The bootcamp is free! Food, accommodation and teaching is all provided at no cost to the participant. Additionally, travel costs can be reimbursed if that would otherwise stop you from attending.
English
There are short breaks throughout the day, and the time from dinner onwards is free time. However, the course is intensive, so don't plan on having time to do something else during the camp.
We expect participants to have some familiarity of the major risks from AI (e.g. misuse from bad actors, extinction risks) and a rough overview of some proposed solutions. We provide a prerequisite reading list to ensure everybody has the same base knowledge to make the most out of the camp.
For the technical bootcamp, you need some experience of coding. All coding is done in Python. We assume you have enough familiarity to work through the introductory PyTorch tutorials, Inspect tutorials, and the OpenAI/Claude tutorials for their APIs. For the governance and strategy bootcamp, you need no coding experience, but it helps to understand some basic principles of computer science.
We assume you have enough understanding of differentiation and linear algebra to understand how a vanilla neural network works, e.g., you can follow 3Blue1Brown’s explanations of neural networks. We provide a prerequisite reading list to ensure everybody has the same base knowledge to make the most out of the camp.
No, our technical bootcamp has some overlap with ARENA in terms of content, but focuses a lot more on developing a strong contextual understanding of the AI Safety landscape, building the knowledge and motivation to start a career in AI Safety, and includes a 2-day independent project.
The ideal candidate is somebody who wants and is committed to contributing to AI safety in a substantial way, whether full-time as a job, as a component of their current work, or as a side project in their spare time. AI is going to impact all parts of society and will require expertise from all fields, so there is no single profile that we are looking for.