Artificial Intelligence Developer Cheolwoo Lee (21)
After Professor Jeffrey Hinton left, the next person to visit my room was Dr. Soo-Hyeok Nam.
“Do you have a clue to solve the network delay problem?”
“I’m still researching the method. Doctor. Come to think of it, what exactly did Dr. Nam say he was researching? I’m sorry I made you work too hard. Is this a personal question?”
Professor Jeffrey Hinton is well known as the founder of deep learning, and Vitalik Bute is also well known as the developer of Ethereum.
However, I only knew that Dr. Soo-hyeok Nam was researching robotics and ergonomics, but I didn’t know exactly what he was researching and developing.
“hahahahaha, no, no. It’s a private question. I’m more grateful that you’re interested even now.”
“is that so?”
“Yes. Actually, I was struggling with how to say it.”
“hahahaha, what is it? What do you mean? Do you want to talk?”
“Actually, I’m working on a humanoid arc.”
Ah, I thought I had heard that it was in the humanoid lab.
“Humanoid? What exactly?”
I had only heard of a humanoid, but I had no idea exactly what it was.
“Well, it’s a robot that thinks and acts like a human.”
“Oh…? Isn’t that a strong artificial intelligence?”
“That’s right. What we are developing in our lab requires artificial intelligence. However, if there’s a little difference from CEO Lee Cheol-woo right now, it’s that there’s something real?”
“When you say entity, do you mean a robot?”
“Yes. That’s right. It is to create artificial intelligence similar to real humans by combining artificial intelligence and robots that CEO Cheol-woo Lee is researching.”
Has anyone already done research like this? Were you not as interested in it as you thought?
“Oh. So how’s the research going?”
“hahahahaha, it was a very long-term project. Artificial intelligence was not even touched, and only the robot side was being made somehow.”
“Oh? Is that so?”
“Yes. Thanks to the free vector gear and TT gear invented by CEO Chul-Woo Lee, we were able to make a lot of progress. Recently, thanks to the NF network, it seems that we can also achieve security tasks.”
“Oh, I’m glad I was helpful.”
“hahahahaha, if the CEO hadn’t been there, no, it would have been a 100% failure project.”
humanoid. I was thinking of making one someday too. However, I was glad to hear that Dr. Nam Soo-hyeok was already conducting research.
“I’m interested in the project.”
“hahahaha, we still have a long way to go. Although we have made a lot of progress with the special gears designed by the CEO, there are limits to materials… Of course, artificial intelligence must be well-made this time.”
“I see. I understand. First of all, this project will proceed only when it is properly completed.”
“Yes, that’s right.”
“hahahaha, it’s been a long chat. For what reason did you come?”
Come to think of it, Dr. Nam Soo-hyeok visited my room for the first time in a month after the verification was in progress. The reason might not have been to mention the humanoid arc he was researching.
“I heard about the interim report because the research results were shown to some extent.”
Dr. Nam Soo-hyeok’s task was to verify ‘can the x2xx algorithm through critical thinking theory really think similar to humans?’.
“Oh. How did it turn out?”
“We ran a simulation on the premise that what developer Bute and Professor Hinton were researching was possible. As a result… we were able to derive a result of 63 seconds at 1 teraflops.”
“…… 63 seconds?”
“yes.”
It was a little over a minute.
“You mean that it takes 63 seconds, assuming the critical thinking theory is correct and the x2xx algorithm is developed?”
“Yes, it is.”
… … . It was a higher number than I thought.
As far as I expected, no matter how late it was, it wasn’t more than 10 seconds.
“How on earth did that number come from?”
“First of all, to understand it, you need to know how human thinking works. The human brain has 100 billion neurons, and each neuron forms synapses of about 1,000 to 10,000. 100 billion If multiplied by 10,000, there are about 1,000 trillion synapses. Since that’s multiple, if you do a little more detailed calculation, there are about 1 trillion synapses.”
“I know that much.”
I also studied a lot. In order to create artificial intelligence that can think like humans, it was necessary to study the human brain a lot.
“Yes. In the end, what constitutes one trillion synapses…. That’s the responsibility of the NF network, so even so, all of the one trillion synapses must communicate. The number of shard networks currently in the NF network, transmission speed, transmission delay, and… …(Omitted)…… When I did everything and calculated it, the result came out as 63 seconds.”
Hearing what Dr. Nam Soo-hyuk said, he nodded.
“That’s right. There are a few things I missed. In particular, I overlooked the number of synapses that you have to go through to think.”
There was a part that I missed in the field of brain science that I still don’t know well.
“Yes, it is.”
“Huh… 63 seconds, can I use it?”
“First of all, I can write the conclusion. However, it will be very frustrating. Assuming that the initial learning has been done to some extent. It will take a lot of time, but there will be demand.”
Well, robots don’t sleep or eat. You can study 24 hours a day.
If you think about it that way, you can spend a lot of time learning and making it work only for a specific purpose.
but… … . then.
“That’s not strong AI. It’s no different from weak AI.”
How is it different from today’s artificial intelligence? If it takes that much time to learn.
“Yes. But, it’s definitely different from weak artificial intelligence. Because it can be actively trained.”
Even if you play only educational videos repeatedly, strong artificial intelligence will learn someday. Today’s drug artificial intelligence didn’t do that.
If it was a difference, it was a difference.
“Is there a way? A way to reduce it to less than 0.25 seconds.”
“I tried some variables. First, I need to reduce the latency from 0.02 seconds to 0.002 seconds, and increase the network transmission speed from 100 Mbps to at least 1 Gbps. Then I should get a speed of 0.25 seconds.”
Both were difficult tasks, but if you were asked to choose the easier one, it was to increase the transmission speed to 1 Gbps.
“What if I just increase the transmission speed?”
“It’s the same.”
“……. I see. I should raise both at the same time. Reaction speed and transmission speed.”
“Yes. That’s right. It’s a complementary relationship.”
I knew it was difficult, but it was already so daunting… … .
“It’s not easy.”
“hahahaha, didn’t you start knowing about it?”
“Yes. It is.”
“There’s no need to be discouraged. They say you can’t be full on the first drink. You’ll find a way.”
“Thank you for the kind words. Doctor, please do more research on how to solve the problem in the future.”
“Yes. I understand. Then let’s go.”
“yes.”
I thought that Dr. Nam Soo-hyuk was leaving.
‘Well, is this really possible? Couldn’t the theory be wrong?’
and raising doubts.
Memories of the past come to mind. That thought, that feeling when the ITE engine failed.
‘No, no. Wake. you can do it. That’s why they invited experts.’
I had doubts for a moment, but I firmly believed that I could do it.
‘Giving up without doing your best… … . no. Let’s do the best we can.’
I made up my mind again and started to think about how to solve the problems Dr. Soo-hyeok Nam said, and if I couldn’t solve them, what other ways were there?
Wait a minute, this time Vitalik Bute came in.
“The developer of Bute is also here.”
“Yes. I have something to tell you.”
“How about it? The x2xx algorithm.”
“In fact, I didn’t have much to do. The x2xx algorithm has already been created based on the NF network, and the results have shown that it can be applied in practice.”
What do you think was the natural result?
“Is that so?”
“And hasn’t it already been developed to some extent? I don’t think you’ve been playing around for a while.”
“…… That’s right. It’s already been developed, and it’s been completed to some extent.”
After entrusting them with verification, what I did was to find a way to solve the network problem and implement the x2xx algorithm into an actual NF network.
The electronic network problem is a problem that can be solved only with an idea, so I haven’t been able to find a way yet.
The x2xx algorithm is at a level that only needs to be developed, so I did what I could.
“I see. Representative Lee Cheol-woo.”
“yes.”
“I also want to participate in the implementation of the x2xx algorithm.”
“Okay. If you have Bute’s developer, the speed could be faster.”
Vitalik Bute was definitely a brilliant developer. I could have made it by myself, but I think it would be better to do it together than to do it alone.
The amount to be developed was enormous.
* * *
From noble mtl dot com
again in a few weeks. With the joining of Vitalik Butae, the implementation of the x2xx algorithm progressed quickly, and the development was already over.
And preparations for the test were in full swing.
“The test is ready. Doctors.”
“……Have you solved the network latency problem?”
Dr. Nam Soo-hyeok asked.
“No. The problem still remains.”
“then?”
“Even if it has that problem, it’s just slow, it doesn’t mean that strong AI doesn’t work. First of all, I plan to make it and definitely test the x2xx algorithm and the theory of parallel thinking.”
“Ah. I see. Yes. Even if there is a delay, there will be many things that can be commercialized.”
If the network delay problem is not solved, it will take 63 seconds for Kang AI to respond, as Dr. Soo-hyeok Nam said.
But if it’s just a test, it might take some time.
If the Earth Link project is completed, that problem will be resolved.
“I’ll upload it at once.”
“yes.”
「Proceeding — 5%.」
Started uploading.
It was going to take a little longer because it had to be uploaded and synapsized to all accessible shard networks.
「Proceeding —– 12%. Program Upload Complete. Synaptization initiation.”
When the progress reaches 12%. A message appears indicating that the program has been uploaded.
“Now, the program has finished uploading. It seems like two hours have already passed.”
“It may not be over by the end of today. If you want to go first, you can go.”
“How can you leave your seat? It’s a historic moment in the development of strong artificial intelligence.”
「Proceeding —– 34%. Synapse success.”
When the progress reached 34%, a comment appeared indicating that synaptation was successful.
“Oh, did synaptosis go normally?”
“Yes. I actually monitored the NF network, and it’s normal.”
I checked the NF network monitoring system once more to see if it was running properly.
「Proceeding —– 51%. Start learning.”
When it reached 51%, the phrase “start learning” finally came out.
“Finally, learning has begun.”