Communications of the ACM
Risks of Automation: A Cautionary Total-System Perspective of Our Cyberfuture
Credit: chombosan
Many computer-related risks discussed in past Inside Risks columns are still present today. These risks (and new ones) are likely to intensify even further as systems provide extensive automated or semi-automated operation. Significantly greater total-system trustworthiness will be required, encompassing better hardware, system software, and applications that are able to tolerate human limitations and environmental factors. Risks will continue to result from inadequate reliability, security, and privacy, as well as gullibility and general inability of users to cope with complex technology. We repeatedly discover unexpected risks resulting from lashing subsystems together (for example, see Beurdouche2), because of unexpected system behavior. Many advances in research, system development, and user friendliness are urgently needed. Also, some middle ground is desirable between the optimists (who believe there are easy answers to some of the problems posed here) and the pessimists (who have serious doubts about increasing uses of automation and artificial intelligence—especially when used by people who are more or less technologically queasy).
In this column, I examine certain approaches that might be economically desirable, but that have serious potential risks. These include aviation safety and security; self-driving and semi-automated vehicles, and eventually automated highways; the so-called Internet of Things; and cloud computing and cloud storage.
No entries found