From Science Fiction to Garage Science: My Journey Building a Farnsworth Fusor with AI
summary
Claims building a Farnsworth Fusor in weeks with AI help and getting detectable fusion events while stressing it is a low‑yield demo not a power source. Workflow uses Claude Cloud Projects with scraped fuser.htm turned into embeddings, plus vision prompts to parse an MKS 91P vacuum transducer datasheet, 12,000 V supply and vacuum pump data, and voice dictation via Whisper. Logs, invoices and expert chats feed the project; safety hazards (lethal high voltage, radiation monitoring) are treated but remain central lessons.
AI-written summary · about
chapters (14)
- 0:00Introduction and philosophy
- 2:24Using embeddings for research
- 6:01Sourcing and validating data
- 7:02Prompting and Claude projects
- 9:13Speedrunning project ideation
- 13:37Visualizing technical data
- 16:02Automating life and school
- 18:48Energy impact and future tools
- 22:53The future of search and AI
- 27:12Voice integration and typing
- 32:24Efficiency and AI limitations
- 35:24Safety and ethical concerns
- 46:35Technical deep dive: The fusor
- 49:24Closing thoughts and Q&A
related talks
37mBuilding a laser for a scientific research facility - without a playbook
30mDevRelling in the age of GenAI
21mHacking AI for creative operations at scale
36mBuilding A Silicon Analysis Toolkit
42mTinkerer’s Toolkit
06mMETASAT Demonstrator: Mixed Criticality, Accelerated AI Computing for Future Space Systems
13mFaster AI time-to-value when planning
36mArt in the Age of Algorithms