My guide to identifying AI risks: Cheat sheet + practice
Plus: Apply to LocalLingua from my AI Ethics Game!
Dear AI ethics enthusiasts,
Identifying ways that AI can cause harm, i.e. AI risks, is usually more challenging than it seems. In my consulting work and research, I often hear from companies that they feel they know which AI risks are relevant to them. But when we dig just a bit deeper we uncover risks they hadn’t thought of.
So today, I am sharing two approaches to risk identifications and a cheat sheet to help with my favorite. Then, for a hands-on exercise, I will guide you through assessing risks in a realistic case study: LocalLingua, a (fictional) language learning platform.
For those who want more, next week I am holding a webinar for learning and practicing risk identification, governance assessment, and guardrails, also practicing on LocalLingua. Register here, free for you lovely paid subscribers!
Enjoy!
Two approaches for identifying AI risks
Two approaches have proven particularly effective:
The Red Team Exercise
Ask "What could go wrong?" but make it specific. For example:
If s…