Sundar Pichai himself, CEO of Google, has published a long post on the company’s official blog to illustrate what are the guidelines that Google intends to follow regarding the development of artificial intelligence, summarized in 7 points. The AI, as Google understands it, should therefore be:
- Socially beneficial
- Avoid creating or strengthening disparities
- Be built and tested with security in the center
- Being subject to appropriate human control
- Incorporate privacy protection principles
- Support high standards of scientific excellence
- Be available for uses that are in accordance with these principles
Each of these points is further clarified by Pichai, but what is almost more important than what the Google AI must do, is what it should not do. The CEO of Google has in fact committed not to develop an AI in the following areas:
- Technologies that cause or can cause generic damages.
- Weapons or other technologies its main purpose is to hurt people.
- Surveillance technologies that violate international standards.
- Technologies whose purpose violates principles widely accepted by international laws and human rights.
As we mentioned a short time ago in our editorial, AI is one of the hottest topics of the moment, and it will certainly be for years to come. The perception that users have today is perhaps a little ‘debased by the continuous use (not to say abuse) of this term for trivial things, but the artificial intelligence referred to Pichai is not the one that allows you to recognize if you are framing an apple or an apple tree, but something more complex and potentially dangerous.
These “7 laws” (parallelism with Asimov – NDR) come at a non-random moment. In the last period there have been many controversies regarding the involvement of Google with a military project of the Pentagon; a project which envisages “the use of artificial intelligence to analyze images of remotely piloted aircraft and identify targets”. Google would, therefore, have decided not to renew the contract, expiring in 2019, but the words of Pichai sound like further reassurance in this regard, rather than as a general guideline. Find the full post in the link to the source, if you would like to deepen the topic.