Page 55 - the SyI Quarterly 14 Booklet Format

Chartered Security Professionals

Lack of Human Oversight: AI algorithms can make decisions
Finally, AI can assist security professionals by automating repetitive without human intervention, which can be concerning in certain
and manual tasks, allowing them to focus on more critical tasks, contexts where human judgment is critical leading to concerns
and reducing the time and resources required to do so. For about fairness and accuracy. For example, It was found that
instance, the creation of risk assessments, a process which an AI system used by the US justice system to make bail and
often involves an element of “best guess” over solid statistics, sentencing decisions, tended to reinforce racial bias in law
has seen a great deal of AI investment within areas such as enforcement data by falsely flagging black defendants as future
fintech and insurance, where fraud detection plays a large part criminals at almost twice the rate of white defendants.
in the risk assessment process. Similarly, extensive research
and development is being conducted in the field of automated Technical Complexity: AI algorithms are often highly complex and
intelligence reporting, such as using feeds from satellite imagery difficult to understand, making it difficult for humans to assess
and street cameras to quickly analyse data. AI-generated risk their decisions and outputs. This can limit the accountability of AI
management systems created with the right quantity and quality of systems, making it difficult to determine who is responsible for
data sources can be highly effective and valuable to both corporate their outcomes.
and national security.

In conclusion, AI has had a significant impact on the field of
However, AI is still an emerging technology and many security security, and this impact is only going to grow in the future. AI
professionals are sceptical of its application with concerns can assist security practitioners in many ways, including physical
regarding the negative consequences of its development and and cyber security, actionable intelligence, biometric security,
deployment, including: incident response, and task automation. With AI, security

practitioners can work more efficiently, respond more effectively
Bias and Discrimination: AI algorithms can perpetuate and amplify to threats, and ultimately create a safer environment for people,
existing biases in society if they are trained on biased data. This infrastructure and assets.
can result in discriminatory outcomes, such as racial or gender
discrimination. In 2015 an Amazon AI-based system used to review However, it also carries significant risks and challenges that must
job application forms was scrapped after it emerged that due to be carefully considered and addressed. It is important to ensure
the majority of CVs being submitted by men, the system taught that the development and deployment of AI technologies are
itself to favour male applicants, it started downgrading applications guided by ethical principles and that appropriate measures are
with wording like “women’s rowing club”. Studies have also shown put in place to mitigate their negative consequences.
that AI used for facial recognition is much less accurate when
analysing black or Asian faces. Ultimately, AI will have many uses that will be valuable to threat
actors which we will have to mitigate. As security professionals,
Privacy Concerns: AI algorithms often collect and process vast it is crucial that we comprehend AI and stay updated with its
amounts of personal data, raising concerns about privacy and the advancements as both an aid and a threat.
security of this information.

Job Displacement: AI algorithms have the potential to automate
many jobs, leading to significant job displacement and
unemployment. This could have a devastating impact on workers
and communities, particularly in lower-skilled occupations.

54 55