How to Regulate Artificial Intelligence
Bill Hibbard Sept 2017
I submitted the following letter to the editor of the New York Times on 2 Sept 2017, although they did not print it:
- Oren Etzioni (How to Regulate Artificial Intelligence, 2 Sept 2017) proposes three rules for regulating A. I. The first two, that A. I. must obey laws that apply to humans and A. I. must disclose that it is not human, are good as far as they go.
- For the deep implications of A. I. the third rule is most important, that A. I. cannot retain or disclose confidential information without explicit approval from the source of that information. However, in order to gain the benefits of A. I. people will need A. I. to know all about them and the people they interact with. But this same information will enable A. I. to manipulate people, as for example Facebook uses information from our pages to target advertising. The third rule should give humans control over how their information is used. But for this, people need to know how their information is being used. And that requires disclosure of the purpose and means of A. I. systems.