Essays - Benedict Evans · Benedict Evans
articleinsightThe problem of AI ethics
23 March 2024Strategy
Source excerpt
About Essays - Benedict EvansCan you write laws, or lay down ethical principles, for a technology that will be used in entirely different ways, for different purposes, in different industries? What does that mean if it’s changing entirely every 18 months?
This is a limited feed-provided excerpt, not the full original work.
Apollo layer
What a founder can learn
Founder takeaway
Treat AI ethics as a use-case-specific, continually updated practice—not a one-time set of universal principles. Evaluate each application by its purpose, industry, users and potential harms, then revisit those judgments as the technology changes.
Why it matters
Static or overly broad rules can become irrelevant as AI capabilities and applications evolve. A recurring, context-based review process helps founders make clearer decisions without assuming today’s standards will fit tomorrow’s product.
Relevant guides
Put it to work
For one AI feature you are building, ask: Who could be affected, what could go wrong in this specific context, and what trigger will force us to review our safeguards?