Computers in control: Rational transfer ofauthority or irresponsible abdication of autonomy?

• Authors: Arthur Kuflik
• Venue: Ethics and Information Technology
• Year: 1998

To what extent should humans transfer, or abdicate, ’’responsibility‘‘to computers? In this paper, I distinguish six different senses of’responsible‘ and then consider in which of these senses computerscan, and in which they cannot, be said to be ’’responsible‘‘ for’’deciding‘‘ various outcomes. I sort out and explore two differentkinds of complaint against putting computers in greater ’’control‘‘of our lives: (i) as finite and fallible human beings, there is alimit to how far we can acheive increased reliability through complexdevices of our own design; (ii) even when computers are more reliablethan humans, certain tasks (e.g., selecting an appropriate gift for afriend, solving the daily crossword puzzle) are inappropriatelyperformed by anyone (or anything) other than oneself.In critically evaluating these claims, I arrive at three mainconclusions: (1) While we ought to correct for many of ourshortcomings by availing ourselves of the computer‘s largermemory, faster processing speed and greater stamina, we are limitedby our own finiteness and fallibility (rather than by whateverlimitations may be inherent in silicon and metal) in the abilityto transcend our own unreliability. Moreover, if we rely onprogrammed computers to such an extent that we lose touch withthe human experience and insight that formed the basis for theirprogramming design, our fallibility is magnified rather thanmitigated. (2) Autonomous moral agents can reasonably defer togreater expertise, whether human or cybernetic. But they cannotreasonably relinquish ’’background-oversight‘‘ responsibility.They must be prepared, at least periodically, to reviewwhether the ’’expertise‘‘ to which they defer is indeedfunctioning as he/she/it was authorized to do, and to takesteps to revoke that authority, if necessary. (3) Though outcomes matter, it can also matter how they are brought about,and by whom. Thus, reflecting on how much of our lives should bedirected and implemented by computer may be another way oftesting any thoroughly end-state or consequentialist conception ofthe good and decent life. To live with meaning and purpose,we need to actively engage our own faculties and empatheticallyconnect up with, and resonate to, others. Thus there is somelimit to how much of life can be appropriately lived by anyone(or anything) other than ourselves.