机器人如海盗

Robots as Pirates

生成式人工智能在文学和图形领域的潜力令人振奋,因为它可以创造新作品。通过分析庞大数据库(公开可用和资料开源)中既有材料元素之间的关系,作为生成式人工智能基础的机器学习技术得以运行。这些算法“学习”预测不同类型表达中“接下来是什么”来进行创作。因此,一个完整的人工智能系统可以熟练地撰写事实性总结、文章、虚构的故事以及生成图像。生成式人工智能引擎会使用一些原始材料,这些原始材料的作者主张机器学习过程中侵犯了他们的版权。对实际和可能的主张进行谨慎评估,可以发现,声称侵权的原告在证明复制、发行、展示以及制作演绎作品方面将面临困难,因此在既定的法律原则下,不太可能构成版权侵权。然而,对未经补偿的使用行为的担忧,可能会促使围绕原创内容增设更多付费屏障,并催生更多的集体许可。

Generative Al has created much excitement over its potential to create new works of authorship in the literary and graphical realms.Its underlying machine learning technology works by analyzing the relations among elements of preexisting material in enormous databases assembled from publicly available and licensed sources.Its algorithms“learn”to predict“what comes next”in different types of expression.A complete system thus can become glib in creating new factual summaries,essays,fictional stories and images.A number of authors of the raw material used by Generative Al engines claim that the machine learning process infringes their copyrights.Careful evaluation of actual and likely claims can show that such plaintiffs claiming infringement will have a hard time proving reproduction,distribution,display,or preparation of derivative works and thus are unlikely to be able to establish copyright infringement under established doctrines.Fears of uncompensated appropriation,however,are likely to fuel the erection of more pay walls around original content and more licensing collectives.