Theory of Change
Overview
This page documents the initial theory of change for the organisation. The purpose of the theory of change is to specify the activities conducted by the organisation, the intended effects of those activities, and the justification for the belief that those activities will bring about those effects.
The structure of a theory of change typically consists of a team or set of teams, a sphere of control, a sphere of influence, a sphere of interest, and a set of assumptions and evidence. This theory of change is subject to change on receiving updated information about the impact of the organisation.
Team
The organisation consists only of one individual at present, Alfie Lamerton. Alfie will have other roles at the organisation, but those roles are not relevant for the theory of change at this time.
Alfie will act as the research team in the proposed organisation for the time being. It is expected that as the organisation evolves, more teams will be created that contribute to the sphere of interest. Those teams will be added as new rows to the theory of change, starting with the organisation's goal and mapping the causal pathway backwards to team design.
Sphere of Control
The sphere of control contains the inputs of the organisation – the activities conducted by the research team. The research team will calculate a forecast for lock-in risk using primary and secondary data, brainstorm lock-in threat models, build an evaluation framework and benchmark for lock-in risk in frontier models, and conduct technical cooperative AI and AI control research.
Sphere of Influence
The sphere of influence contains the measurable effects of the activities conducted by the research team – these are the products of the organisation. The research activities will be channelled into publishing a lock-in risk forecast, creating threat models for lock-in and maintaining a public list of threat models, publishing a lock-in evaluation framework and benchmark, producing technical research papers, and producing reports making recommendations about lock-in risk reduction.
Sphere of Interest
The sphere of interest specifies the events the organisation is trying to elicit in the world; it contains two columns in the causal chain of events: the outcomes – the intended effects of the outputs, and the goal – the main objective of the theory of change.
Summary
This page aims to outline the purpose and operation of the organisation by decomposing its initial theory of change and explaining each element in detail. It also identifies the assumptions on which the theory of change rests, outlines the bases for these assumptions and expresses a confidence level in each of them. Lastly, the report illuminates potential next steps for the improvement of the theory of change.