How to Build Ethical and Reliable AI: Learn from Search Engines, Demand Transparency, and Approach with Healthy Skepticism
-
Learn from search engines - use ranking techniques and favor trustworthy data sources to build reliable generative AI apps
-
View large language models as an interlocutor, not an oracle - they should not be the canonical source of knowledge
-
Employ a layer to determine confidence in answers - and disclose when high confidence can't be reached
-
Demand explainability and transparency - show sources of information so users can evaluate trustworthiness
-
Approach AI tools with educated skepticism - look for signs of bias and lack of explainability