AI algorithms that ingest real-world data and preferences as inputs run a risk of learning and imitating our biases and prejudices.
Performance risks include:
- Risk of errors
- Risk of bias
- Risk of opaqueness
- Risk of instability of performance
- Lack of feedback process
For as long as automated systems have existed, humans have tried to circumvent them. This is no different with AI.
Security risks include:
- Cyber intrusion risks
- Privacy risks
- Open source software risks
- Adversarial attacks
Similar to any other technology, AI should have organisation-wide oversight with clearly-identified risks and controls.
Control risks include:
- Risk of AI going “rogue”
- Inability to control malevolent AI
The widespread adoption of automation across all areas of the economy may impact jobs and shift demand to different skills.
Economic risks include:
- Risk of job displacement
- Risk of concentration of power within 1 or a few companies
- Liability risk
The widespread adoption of complex and autonomous AI systems could result in “echo-chambers” developing between machines, and have broader impacts on human-human interaction.
Societal risks include:
- Risk of autonomous weapons proliferation
- Risk of an intelligence divide
AI solutions are designed with specific objectives in mind which may compete with overarching organisational and societal values within which they operate.
Ethical risks include:
- Values misalignment risk