Go: Is the portable old man the nine-time champion?

Chapter 438 Version Enhancement
"This……"

The members of the Chinese Chess Academy couldn't help but look at each other, exchanging glances in unison. They were all somewhat hesitant, but none of them dared to speak out their true thoughts.

After all, it would be too hurtful to say it out loud.

What kind of foreign AI is this? Trying to mess with us Go players now?!

There's no need for Yi Zheng to even step in; every one of the old men here feels absolutely confident that they can defeat the so-called artificial intelligence.

Having Yi Zheng take action is, in a sense, overkill. Was it really necessary? Wouldn't it lower Yi Zheng's standing?

In the midst of this delicate silence, Yi Zheng broke the deadlock. He said calmly, "Mr. Zhang, last time I tested your company's artificial intelligence with Jiang Mobai, a 9-dan professional. However, at that time, its Go strength was roughly equivalent to that of an average professional 1-dan player, which is still quite far behind ours."

If artificial intelligence remains at this level, I don't think there's any need to hold large-scale competitions. Perhaps incorporating it into some commercial exhibition matches would be a more interesting option.

Yi Zheng was actually very clear about the future development potential of artificial intelligence, but at this moment he wanted to take this opportunity to test the level of the artificial intelligence developed by Muye Fox Company.

Zhang Xinping was not surprised by Yi Zheng's words. After all, in the last test, the artificial intelligence was no match for Yi Zheng, let alone Jiang Mobai, a ninth-dan player.

Therefore, it is normal for Yi Zheng to underestimate artificial intelligence, but this time the artificial intelligence is completely different.

Zhang Xinping smiled slightly and said firmly, "The last Go AI was only the third iteration, but this AI has already reached the eighteenth iteration."

The difference in strength is completely incomparable. According to our internal assessment, this version of the AI ​​has surpassed the skill level of a professional Go player at the 9th dan level, achieving a score of 3900 in our self-assessment.

The score of 3,900 is calculated according to the world Go rating system. The highest rating in human history is currently held by Lee Jun-hyuk, who achieved 3799 points, just shy of 3,800 points.

As for Yi Zheng, he has only reached 3779 points so far. After all, Yi Zheng has only been in the industry for a short time and the number of games he has played does not support him to reach a higher position, so his score does not yet fully match his performance.

Zhang Xinping spoke with great confidence, after all, people in science and engineering believe that data doesn't lie.

If an artificial intelligence claims to have 3,900, then it definitely has 3,900.

Therefore, the strength of the artificial intelligence developed by our own research lab is absolutely beyond doubt.

Although Zhang Xinping spoke with great confidence, and theoretically he is not a complete novice in Go, but has a deep understanding of the game, he should be aware of the strength of a professional 9-dan player.

His words were somewhat credible, but the people present still found it hard to believe that artificial intelligence could have the strength of a professional ninth dan, let alone surpass it.

Anyway, many people in the room already had that look on their faces: "Really? I don't believe it."

Yi Zheng silently sipped his tea and calmly exchanged words with Zhang Xinping: "Mr. Zhang, theoretically speaking, the training of artificial intelligence seems to have never broken through the field of Go. How did your company manage to do it?"

Zhang Xinping didn't hold back when he heard this, since everyone was roughly the same in the general direction of artificial intelligence. What really mattered were the algorithms that each company kept to itself.

He was more worried that Yi Zheng and the others wouldn't understand. After all, in the sports industry, let alone science subjects, most people haven't even had much formal education.

"Games are a very convenient thing. Many games have a scoring system, so it is easy to measure progress."

Go is probably the most complex game ever designed by humankind, so defeating professional players in Go is a long-term and enormous challenge for artificial intelligence research.

Not only us, but most AI companies worldwide are researching related topics. Our AI's training in Go is a project of our parent company.

As Zhang Xinping explained, Yi Zheng nodded repeatedly, but most people looked completely confused.

Professor Ke once introduced the artificial intelligence in their world. The first generation of Go was, in a sense, the pinnacle of human Go wisdom. Although there were some unexpected moves, the overall framework did not deviate from human Go, and basically every move was within everyone's understanding.

So to some extent, it has not escaped the realm of humanity, and sometimes it even falls into the same mistakes as humans.

However, the second-generation dog was completely different. Although the first version of the second-generation dog also used human chess records, it chose a newer algorithm and training method.

This method not only trains AI systems more effectively but also yields better training results. It can be used for self-play, Monte Carlo tree search, and deep reinforcement learning.

Later, the AI ​​company went even further, abandoning human game records altogether and allowing the AI ​​to train itself with only a basic understanding of the rules of Go. This was equivalent to having a human with superhuman computing power start playing Go again in another world, forging a completely different path from the original approach.

It even completely denies humanity's previous belief in the correct path in various choices and directions, which is the so-called AI chess.

This refers to the techniques that Master Ke taught Yi Zheng, and these techniques have already begun to spread widely throughout the world.

Thinking of this, Yi Zheng's expression couldn't help but become a little strange.

According to Professor Ke, the first generation of artificial intelligence was trained using chess records, but because human chess records were too poor, the development of artificial intelligence was hindered.

However, everyone is now vying to study the correct game record brought by Professor Ke, and the world's artificial intelligence will learn from this game record...

Theoretically speaking, the starting point should be stronger than what Ke Lao mentioned about the so-called first-generation dog.

However, Yi Zheng was not sure how powerful the artificial intelligence that obtained such partially correct data could be, since his strongest understanding of computers was that they could be restarted when something went wrong.

What would an artificial intelligence that is on the right track from the very beginning look like? What amazing power would it possess? Yi Zheng could not say for sure.

But precisely because of this, he was even more excited.

The strong never care about any challenge.

(End of this chapter)