With this level of artificial intelligence, bugs are not a common occurrence.

Over the years, AlphaGo has only had one obvious bug when playing against chess players from South Korea.

After that, AlphaGo's team optimized the algorithm and this bug was fixed.

"Is it because of another upgrade?"

"No, we just upgraded the computing power, and the core algorithm has not changed from before."

"In that case, why does this happen?"

"When AlphaGo encounters a very rare move of its own, it will fall into a search dilemma. And if AlphaGo feels that its winning rate is already very low, it is easy for this kind of program crash."

After hearing this explanation, the leader of Google was also very puzzled.

"Hasn't our artificial intelligence already played countless games? How come it still encounters some rare moves? Could it be that they have caught our shortcomings and deliberately made a rare wrong move?

"

"No, this cannot be said to be a wrong move, but a very rare coup."

At this time, a staff member said.

This staff member is a software engineer and an amateur six-dan chess player. He plays a very important role in the entire AlphaGo development team.

"This move is almost never seen in professional chess players, because under normal circumstances, humans think that this is a move that will make them passive. The complexity of its changes is simply not something that the human brain can handle. However,

For an artificial intelligence with extremely strong computing power, this is probably a good trick."

As this staff member said, with the emergence of this bug, AlphaGo was in a state of collapse and made a series of unlucky moves.

This is also the difference between artificial intelligence and humans.

When the artificial intelligence determines that its winning rate is very low, it will start to play randomly, because no matter where it plays, it will have no impact on the final outcome.

Seeing that AlphaGo's tricks began to become increasingly outrageous, AlphaGo's team had no choice but to quickly accept defeat.

If this continues, AlphaGo will probably fill up all its eyes, and eventually the entire chessboard will be black's territory.

Seeing such a scene, Nie Jie, the commentator in the CCTV live broadcast room, was undoubtedly the happiest person.

"It seems that AlphaGo suddenly has serious bugs like before. I don't really understand the logic behind these artificial intelligences, but I guess the big reason is because our island artificial intelligence has gone out of AlphaGo.

Fagou made an unexpected move. This shows that the artificial intelligence of the island greatly surpasses Fagou in terms of game experience and computing power. It is just a pity that this game ended too quickly."

Nie Jie couldn't help but secretly rejoice that this time the upgraded version of AlphaGo actually didn't last as long as him and the island's artificial intelligence.

If you round up, Nie Jie's strength is already above AlphaGo.

Although he thought so in his heart, Nie Jie also knew that he would definitely not be able to defeat either of these two artificial intelligences.

But no matter what, as long as China's artificial intelligence can defeat the United States' artificial intelligence, it will be a satisfying thing.

"As a professional software engineer, let me explain the possible reasons for bugs in Alfa Dog."

At this time, another commentator suddenly said.

This commentator is called Chen Yuchang. He is not only a professional software engineer, but also participated in the research of Penguin Go artificial intelligence.

"Because all current Go artificial intelligences are based on deep learning systems to improve their chess skills, rather than using exhaustive methods. This makes every artificial intelligence likely to have bugs. To be more precise,

, in this type of algorithm model, this bug is called overfitting."

"Teacher Chen, can you explain what overfitting is?" Nie Jie asked.

While Nie Jie was cooperating with Chen Yuchang, he did have some interest in this issue.

"The so-called overfitting can be understood as decision-making errors caused by imbalanced AI data. I just said that the improvement of AI's chess ability comes from its own deep learning. For example, when AI plays against itself

When playing chess, we have accumulated 10,000 games. Some chess shape changes have been accumulated more, and some chess shape changes have been accumulated less. For example, in a duel with few chess shape changes accumulated, the overall winning rate reference is relatively small, because

Too much chance..."

After listening to Chen Yuchang's explanation, Nie Jie basically understood the reason.

"Can it be understood this way? For example, for a change, if the AI ​​only accumulates ten games, wins three and loses seven, it will think that the winning rate of this change is only 30%. But in fact, if the accumulated

If there is enough data, its true winning rate may be much higher than this data?"

"Yes, that's what it means. Simply put, as long as you keep making changes that the AI ​​has never seen, it will easily lead to bugs."

"I see. If I had known about this bug earlier, when I played chess with AlphaGo, I would have used more odd moves. For example, if I made a Tianyuan or something like that on the first step, I might have been able to steal a game.

Victory." Nie Jie joked.

At this time, the atmosphere in the Island Quantum Laboratory was also peaceful.

As everyone expected, Island Artificial Intelligence won the first game of chess.

Even if Black does not have an advantage, the island artificial intelligence still has a crushing advantage.

However, Xu You did not expect that the upgraded version of Alpha Dog would be unable to withstand it so quickly.

"Such a chess game is difficult to test the true strength of the island's artificial intelligence." Xu You said helplessly.

As the creator of the island's artificial intelligence, Xu You understood why AlphaGo suddenly collapsed just now.

It must be the island's artificial intelligence, which accidentally made a rare move in AlphaGo, directly triggering a hidden bug.

This kind of bug cannot be completely avoided. Only by relying on a large number of duels can the frequency of bugs be reduced.

Compared with supercomputers, the calculation speed of quantum computers is much faster, so it can accumulate a huge number of games in a short period of time.

It can be said that the probability of bugs appearing in island artificial intelligence is extremely low.

"If it weren't for the fact that this was a game that was broadcast live on the entire network, I would have just let the island artificial intelligence give up." Xu Youxin said.

But after only one game of three rounds of chess was played, Xu You still couldn't let down his guard.

Xu You decided that if he could win easily in the second set, he could give AlphaGo a little more in the last set to test the upper limit of the island's artificial intelligence.
Chapter completed!