The Dual LLM pattern for building AI assistants that can resist prompt injectionsimonwillison.net·201 pts·simonw·109
The Dual LLM pattern for building AI assistants that can resist prompt injectionsimonwillison.net·6 pts·simonw·0
We need to tell people ChatGPT will lie to them, not debate linguisticssimonwillison.net·546 pts·simonw·457
I built a ChatGPT plugin to answer questions about data hosted in Datasettesimonwillison.net·6 pts·simonw·3
Don’t trust AI to talk accurately about itself: Bard wasn’t trained on Gmailsimonwillison.net·5 pts·simonw·2
Could you train a ChatGPT-beating model for $85,000 and run it in a browser?simonwillison.net·7 pts·simonw·1
I promise ChatGPT can’t access the internet, even though it looks like it cansimonwillison.net·4 pts·simonw·0
AI chatbot “Sidney” is misbehaving – Nov 23 2022 Microsoft community threadlogin.microsoftonline.com·1 pts·simonw·3
Exploring MusicCaps, the training data behind Google’s MusicLM text2music modelsimonwillison.net·1 pts·simonw·0
Semantic search answers: Q&A against documentation with GPT3+OpenAI embeddingssimonwillison.net·2 pts·simonw·0
Increased productivity on personal projects with comprehensive docs and testssimonwillison.net·2 pts·simonw·0