Engineering Intent and Intent Measurability
summary
Claims that intent is fully measurable give way to specific practices: treat intent measurability as approximate, stop flattening decisions to event sequences, and design for concurrent feelings and context. Recommendations include process ambiguity as a feature, set goals to elicit emotions not to predict certainty, offer explicit opt‑in intent buckets (example: Yelp), and use on‑device inference like TinyML and local-first voice processing (iOS 15 example) to reduce distant profiling.
AI-written summary · about
chapters (7)
- 0:00The challenge of intent
- 2:50Intent in action
- 5:27Limitations of data metrics
- 8:48The human element
- 12:05Challenging industry habits
- 15:43Embracing ambiguity
- 20:33Q&A and practical insights
related talks
30mMeasurement in a Privacy First World
01h 04mHuman Intent and Text Understanding
41mIf We Want AI to be Interpretable, We Need to Measure Interpretability
32mJust ask: designing intent-driven algos- 30mMeasurement in a Privacy First World
39mMind-Reading with Adaptive and Intelligent User Interfaces
25mWhen design meets machine learning
40mMeasuring the Ethical Behavior of Technology