https://mp.weixin.qq.com/s/97QNPbJkeUOh6nQj1ymy6A
By 超神经
场景描述：人工智能创作的作品是否拥有版权？这些作品版权归属于谁？而人工智能犯错后果又该由谁承担？众多问题亟待法律法规的完善。
关键词：人工智能 版权 世界知识产权日 Have you noticed that in recent years, domestic copyright awareness has begun to awaken.
Nowadays, many works on various music and video platforms require paying members to listen and watch. Creators have "zero tolerance" for piracy, and the country's laws and regulations in this area are constantly strengthening and improving.
However, with the development of artificial intelligence, some new questions have begun to appear, such as: Does artificial intelligence that can create works have copyright?
Today, April 26, the 19th World Intellectual Property Day, we will discuss the issue of artificial intelligence and copyright together.
Copyright storm caused by black holes
Not long ago, the black hole photos triggered a wave of public opinion about copyright. The photo was taken by Visual China as soon as it was released, and the copyright watermark was added.
Visual China included the black hole photo in its own copyright library on the night it was published
The end result is that Visual China apologizes, and the black hole photos, because behind the efforts of digital astronomers, software engineers, astronomical telescopes and computers around the world, belong to cooperative works and cannot be divided, so they are not protected by copyright law. Its copyright is recognized as belonging to All mankind.
The copyright turmoil of black hole photos has subsided, but from this we thought of artificial intelligence that can create works independently. Whom should the copyright of the works they create belong to?
Does the work created by AI have copyright?
Since artificial intelligence began to emerge in the field of cultural creation, the copyright ownership of their works has been controversial. With the development of related technologies, the independent creative ability of artificial intelligence is getting stronger and stronger.
In 2016, artificial intelligence developed in Japan created the science fiction novel "The Day When Computers Wrote a Novel", which not only deceived all human judges, but also successfully entered the Japan Micro Literature Award.
In May 2017, the poetry collection "Sunshine Lost the Glass" created by Microsoft's artificial intelligence system "XiaoIce" was officially published.
In April 2018, Korea used artificial intelligence technology to participate in song production and succeeded for the first time, and independently developed "K-pop AI" that can sing.
In the past, Google artificial intelligence was able to create paintings, and some paintings were even sold at a high price of $8,000.
The day before yesterday, the first book written by artificial intelligence was published, titled "Lithium Ion Batteries: A Summary of Cutting-Edge Research in Machine Generation".
Such examples are too numerous to list. Artificial intelligence creation is nothing new. But for these works, to whom does the copyright belong? Who should own the manuscript fee and auction fee?
Some people think that these works cannot be considered as creations, but can only be said to be calculated, or that they are imitated and pieced together, so there is no copyright issue;
Another point of view is that artificial intelligence with natural language understanding and generation as its core technology has the ability to generate original language. Although the work is still far from the master, it is also a creation, and artificial intelligence should own the copyright.
The world is inconclusive
In this regard, various countries have different legal regulations and measures.
United Kingdom: As early as 1988, the "Copyright, Design and Patent Act" promulgated by the United Kingdom recognized that "computer-generated" works are works without "human authors" and explicitly granted copyright protection to such works.
Europe: In February 2017, the European Parliament advocated granting autonomous robots the legal status of "cyborgs" for the purpose of copyright protection.
Japan: In 2016, Japan adopted the Intellectual Property Promotion Program, the highlight of which is to establish a legal system for the protection of “copyright” of artificial intelligence products. The Japanese government stated that it will establish a new registration system through this plan to protect the copyright of artificial intelligence works.
China: my country's "Patent Law" is also gradually liberalizing patent protection for computing people's "algorithms", but for patent applications for artificial intelligence products, the "Patent Examination Guidelines" still need to be further improved and supplemented.
United States: According to the intellectual property clause in the US Constitution, the constitution maker did not include copyrighted works created by robots in this clause, because machines simply act according to program instructions without any material rewards.
Today, such discussions are still inconclusive on a global scale.
Who gains the glory of AI, and who bears the blame for causing trouble?
As we all know, with deep learning, neural network and other technical training, artificial intelligence becomes smarter and smarter, and it may do things that "guardians" would not expect.
Take the previous Uber self-driving car accident as an example.
In March 2018, a woman crossing the road was knocked down by an Uber driverless SUV and eventually died. This was the first fatal accident caused by an unmanned vehicle hitting a person, which caused an uproar at the time.
Finally, after investigation, the responsibility lies not with Uber, but with the victim. Uber does not bear criminal responsibility, and the back-up driver in the car needs to be further investigated.
In the same year, Tesla's unmanned driving caused another accident in which the autopilot system caused death. Eventually, the traffic police determined that the driver was primarily responsible.
In these two accidents, the autonomous driving system was not assigned responsibility. However, if one day, autonomous driving is completely unmanned, and if an accident occurs, who should be responsible? This is an unsolvable problem so far.
Similarly, if works created by artificial intelligence are awarded and sold, who will benefit?
The copyright issue of artificial intelligence still needs to be improved
In the face of more and more artificially created content and other products, copyright-related issues need to be resolved by clear laws and regulations.
If the content created by artificial intelligence has the appearance of a "work" and thus constitutes a "work" in the sense of copyright law (copyright law), the copyright system must respond to how to define the ownership of the artificial intelligence work.
If the rights of artificial intelligence works are not clearly defined and allowed to become "ownerless works", it will cause a lot of legal disputes, impact the existing copyright system, and even cause a moral crisis that "piracy" cannot be controlled.
Since the birth of the copyright system, every major breakthrough in science and technology has been accompanied by profound changes in the copyright industry, and the copyright system has also been adjusted accordingly.
In terms of intellectual property protection, we still have a lot to do. On the issue of AI's intellectual property rights, there is also much to be improved.
HyperNeuropedia
Ensemble learning
Ensemble learning is an idea in machine learning, through the combination of multiple models to form a higher accuracy model.
Integrated learning itself is not a single machine learning algorithm, but to complete the learning task by constructing and combining multiple machine learning devices.
Integrated learning can be used for classification problem integration, regression problem integration, feature selection integration, outlier detection integration, etc. It can be said that integrated learning can be seen in all machine learning fields.
There are two main problems to be solved in ensemble learning. The first is how to obtain several individual learners, and the second is how to choose a combination strategy to assemble these individual learners into a strong learner.
