Page 53 - ITGC_Audit Guides

 Improved employee morale – employees may be freed from conducting repetitive tasks.
 Productivity – automating simple tasks allows employees to increase productivity in other
areas.

 Reliability – with proper programming, RPA may produce more dependable results.
 Consistency – bots can be programmed to work nonstop and perform repeatable processes,
ensuring consistent results over time.
 Non-invasive technology – disruption to existing systems isn’t an issue.
 Compliance – audit trails can be documented to satisfy regulatory requirements.
 Low technical barrier – configuration is relatively simple.

 Accuracy – bots are less prone to human error.

Risks may include, but are not limited to:
 Segregation of duties issues – bots may have excessive authority.
 Poorly scripted processes – as with any computer program, attention must be paid to what
the bot is being requested to do.
 Existing process not improved before being automated – if a process was flawed before
automation, simply transferring the same rule set to an automated program will continue to
produce flawed results.
 Poor monitoring of bots and administrators – though automated, bots need occasional
maintenance, and administrators should be kept apprised of new processes, compromised
output, etc.

 Cyberattacks – anything in the IT environment is subject to cyber issues. Bots are no
exception.

Machine Learning and Artificial Intelligence

Cognitive automation combines advanced technologies such as natural language processing
(NLP), artificial intelligence (AI), machine learning (ML), and data analytics to mimic human
activities such as inferring, reading emotional cues, reasoning, hypothesizing, and communicating
with humans.

The value goes beyond the ability to automate business processes; cognitive automation may
also serve to augment what humans do, making employees both more informed and more
productive. Within cognitive automation, there is an important difference between learning and
reasoning. Learning is about recognizing patterns from unstructured data and the correlated
automation is based on accuracy ratings. In contrast, hypothesis-based reasoning is based on
confidence ratings.

Risks related to cognitive automation include but are not limited to:

 Bad practices can be interpreted as acceptable by AI.
 Poor understanding by designers is reflected in systems.

45 — theiia.org