Cooperative Design Optimization through Natural Language Interaction
summary
A system combines batch Bayesian optimization with a large language model to let designers give outcome-focused natural language directives and receive a chosen parameter set plus the LLM’s rationale. Two user studies use relative hypervolume to measure optimization and NASA-TLX for workload: the method yields higher agency than a system-led BO baseline, matches manual design agency, produces better median optimization than manual (but below pure system-led), and equals an explicit-constraint cooperative method while reducing workload.
AI-written summary · about
chapters (8)
- 0:00Introduction and background
- 1:27Cooperative design optimization
- 2:51Technical background
- 4:16Natural language integration
- 5:19Study one results
- 6:44Study two comparison
- 8:11Conclusion and findings
- 8:53Q&A session
related talks
01h 04mBuilding Expertise Into Generated Conversations
01h 04mBuilding Expertise Into Generated Conversations
41mPlanes, Trains, and Skateboard Shoes - Bayesian methods in engineering and product design
14mUnconference: Self-optimizing deep research agents
26mMastering Large Language Models: Prompt Engineering for Precision & Performance
51mOrchestrating Large Language Models (LLMs) with CrewAI
42mKeynote: Designing LLM Native systems
24mBring the Power of AI to Your Application