Resolving data barriers
Aviv H. Gaon · Edward Elgar Publishing eBooks · 2021
Chapter nine discusses data barriers to artificial intelligence. Artificial intelligence requires a massive amount of data to develop. The mere availability of data is not enough. Artificial intelligence needs good data. Incomplete or biased data can exacerbate problems. Good data, however, might be copyrighted or restricted, which poses a problem: where the artificial intelligence’s input infringes copyright, the output may infringe too. This chapter touches on the ways copyright and privacy laws might adapt to accommodate the expected changes. It discusses legal alternatives and exceptions such as fair dealing, sui generis legislation or a specific exemption for data mining and AI ‘training’.