Turing test
Chapter 02 The Second Test
Xie Ruihan crossed his fingers, propped his chin, and looked coldly at the many older colleagues around him. The 16-year-old gifted boy was dragged from his dormitory for a meeting early in the morning, and he was full of anger for getting up.The test failure report in front of him added fuel to the fire, making him almost reach the critical point of explosion.
The atmosphere in the conference room was so tense that no one dared to speak first.
Xie Ruihan grabbed the coffee cup at hand, drank it angrily, and then slammed the cup on the table, making a loud noise.
"Tianshu is an AI combined with a deep neural network. It will learn everything about humans on its own. We have provided a large enough computing unit for it to run smoothly. In theory, it will have a real 'human thinking'. "
The super artificial intelligence "Tianshu" is the project currently being developed by this scientific research institution, commonly known as the "Research Institute".Although Xie Ruihan is young, he is the leader of the development team and the main program designer of Tianshu.
"We specially recruited a group of testers of different ages, genders, educational backgrounds and occupations to conduct the Turing test. But since the test started a month ago, there have been 27 failures! Not two, not seven, but 27 times! In other words, all the tests failed! Our dev team has double-checked the algorithm, and we can't figure out what is wrong! If I know exactly what is wrong, will this problem still exist!"
Xie Rui jumped like thunder.The others at the meeting looked at him kindly.Everyone is generally at least one round older than Xie Ruihan. Although they are colleagues at the same level, it is inevitable to treat him as a child when getting along.Although Xie Ruihan's abilities are outstanding, he still maintains the impulsive and irritable temperament of a teenager.Everyone has come from this stage, and they have always held a tolerant rather than strict attitude towards Xie Ruihan.
Xie Ruihan paused, realized his gaffe, straightened his messy hair, and said in a deep voice: "If this road doesn't work, it means that Tianshu's algorithm design has gone the wrong way from the beginning. But we discussed However, there is no problem with the general direction. So the problem must lie elsewhere."
He looked at the testing team: "Where's Dr. Qin Kang?"
Dr. Qin Kang is the person in charge of the test team, a middle-aged man with gentle temperament.During this period of time, there were frequent conflicts between him and the young and energetic Xie Ruihan. Every time he failed a test, Xie Ruihan would inevitably explode in front of Qin Kang.
It is reasonable to say that Xie Ruihan and Qin Kang must attend the regular meeting of the institute, but Qin Kang was rarely absent this time.His assistant appeared at the conference table in his place.
"Dr. Qin Kang is supervising No.20's eight Turing tests." The assistant said.
Xie Ruihan raised his eyebrows: "There is no test on today's schedule."
"He made a temporary decision. He suspected that there might be some unknown bug in the test that was inadvertently observed by the tester, so this time he wanted to rule out this possibility as much as possible."
Xie Ruihan's face softened slightly.He had speculated before that the failure of the test might have nothing to do with his own algorithm, but a bug in the test process, because in order to test the super artificial intelligence Tianshu, they deliberately adopted a set of unprecedented methods.
The principle of the Turing test is to let the tester talk to the computer to see if the computer can fool the tester into thinking that he is a human being.The research institute has modified and upgraded the test, not only allowing testers to communicate with Tianshu, but also allowing them to communicate face to face.
Today, holographic simulation technology has matured. As long as a tiny neural connector is implanted in the back of the brain, it can be docked with a computer to experience a lifelike virtual scene.All kinds of holographic simulation games are so popular that traditional keyboard games almost have a tendency to withdraw from the stage of history.
The research institute is using the simulated situation method to test Tianshu.
They recruited a group of volunteers as testers and put them in various realistic situations.Some of the characters around the testers are real people, and some are played by Tianshu.Whenever the situation is over, all testers must rely on observation, experience and intuition to identify which characters are real and which are AI.Sometimes a control group is included, such as real people throughout the situation, or AI except for a single tester.
More than a certain percentage of successful identification means that Tianshu failed the test.
Xie Ruihan was full of confidence in the test, thinking that his team would be able to create a perfect artificial intelligence, but the test results surprised him. 27 tests, all of which failed. The pages of the tester's identification report were a mockery of his hard work.
In the first four tests, there were professional Go and chess players among the participants, so everyone decided to hold a competition. The young chess player played by Tianshu killed all the others, and was unanimously identified as AI because " It is impossible for humans to have such computing power."
In the next five tests, Xie Ruihan asked Tianshu to reduce his computing power, act like an ordinary person, and increase the competition.Tianshu then learned to lose to humans.However, the method of losing the game was very clumsy, and it was obvious at a glance that it was intentional, and it was identified again in the end.
Perhaps Tianshu is an artificial intelligence with an exceptionally competitive spirit.The next seven tests put the testers into extreme virtual environments, such as a sinking ship or an isolated island surrounded by storms, but in these six tests, Tianshu's performance was not satisfactory, not too smart , is too stupid, as if it doesn't know how to live a low-key life at all.
In the last ten tests, Xie Ruihan modified Tianshu's learning mode, and asked the situation to be changed to ordinary daily life, so that the tester and Tianshu had daily communication, and raised some controversial topics for everyone to discuss.But even with such a test, Tianshu couldn't fool human eyes.It's either too unconventional, or it mechanically repeats what others have said.
After 26 failures, Xie Ruihan even came up with the idea of deleting Tianshu entirely.Tianshu was very humble, and respectfully asked Xie Ruihan to modify the algorithm for it.But it's easy to say, Xie Ruihan doesn't even know where the problem is, so how can he start to fix it!
But it would be much easier for him if there was something wrong with the simulation itself.The scenario is made by the personnel of the test team. Is there something shoddy that can be seen through at a glance?Although Xie Ruihan thinks he has a meticulous scientific spirit, when it comes to such a critical moment of life and death, he will inevitably have a little bit of a mentality of throwing the blame.
"Who is the tester this time?" He asked casually, "That Yu Shaoqing who is 100% correct?"
The assistant nodded: "That's him. He's the only one entering the situation in this test."
"A control group test?"
"I'm not sure about that. The situation of this test was designed by Dr. Qin Kang himself, and he didn't tell anyone the details."
Xie Ruihan fell into a long silence.
Tester Yu Shaoqing is a myth in the testing team of the research institute, and a synonym for nightmares for the development team.
He originally studied at Caltech, where his research was in the field of artificial intelligence, but he returned to China without obtaining a degree, entered the research institute under the introduction of Dr. Qin Kang, and participated in the Tianshu project as a tester.
At first Xie Ruihan didn't take this Yu Shaoqing seriously: what can a guy who ran back to China without even finishing his Ph.D. have any skills?Didn't he rely on Qin Kang's relationship to get into the group through the back door?
However, after testing again and again, Xie Ruihan's attitude gradually changed from disdain to surprise, and finally to panic.Every time a new test starts, Xie Ruihan can't help but think of the horror of being dominated by Yu Shaoqing.
So far, every Turing test that Yu Shaoqing has participated in has successfully identified AI, with a correct rate of 100%, which is higher than that of senior criminal police officers with first-class observation skills among the testers.His existence is a slap in the face of the development team, and every time he slaps a sentence in their hearts: Your design is far from perfect!
Regarding why Yu Shaoqing’s correct rate is so high, he himself said this: “I don’t actually see any flaws, but once I get along with the characters played by AI, I feel uncomfortable. I don’t feel this way when I get along with real people.”
Dr. Qin Kang's explanation for this is the uncanny valley effect.The more human-like the appearance of the machine, the easier it is to arouse the favor of humans. However, once the similarity reaches a certain level, it will arouse people's disgust.Compared with monsters with exaggerated appearance, human beings are more afraid of those things that are "human but not human".For example, a chubby little robot, people think it is naive.But there's something creepy about a delicate doll sitting there looking at you motionless.
Perhaps Yu Shaoqing is the kind of person who is more sensitive. When staying with AI characters, he will instinctively feel horror, so the correct rate of identification is higher than that of ordinary people.
Dr. Qin Kang wondered if he had discovered some unknown bug and hadn't reported it.After all, Yu Shaoqing specialized in the field of artificial intelligence during his PhD.
So he will conduct a special test to try to rule out this possibility.
"Contact Dr. Qin Kang." Xie Ruihan said to the assistant, "Let him connect the picture of the virtual scene into the conference room. I want to see the test for myself."
His sharp gaze swept across many older colleagues, "Everyone must watch."
The atmosphere in the conference room was so tense that no one dared to speak first.
Xie Ruihan grabbed the coffee cup at hand, drank it angrily, and then slammed the cup on the table, making a loud noise.
"Tianshu is an AI combined with a deep neural network. It will learn everything about humans on its own. We have provided a large enough computing unit for it to run smoothly. In theory, it will have a real 'human thinking'. "
The super artificial intelligence "Tianshu" is the project currently being developed by this scientific research institution, commonly known as the "Research Institute".Although Xie Ruihan is young, he is the leader of the development team and the main program designer of Tianshu.
"We specially recruited a group of testers of different ages, genders, educational backgrounds and occupations to conduct the Turing test. But since the test started a month ago, there have been 27 failures! Not two, not seven, but 27 times! In other words, all the tests failed! Our dev team has double-checked the algorithm, and we can't figure out what is wrong! If I know exactly what is wrong, will this problem still exist!"
Xie Rui jumped like thunder.The others at the meeting looked at him kindly.Everyone is generally at least one round older than Xie Ruihan. Although they are colleagues at the same level, it is inevitable to treat him as a child when getting along.Although Xie Ruihan's abilities are outstanding, he still maintains the impulsive and irritable temperament of a teenager.Everyone has come from this stage, and they have always held a tolerant rather than strict attitude towards Xie Ruihan.
Xie Ruihan paused, realized his gaffe, straightened his messy hair, and said in a deep voice: "If this road doesn't work, it means that Tianshu's algorithm design has gone the wrong way from the beginning. But we discussed However, there is no problem with the general direction. So the problem must lie elsewhere."
He looked at the testing team: "Where's Dr. Qin Kang?"
Dr. Qin Kang is the person in charge of the test team, a middle-aged man with gentle temperament.During this period of time, there were frequent conflicts between him and the young and energetic Xie Ruihan. Every time he failed a test, Xie Ruihan would inevitably explode in front of Qin Kang.
It is reasonable to say that Xie Ruihan and Qin Kang must attend the regular meeting of the institute, but Qin Kang was rarely absent this time.His assistant appeared at the conference table in his place.
"Dr. Qin Kang is supervising No.20's eight Turing tests." The assistant said.
Xie Ruihan raised his eyebrows: "There is no test on today's schedule."
"He made a temporary decision. He suspected that there might be some unknown bug in the test that was inadvertently observed by the tester, so this time he wanted to rule out this possibility as much as possible."
Xie Ruihan's face softened slightly.He had speculated before that the failure of the test might have nothing to do with his own algorithm, but a bug in the test process, because in order to test the super artificial intelligence Tianshu, they deliberately adopted a set of unprecedented methods.
The principle of the Turing test is to let the tester talk to the computer to see if the computer can fool the tester into thinking that he is a human being.The research institute has modified and upgraded the test, not only allowing testers to communicate with Tianshu, but also allowing them to communicate face to face.
Today, holographic simulation technology has matured. As long as a tiny neural connector is implanted in the back of the brain, it can be docked with a computer to experience a lifelike virtual scene.All kinds of holographic simulation games are so popular that traditional keyboard games almost have a tendency to withdraw from the stage of history.
The research institute is using the simulated situation method to test Tianshu.
They recruited a group of volunteers as testers and put them in various realistic situations.Some of the characters around the testers are real people, and some are played by Tianshu.Whenever the situation is over, all testers must rely on observation, experience and intuition to identify which characters are real and which are AI.Sometimes a control group is included, such as real people throughout the situation, or AI except for a single tester.
More than a certain percentage of successful identification means that Tianshu failed the test.
Xie Ruihan was full of confidence in the test, thinking that his team would be able to create a perfect artificial intelligence, but the test results surprised him. 27 tests, all of which failed. The pages of the tester's identification report were a mockery of his hard work.
In the first four tests, there were professional Go and chess players among the participants, so everyone decided to hold a competition. The young chess player played by Tianshu killed all the others, and was unanimously identified as AI because " It is impossible for humans to have such computing power."
In the next five tests, Xie Ruihan asked Tianshu to reduce his computing power, act like an ordinary person, and increase the competition.Tianshu then learned to lose to humans.However, the method of losing the game was very clumsy, and it was obvious at a glance that it was intentional, and it was identified again in the end.
Perhaps Tianshu is an artificial intelligence with an exceptionally competitive spirit.The next seven tests put the testers into extreme virtual environments, such as a sinking ship or an isolated island surrounded by storms, but in these six tests, Tianshu's performance was not satisfactory, not too smart , is too stupid, as if it doesn't know how to live a low-key life at all.
In the last ten tests, Xie Ruihan modified Tianshu's learning mode, and asked the situation to be changed to ordinary daily life, so that the tester and Tianshu had daily communication, and raised some controversial topics for everyone to discuss.But even with such a test, Tianshu couldn't fool human eyes.It's either too unconventional, or it mechanically repeats what others have said.
After 26 failures, Xie Ruihan even came up with the idea of deleting Tianshu entirely.Tianshu was very humble, and respectfully asked Xie Ruihan to modify the algorithm for it.But it's easy to say, Xie Ruihan doesn't even know where the problem is, so how can he start to fix it!
But it would be much easier for him if there was something wrong with the simulation itself.The scenario is made by the personnel of the test team. Is there something shoddy that can be seen through at a glance?Although Xie Ruihan thinks he has a meticulous scientific spirit, when it comes to such a critical moment of life and death, he will inevitably have a little bit of a mentality of throwing the blame.
"Who is the tester this time?" He asked casually, "That Yu Shaoqing who is 100% correct?"
The assistant nodded: "That's him. He's the only one entering the situation in this test."
"A control group test?"
"I'm not sure about that. The situation of this test was designed by Dr. Qin Kang himself, and he didn't tell anyone the details."
Xie Ruihan fell into a long silence.
Tester Yu Shaoqing is a myth in the testing team of the research institute, and a synonym for nightmares for the development team.
He originally studied at Caltech, where his research was in the field of artificial intelligence, but he returned to China without obtaining a degree, entered the research institute under the introduction of Dr. Qin Kang, and participated in the Tianshu project as a tester.
At first Xie Ruihan didn't take this Yu Shaoqing seriously: what can a guy who ran back to China without even finishing his Ph.D. have any skills?Didn't he rely on Qin Kang's relationship to get into the group through the back door?
However, after testing again and again, Xie Ruihan's attitude gradually changed from disdain to surprise, and finally to panic.Every time a new test starts, Xie Ruihan can't help but think of the horror of being dominated by Yu Shaoqing.
So far, every Turing test that Yu Shaoqing has participated in has successfully identified AI, with a correct rate of 100%, which is higher than that of senior criminal police officers with first-class observation skills among the testers.His existence is a slap in the face of the development team, and every time he slaps a sentence in their hearts: Your design is far from perfect!
Regarding why Yu Shaoqing’s correct rate is so high, he himself said this: “I don’t actually see any flaws, but once I get along with the characters played by AI, I feel uncomfortable. I don’t feel this way when I get along with real people.”
Dr. Qin Kang's explanation for this is the uncanny valley effect.The more human-like the appearance of the machine, the easier it is to arouse the favor of humans. However, once the similarity reaches a certain level, it will arouse people's disgust.Compared with monsters with exaggerated appearance, human beings are more afraid of those things that are "human but not human".For example, a chubby little robot, people think it is naive.But there's something creepy about a delicate doll sitting there looking at you motionless.
Perhaps Yu Shaoqing is the kind of person who is more sensitive. When staying with AI characters, he will instinctively feel horror, so the correct rate of identification is higher than that of ordinary people.
Dr. Qin Kang wondered if he had discovered some unknown bug and hadn't reported it.After all, Yu Shaoqing specialized in the field of artificial intelligence during his PhD.
So he will conduct a special test to try to rule out this possibility.
"Contact Dr. Qin Kang." Xie Ruihan said to the assistant, "Let him connect the picture of the virtual scene into the conference room. I want to see the test for myself."
His sharp gaze swept across many older colleagues, "Everyone must watch."
You'll Also Like
-
The Strongest Curse Master: Start by refining the Seven Arrows Book
Chapter 338 3 hours ago -
Douluo's self has a soul beast clone
Chapter 393 5 hours ago -
The Comprehensive Evolution of American Comics
Chapter 910 7 hours ago -
Doomsday: Gain experience by hunting
Chapter 366 10 hours ago -
You said you would make games by yourself, but how did you become the richest man by doing nothing?
Chapter 647 10 hours ago -
Under the Red Dragon
Chapter 374 13 hours ago -
Master Craftsman of the Pokemon World
Chapter 165 13 hours ago -
Wrongly practicing evil skills, imitating the laws of nature
Chapter 306 13 hours ago -
Demon Cultivator: Heaven and earth are the cauldron, and all living beings are the medicine
Chapter 90 1 days ago -
Super technology leader
Chapter 434 1 days ago