramatic that could have obviously huge uh impact. And it's especially because it's not up in space, it's actually here on Earth. This is all the other stuff that you've probably heard about. So this is this is awesome. What um what about so, like a you know, a bio digital digital twin? Can you talk to us a little bit about what you guys are doing?

SPEAKER_00 22:52

Sure. Yeah, yeah. Um, so let me explain uh this way. Uh digital twin is uh kind of um not just a concept, uh actual solutions uh out there. Uh typical digital twin technologies applied to something like a jet engine, uh, very expensive physical system. And uh, you know, when the jet engine is created in the factory, coming out of the factory, all the same spec engines are well basically the same. But when it is uh uh mounted to the uh to the airplane and stuff flying all over the place in the world, and each engine starts um taking a different uh trajectory, right? That uh then uh how many hours are flying and the temperature of that the airs, the airspeed, uh, and so on. So depending upon those um um those conditions that the engines were operated, um the wear and tears in the engines are you know happening in the different parts. So um so to monitor those stuff, uh digital twin, digital the twin of the physical item uh is uh created on the computer simulation base, and then start tracking those each engine's their trajectory. By doing so, you can predict that the which parts is most likely uh you know uh where and tears are happening. So you can do the preventive maintenance, you know, actively and efficiently, so on. So that's the twin uh digital twin concept, and you can apply to the city planning, uh to uh like an oil rig, and in many different uh ways. Now, uh our concept of biodigital twin is that instead of uh taking a trajectory of the physical system, let's do the same concept for the human body. Uh so you can, so ideally, right, when this is completed, you can really traject uh and predict that uh you know what kind of disease or illness uh that you may uh get into uh if you just keep on doing whatever you are doing right now. Therefore, uh you should change this or change that. Uh that's the idea. But as you can imagine, that's a that's a that's cool, but that's a little bit too difficult to achieve. Uh so what we are focusing on is uh that uh just the heart and the cardiovascular system, uh biodigital twin. So uh instead of the human body entirely, you just focus on the heart. And uh currently, what we have uh achieved over the last five years or so, we are doing this project. Um uh what we called acute cases, meaning that you already uh taken to the hospital, you got uh heart failure uh or heart attack in that type of scenario. You are taken to the hospital. Uh, and uh uh we can apply this digital twin uh concept and approach, and uh you can really predict uh how this heart is gonna uh behave uh in in uh and we can simulate different kinds of drugs being prescribed. Uh what how do they uh react to it? So so in a way that the system helps the uh physicians uh to determine what to do in uh in their obviously that in uh in the ICUs or in a in a hospital environment. So that's that's where we are, but also pivot from there is that because we can predict those um behavior of your uh of the patient's heart, uh what we are trying to achieve is that we may potentially do that uh more uh we can um uh do this medical treatment more automated fashion. Uh so drug infusions and stuff like that, we can just obviously we're not there yet, and then this requires a lot more testing and so on, but uh potentially it's like an autonomous driving car, it's an autonomous driving uh drug infusions and so on. That's what uh we're trying uh to accomplish. And then to do that, this uh digital twin technology is that the core uh core of it. You can simulate how uh well precisely simulate that how uh the patient's uh heart um is uh reacting uh to uh to different drugs. So that's cool.

SPEAKER_03 27:53

So like that even goes into more of like longevity of life, which obviously is is kind of like one of those things which you know we've heard people will live longer. I mean, this is kind of in that tech, in that world where yeah, yeah, yeah.

SPEAKER_00 28:05

And then another if I may add uh one more thing is that uh by doing so, uh our real hope is that by doing so, that uh the medical system can be hopefully cheaper, uh more, you know, that the less human uh involved in the in the entire care. Um so um more affordable at the end of the day to everybody, and then uh in the in a rural area, rural settings, those uh top of the notch that the medical uh care can be uh available uh to the to the to the rural areas and so on. So uh yeah, longevity is the goal, but uh to get there, we can uh provide the medical system, uh medical treatment a bit uh cheaper, and then the more uh availability is going to be uh you know, wider range that uh so those those are the types of things that we are shooting for.

SPEAKER_03 29:08

And I'm assuming there's quite a bit of machine learning in all of that as well, right?

SPEAKER_00 29:12

Yes, yeah, yeah.

SPEAKER_03 29:13

Yeah, very cool. So, what about um, you know, so we hit on the other topic, which would be crypto cryptology.

SPEAKER_00 29:20

Cryptography, yeah, cryptography.

SPEAKER_03 29:22

Yes. Let's talk to us a little bit about what you guys are doing there.

SPEAKER_00 29:24

Yeah, so cryptography, um well, this topic is alone, is uh a lot of things are going on, but uh something I can uh explain, the type of things that we are doing, exciting, is that uh yes, just assume something like that. Uh so when you well, probably everybody knows about cryptography now. So you encrypt the data, uh say, and then if you have a right key, you can just open the file. If you don't, you don't see anything. So so today's most of the cryptography is uh you see you can access to the file 100% of it or 0% of it. So either all or nothing. Uh assuming there is a cryptography technology that uh you can uh well you lock the file, but uh to open the file, you can issue uh different keys. And the mic, you get one key, and uh I get different key. And with the mic's key, you can, well, when you open the file, you can say, assume this is a book. Uh, and then mic's key allows you to access to the odd number of pages, and the my mic key allows only the first 10 pages or or something. You you can just control everything like that in the as a part of the cryptography. This type of thing is possible. So many interesting crypto technologies uh is kind of screwing out. Um so give given that uh so many machine learnings and you have to you have to share data amongst ourselves to make the system more intelligent and then do a do a good thing for for us, you you need to also hide the privacy and uh confidentialities and so on. So so this technology, we believe, can play a pretty important uh piece of it. Uh so but yeah, anyway, so the this is one example, but the crypto uh uh just looking at uh um um the output possible for the next generation of cryptography. People are thinking about those type of uh new uh new type of cryptography.

SPEAKER_03 31:48

Very cool. That's that's fascinating. I mean, when I think of um, you know, sharing files or file sharing or even segments. I mean, I'm sure there's so many ways to slice it and dice it, like you're saying. Yeah, the book example is a good example. But you know, that's just you know, if you've got large file structures or large things that you're just trying to, you know, keep prying eyes from, I mean, I could see where that would be really important.

SPEAKER_00 32:15

Yeah. So what what we are trying to actually actually, this is one area that uh we coming out from the basic research, you know, we are um trying to productize this technology. This is a technology that I'm actually pushing for the productizing it, making a business out of it.

SPEAKER_04 32:32

Yeah.

SPEAKER_00 32:33

And uh the real use case is going to be that uh if there is a file, and then uh if it's a you know, in a in the old days or the today, even if you circulate the documents with uh the privacy or confidentialities included in a document, what you do is put the black mark on on uh you know social security numbers, credit card numbers, whatever we include in the document, right? So we can do the same thing, digital basis, yeah, uh using this technology. So interesting. Uh yeah.

SPEAKER_03 33:07

Very cool. And then so the last section we talked a little bit about was AI, correct?

SPEAKER_00 33:11

Yeah.

SPEAKER_03 33:12

So in our last maybe five or 10 minutes here, yeah, like walk us through that. What are you guys working on there?

SPEAKER_00 33:17

Yeah. So once again, uh what we are trying to do is more basic science uh approach. Uh so uh we're trying to understand how that the machine learning is happening. Uh and uh something interesting these researchers have found, I thought that was fascinating, is that uh how much does the AI know? Well, say the chat chatbot type of um applications, how much uh AI understands, and then uh what do you mean by that? How much feelings that the AI chatbot understand uh does you know, the other person's feeling. So it was a fascinating uh results. Uh uh when you talk about feelings, uh a little bit different uh uh discipline, uh psychology, there's a lot of studies going on. Uh when the baby was born, obviously the baby has a feeling, but very few dimensions, uh, either sad or happy, hungry, full, sleepy, maybe that's about it. Um but as you grow old, uh your feelings become much more sophisticated. Uh, happy can be happy because you are recognized, or happy because you see interesting movies, you know, something pretty, something interesting. So the happy itself becomes very, very granular. Uh same can be said for the sadness, same can be said for angry and everything. Uh we so these researchers try to understand the AI and analyzing how much do they understand speakers' feeling? And then actually, in the uh as the AI model becomes bigger, uh AI does understand very similar to what the adult uh human beings uh feeling, uh so similar granular. Uh in the back in the day, say 10 years ago, when the model was much smaller, uh, that uh this granularity was very small. So it's like a baby uh baby stage, either happy or angry, uh, or whatever. Uh so so that's one of the uh so you know you may ask so what how do you make money out of it? I don't know. It's probably you cannot make money out of this uh discovery, but uh this is uh interesting. Well I think it's an interesting discovery that to understand uh how much AI is, how how far AI has gone uh already so far that that similar uh really understands the feelings of uh of the the uh once again, you know, chatbot uh type of AI. Uh AI does understanding, does understand that the people's uh feelings uh by speaking for a long, long time. Yeah.

SPEAKER_03 36:28

That's amazing. I you know, I just was having a conversation with a colleague yesterday about how much AI will or won't impact you know all industries, specifically, you know, in my industry. And I just sat there and said, you know, we have no shot because if you look at how smart AI is, right, and it's got some sort of an IQ, and it's just gonna progress. It's never gonna go lower. I could read every book in the world, oh, and it's across all languages. It's a so when you talk about it, it understands feelings. I don't think people understand like just how sophisticated the systems are or and how much more sophisticated. And it's only a matter of time before it does. I don't want to say replace us, it's gonna be an assistant, but it could. And it could do all the math in the world, right? Like, you know, all the all the physics and all the engineering and everything that I took, and all my, you know, all of our employees and your employees. It I mean, you we just don't know. We're just at the right, we're just kind of getting you like you said, 10 years ago, here we are 10 years later. Who knows what it's gonna be like in 10 years?

SPEAKER_00 37:28

Right, right. It it it to some extent, it's kind of scary, like everybody feels. I I feel the same. Where are we going from here? Um, and uh scary scenario, AI takes over all the human beings. And well, I don't think that will happen. I I don't think that will happen, but uh uh but uh you know, like I yeah, like you said, that 10 years, so much progress has been made. So what's next in the next 10 years?

SPEAKER_03 37:55

And uh it just tells our well, I just I just think of like my kids, I have little kids, right? And I sit there and I think like for them, what's important to study? Because what is it that's gonna be differentiating them maybe from AI or from not from AI, or how do you augment to AI, or how do you do you have to be even more smart? Because I, you know, my my my daughter, she doesn't like math, and then my son, math whiz. So I'm just like, you know, do you tell her, hey, you know, AI is gonna help you through college on some of your math, right? Like I I just think about that, but anyway, I don't want to get off track here, but that's the kind of stuff with the AI thing that I think is is honestly fascinating. Yeah. And and I think, you know, in terms of monetization, I think you're validating. You're just talking about, well, how do you make money off of it? Well, no, you're validating what could be and how do you take all that and you know you spin it into something else.

SPEAKER_00 38:44

So yeah, exactly. So yeah, um, I I think I it's a building blocks that uh uh nobody really thinks about this angle that uh how you know it seems like AI understands your feeling, but uh yeah uh this is a kind of scientific approach, then all right, let's just just study how much uh how much granulity that uh the of the feelings that the AI understands, and then boom, they have uh they have found something like this. Interesting. So yeah.

SPEAKER_03 39:15

Culturally, I was gonna ask, since you are, you know, the first kind of Japanese, you know, background, I was just curious to know when we talk about like an American company, you kind of hit on you know, Bell Labs and ATT and how that's they've divested from that. And maybe that that exists maybe in Apple and a couple of places, like you're saying, but really they've divested. And you know, do Japanese kind of companies do they still spend a lot of RD time? And is that is that kind of cultural, do you think?

SPEAKER_00 39:47

Uh yeah, I believe so. I I don't have that uh actual you know data to show you, but uh I think many Japanese companies tend to uh stick with that uh kind of kind of company culture, uh tip definitely a cultural thing that you should you should uh work hard and come up with uh your own technology. Um so it takes a longer time, but uh yeah, that's the that that should be the core uh uh yeah, things move forward.

SPEAKER_03 40:20

That's amazing. Well, so before we leave, is there, you know, I I have one more question or two more questions, I guess. Would you know um I guess the first question would be uh is there anything else you want to share before we let you go? And then also the second question to piggyback off of that, and I'll ask it now, is what would you have told yourself, your younger self, you know, throughout your career journey if you could go back?

SPEAKER_00 40:47

Yeah. Um so I so I uh before I'm I'm doing this management uh job for the RD team. Uh and then prior to that, uh I was managing a large business uh business team. Now um and about my background is uh engineering geeky side. Uh something I learned over the time is that the two uh having uh one strong core, uh in my case, that uh I'm I'm uh electrical engineering background. And uh so I most of the time I understand what's going on, I I claim myself, uh, in the computer science computers, you know, how things are made, and then uh both software and hardware, what the core is. Having one strong core is very, very helpful. So why when I assume the job uh to run this business in uh for NTT uh North America, obviously being a CEO, you have to you have to know, you don't have to know everything, but you need to understand that there's so many other things. Mike, you probably have uh similar thing that you're dealing with, that you have to understand accounting, you have to understand the HR rules, labor laws, and all the other kinds of regulations, laws, you know, those kind of things, which I hated, I have to admit, that when I was studying uh in a school, uh I kind of intend in one, yeah, in uh yeah, I try to avoid those uh those studies, but but it is very important. Um but if you have a core strength, you can spend time to learn about those things that after you became a CEO. I I felt like uh so uh some of those uh technical things, I most of the time I understand. So I don't have to spend too much time on it, but so I can spend time to account the rules, which I didn't know much at all. So so I I really felt that having one string is a really good, and then it's impossible to know everything from the get-go. So just uh as um yeah, uh I I tell yeah, young people that uh just uh stick with your measure and then really you should perfect it. Um then that will help you uh take you to the next level, whatever it is. So uh so that's my thinking. And then um yeah, thank you very much for the opportunities. I really appreciate it.