Policy (economic/social impacts of ML)
Concrete Problems in AI Safety — inherited
An adjacent research area concerned with predicting and responding to the economic and social consequences of machine learning.
The economic and social consequences of machine learning, for labor, industries, and governance, have a research area of their own. This corpus calls it policy, following Concrete Problems in AI Safety (2016), which mentions the field in one line while marking the edge of its own technical territory.
What follows describes the adjacent field the paper declines to cover, then the boundary that one brief mention draws around the paper's own territory.
An adjacent field the paper declines to cover
Policy, in Concrete Problems' usage, names the pre-existing research area concerned with predicting and responding to the economic and social consequences of machine learning, questions of labor displacement, industry structure, and governance that predate this paper and that it does not attempt to address. The paper lists it in a single line among six adjacent research areas: "How do we predict and respond to the economic and social consequences of ML?" (concrete-problems, §"Policy:", p. 21).
Marking the edge of the paper's territory
Policy belongs to Drawing the boundary: what accident risk is not alongside Privacy (in ML), Fairness (in ML), Security (attacks against ML systems), Abuse (of ML systems), and Transparency (in ML), six fields the paper places next to its own accidents framework specifically to distinguish them from it: none of these six count as "accidents" in the paper's technical sense, since accidents are defined by unintended harm from poor design rather than social consequence, discrimination, external attack, or opacity. The paper closes this section by noting that it believes "research on these topics has both urgency and great promise" and that "fruitful intersection is likely to exist" between them and the accidents framework (concrete-problems, §"Related Problems in Safety:", p. 21), without claiming to have explored that intersection itself.