Announcing Braithwate’s AI/ML risk management service
UK supervisory authorities recently published a Discussion Paper (DP5/22) about the impact of Artificial Intelligence 🤖 on the UK 💷 financial services market.
72% of UK financial services firms surveyed by BoE reported using or developing ML applications in house and expect the number of ML applications to increase by 3.5x in the next 3 years.
Given the large-scale adoption of AI tools in the financial services infrastructure, it is important to understand what this means for consumers, firms and wider markets.
Here's a quick summary of DP5/22 views of AI risks on UK financial services ecosystem.
AI risks can materialise at different levels within AI systems:
🗂 Data: AI can analyse volumes and ranges of data beyond human capabilities; however AI can also perform not as intended due to data quality issues or biased training data sets
🔍 “Blackbox” models: AI disrupts traditional rule-based models and alters parameters iteratively, making them opaque and hard to explain
🤝 Governance: the removal of human judgement in decision-making and oversight poses accountability challenges and has substantial implications for existing governance models
Sample AI risk types against UK financial services supervisory objectives: