Abstracted Away: Resisting Alienation and Ungrounded Abstraction in AI Research Communities

Abstraction in AI research creates distance from real-world harms: The paper argues that dominant practices in computational AI research prioritize abstract standards of legitimacy that become separated from the lived experiences of people affected by AI systems, allowing researchers to overlook concrete harms their work may cause.

Career incentives discourage critical reflection: The authors identify how professional demands and advancement pressures in AI research communities crowd out the time and space needed for researchers to critically examine the societal impacts of their work, even when they recognize potential problems.

Marginalized researchers face systematic alienation: Early-career researchers from underrepresented backgrounds experience being simultaneously overlooked and tokenized in AI research spaces, creating a form of alienation that mirrors the abstraction mechanisms embedded in the field itself.

"Affect abstraction" suppresses emotional and ethical responses: The paper identifies "affect abstraction"—the suppression of emotional and ethical dimensions of research—as a particularly high-leverage mechanism through which AI research communities maintain distance from the human consequences of their work.

Existing efforts to address these issues remain insufficient: Despite prior academic and community-oriented attempts to recontextualize AI research practices and challenge abstraction norms, exposure to sociotechnical harms and epistemic injustice continues to persist in these communities.