Essays - Benedict Evans · Benedict Evans
articleinsightBuilding AI products
8 June 2024Strategy
Source excerpt
About Essays - Benedict EvansHow do we build mass-market products that change the world around a technology that gets things ‘wrong’? What does wrong mean, and how is that useful?
This is a limited feed-provided excerpt, not the full original work.
Apollo layer
What a founder can learn
Founder takeaway
Design AI products around explicit tolerances for error, not an assumption of perfect accuracy. Define what “wrong” means for a specific user and task, then focus on use cases where imperfect outputs still create meaningful value.
Why it matters
AI mistakes have different consequences across contexts. A tolerable suggestion in a creative workflow may be unacceptable in a high-stakes decision. Product adoption depends on matching the technology’s uncertainty with appropriate use cases, safeguards and user expectations.
Relevant guides
Put it to work
For your core use case, list the ways the AI can be wrong, the cost of each failure and how users can detect or recover from it. Which failures are acceptable, and what product design is needed for the rest?