This idea started given I am flying to the UK/China for 2mths, and with initial conversations with ‣.
Premise: A series of interviews of thinkers in the UK (and maybe China) on building LLMs which are fit for human interaction.
Context: Concerns of AI psychosis, young people and their addiction to platforms are ever increasing. Frontier AI companies have all pushed new articles, grants, and initiatives to reduce the possibility of this happening. The question is: What are leading builders currently doing to solve this?
Steve’s Questions
Advice: There is a step between transcription and end-written product that is non-insignificant.
Interviewees
| Name/Profile | Affiliation | Scheduled Dates | Notes |
|---|---|---|---|
| Max Rollwage | Limbic AI | 08/01/26 + 12/01/26 | Done! ‣ |
| Elliot Taylor | Throughline | Done | |
| Matthew Nour | Microsoft AI + Oxford/UCL | Done | |
| Keno Jüchems | AISI, Ex-Limbic | Contacted | |
| Kira Riehm | Anthropic | Mention recent article | Contacted |
| Daniel Reid Cahn | Slingshot AI | Mention cool podcast in cold! | Unsure → Steve intro? |
| Jade Leung | UK PM Advisor for AI, Ex-OpenAI, AISI | NZ connection? | Unsure - Her Thesis + recent interview |
| Declan Grabb | OpenAI, ex-Stanford | Read his earlier paper on SAEs | Unsure |
| Deepmind? | |||
| Tim Hua | MATS | Wrote email to him last year | Unsure |
| Australian AI Safety Institute? | Get intro via Greg Sadler? | ||
| Xuan Zhao | Flourish AI | Steve? |