Eudoxa Policy Study #5:
Autonomy, Loyalty and Trust in the World of Machines
We are surrounded by machines that are increasingly autonomous, although
the autonomy we have managed to create is far from robust or comprehensive.
Simple devices such as automatic doors might be broken. In other instances
the automation will behave wrong, because it does not interpret the input
correctly. Or the automation will behave according to specification, but
the interaction between the machine and the human will produce unexpected
behavior. Who is to blame when autonomous technology fails?
We are used to our tools being loyal to us, but will that always be the
case in autonomous machines? Our computers provide numerous examples of
machines that communicate with its creator without the knowledge or consent
of its user. Malicious code has even sent people to jail until it could
be proved that the software acted on its own accord. The question of who
an autonomous machine is loyal to is eventually a question about security.
It all boils down to the eternal issue of trust. The author looks at
trouble with current Trusted Computing solutions, such as monopolization
and less dynamic markets.
Increased autonomy leads to reduced trust and loyalty. Trust is something
you earn, machine or human. Increased transparency in the ability of a
machine is one of several keys to the solution. Another one is to redefine
how we look at machines, the author suggest that we could gain from considering
them as, and treating them as pets. They are not completely in our control,
but we have a certain degree of authority over them based on a mutual
relationship. He concludes the study this way: "The problems are
stupid machines, stupid laws and stupid people. Let us try to fix all
three of them!".
Download
the PDF-file here >>
Table of Contents
- About Eudoxa
- Executive Summary
- Autonomy
- Loyalty
- Trust
- Autonomy, Loyalty and Trust - all together now!
|