Artificial Intelligence (AI), has been met with apprehension from experts in the field of cyber risk. Seeking to shed light on their concerns, I conducted a survey of guests who attended this year’s International Security Forum to obtain their views on the most pressing issues surrounding AI.
Expert’s 5 biggest concerns about AI
1. Unclear Legal Obligations
Risk specialists wanted clarity on the legal obligations related to AI, especially for high risk use cases. Defining accountability and liability in case of adverse outcomes remains a challenge.
2. Securing AI from Cyber Threats
Cybersecurity is the central focus for ISF members, and with the increasing integration of AI, safeguarding AI systems from potential cyber threats is paramount. Experts emphasised the urgency to understand ways in which AI could be tricked or spoofed and with that concern how cyber attacks could be detected, monitored and mitigated.
3. Interpreting AI Decisions
AI algorithms, particularly those using deep learning and neural networks, often generate decisions that are difficult for people to fully understand. This lack of interpretability raised concerns about the transparency and accountability of AI-driven decisions.
4. Incorporating Bias into Decision-Making
Hand-in-hand with lack of interpretability is concerns about the potential for AI to perpetuate and amplify existing biases in datasets. Experts highlighted the need to be able to monitor, identify and mitigate bias to ensure AI is equitable and fair.
5. Compliance with Regulations
Experts expressed difficulty keeping track of the various regulatory frameworks. Ensuring compliance with existing and future regulations is seen as a significant challenge by risk experts, who advocate for robust governance mechanisms.
Who responded to the survey?
The survey captured a range of perspectives from experts in the field representing different industry sectors. They used AI in a variety of domains for various purposes.
Addressing their concerns will be pivotal in harnessing the benefits and potential of AI safely and responsibly.
