AI Trustworthiness Challenges and Opportunities Related to IIoT
which, once properly assembled, result in
the privacy violation. AI systems may
increase the ability to identify and leverage
such data to compromise privacy.
AI AND R ELIABILITY
AI can improve reliability of systems through
the use of predictive analytics such as AI
prediction of required maintenance and
detection
of
earlier-than-expected
malfunctions (for example, hard disk
S.M.A.R.T. (Self-Monitoring Analysis and
Reporting Technology) AI analysis to
recommend replace RAID hard disks before
they fail). An example of this is the
electronic-model surveillance system of
ABB 19 that contributes to improved
reliability of systems by drawing on external
resources outside of traditional control
systems, such as sensors of vibration to
predict engine failure.
IoT processes that involve exchanges of
Personal Data between Data Subjects, Data
Controllers and Processors 17 must be
designed to cater for the requirements of
data privacy laws 18 . Privacy by Design is a
term that refers to an approach in systems
engineering that calls for privacy to be taken
into account during the design stages of
system and throughout the lifecycle of that
system. Privacy relies on Data Security
mechanisms as well as well-established
principles such as Data Minimization, Data
Anonymization and Data Pseudonymization.
AI AND R ESILIENCE
The role of AI in privacy is in the analysis of
IoT sensor data (in-motion, at-rest, in-use) to
detect situations that constitute violations of
data protection requirements. This is critical
with data privacy laws like the GDPR where
tight notification and reporting windows (72
hours) are mandated for data privacy
violations.
Resilience is defined in the Industrial
Internet Consortium Vocabulary 20 as the
“ability of a system or component to
maintain an acceptable level of service in the
face of disruption”. This means avoiding
complete failure and maintaining some
service, even if reduced from the optimal
level.
The use of AI can greatly augment the
organization’s ability to meet this important
objective.
According to the UK Government guide to
improving the resilience of critical
infrastructure, 21 there are four aspects of
17
GDPR terminology https://eur-lex.europa.eu/legal-
content/EN/TXT/PDF/?uri=CELEX:32016R0679&qid=1473816357502&from=en
18 GDPR (EU), CCPA (California), PIPEDA (Canada)
19 https://new.abb.com/future/smartsensor
20 https://hub.iiconsortium.org/portal/Glossary/59ad7c93c36c57490631d340
21
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/61342/natural-hazards-
infrastructure.pdf
- 87 -
June 2019