Context Week
A 4 day residential program for people who want to understand the AI safety field and find their highest impact path.
Cohort
Around 20 people
Dates
Aug 30–Sept 4, 2026
Location
Berkeley
Covered
Travel, accommodation, meals
About Context Week
Context Week brings together around 20 people who are considering work in AI safety. Participants study how the field developed, the risks people are concerned about, and the work being done in response. They also compare different theories of change and meet people working in several parts of the field.
The programme includes seminars, 1:1 conversations, sessions with guests, and time with other participants. Sessions examine the evidence and assumptions behind different views. Participants can use this material when applying for roles or programmes and when choosing between jobs, training, or projects.
Participants arrive on August 30. Programming runs from August 31 to September 3, and participants depart on September 4. Before arriving, participants complete about ten hours of reading selected by the programme team.
Participants
We are looking for people with a strong record of study or work who are seriously considering contributing to AI safety. They may be applying for a job or training programme, choosing a research area, developing a project, or moving into AI safety from another field.
Applicants may still be considering several options. They should be able to explain what they want to understand about AI safety and how it could affect work they are considering. We also expect them to discuss disagreements, say when they are uncertain, and change their mind. Relevant experience in other fields is useful.
Context Week and Incubator Week
Context Week is for people who are still working out which part of AI safety to focus on and where their skills would be useful. They may be choosing between problems, roles, organisations, or project ideas. Some mainly need a firmer view of the field before making that choice.
Incubator Week is for people who have already chosen a specific intervention and know who it is meant to help. The programme gives them time to test an important assumption about that intervention.
Programme
Most scheduled time will be spent in facilitator-led seminars, generally in small groups. Participants will also have 1:1 conversations and sessions with guests who work in AI safety. Meals and evenings leave time for less structured conversations.
The detailed schedule is still being developed. Sessions will cover the history of AI safety, current risks and responses, disagreements between different approaches, the work of existing organisations, and how participants might use what they learn when choosing their next steps.
After Context Week
By the end of the programme, participants should be able to explain the main concerns about AI risk, the work being done in response, and why people support different approaches. They will have considered which roles or projects suit their skills and how to explain their interests and reasoning in applications or interviews.
Each person will record what they plan to do next. Their next steps may include applications, conversations, further reading, or project work. They may make a decision during the week or leave knowing what they still need to learn.
BlueDot will assess participants' knowledge before and after the programme and contact them 90 days later to ask what they did next. The evaluation will also look for overconfidence and pressure to agree with the group.
About BlueDot
BlueDot Impact is a nonprofit building the workforce and organisations needed to safely navigate AGI. We've raised over $35M and trained over 8,000 people since 2022.