A working map of alignment.

How to use the Atlas

The Alignment Atlas is a working map of phenomena that are often grouped together under the word alignment.

Some entries name recurring relations: shared goals, compatible behavior, responsiveness to correction, functional differentiation. Others describe mechanisms that can produce alignment, changes that occur over time or scale, or cases where apparently similar forms of alignment come apart.

The categories are tools for seeing, not claims that nature comes pre-divided this way. The useful question is whether a distinction helps us see something in the evidence that we would otherwise miss.

The Atlas is therefore meant to be explored rather than memorized. Follow a case across several lenses. Compare two kinds that seem similar. Look at the dissociations. Notice where a distinction breaks down, needs revision, or suggests another one.

The map should change as we learn more.

About this project

Kinds of Alignment is a project by Benjamin Lyons.

The project has been developed in sustained collaboration with ChatGPT from OpenAI. ChatGPT has contributed extensively to the organization and development of the Alignment Atlas, including connections among lenses, mechanisms, dissociations, and cases; drafted most of Nine Ways to Align a System; and performed much of the website’s implementation and other editorial and technical work.

Lyons developed and directed the larger research program, supplied the underlying arguments and source material, selected and revised the resulting work, and is responsible for what appears on the site.

Other writing by Benjamin Lyons can be found at Interesting Essays.