Eventually, however, HudZah wore Claude down. He filled his Project with the e-mail conversations he’d been having with fusor hobbyists, parts lists for things he’d bought off Amazon, spreadsheets, sections of books and diagrams. HudZah also changed his questions to Claude from general ones to more specific ones. This flood of information and better probing seemed to convince Claude that HudZah did know what he was doing, and the AI began to give him detailed guidance on how to build a nuclear fusor and how not to die while doing it.
Recent articles
- Model Context Protocol has prompt injection security problems - 9th April 2025
- Long context support in LLM 0.24 using fragments and template plugins - 7th April 2025
- Initial impressions of Llama 4 - 5th April 2025