Daniel Fitzpatrick & Maria Müller | AI Track | Compass AI & Data Summit 2025
summary
Presents a concrete method called Pragmatic Metacognitive Prompting that makes LLMs run a three‑step pipeline—preliminary analysis, reflection, map to a computational return value—to detect implicature and sarcasm. Shows practical tooling: chaining prompts, extracting JSON, and mapping scalar implicature adjectives to numeric ranges. Deployments include a conversational assistant and a digital human tied to a vehicle options database that negotiates buildability conflicts, addressing a reported 30% drop per conflict and clearing legal display rules so dealers accept the system.
AI-written summary · about
chapters (10)
- 0:00AI and sarcastic communication
- 2:58Linguistics and conversational nuances
- 7:29Humans vs computers communication
- 10:45Grice’s cooperative principle in AI
- 14:37Detecting sarcasm and implicature
- 17:50AI challenges with convoluted speech
- 20:19Conversational AI in automotive
- 26:52Value from unstructured speech
- 29:19Prize announcement
- 30:23Q&A on voice interfaces
related talks
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
19mUnlocking the Full Potential of LLMs: Advanced Prompt Engineering Techniques
14mEdge of the Present: Pursuing AI’s Frontier in the Enterprise
40mBeyond the Hype: Realistic Expectations of AI
26mMastering Large Language Models: Prompt Engineering for Precision & Performance
50mFrom Prompt Hacking to Architectural Determinism: Engineering Reliable GenAI Systems
43mBuild and Maintain your Company Copilot with Azure ML and GPT-4 Q&A
01h 08mIn Prompts We Trust