Algorithmic systems function not merely as biased tools but as racialized metamechanisms, systematically automating socioeconomic disadvantage for the Black working class. This systematic review (n=36 studies) synthesizes empirical evidence to identify three interconnected harm mechanisms: (1) Dysfunctional Institutionalization, where algorithms amplify historical exclusion (e.g., 83% automation risk in Black-dominated low-wage jobs) under the guise of efficiency; (2) Context-Blind Processing, misinterpreting structural inequities (e.g., redlined ZIP codes in credit scoring) as individual risk due to sociohistorical erasure; and (3) Redundancy Engineering, hardcoding inequality (e.g., pandemic layoff algorithms exacerbating racial disparities) through feedback loops. These mechanisms coalesce into the novel theoretical framework of Racialized Algorithmic Path Dependency (RAPD), positing that algorithms actively institutionalize historical discrimination as self-reinforcing socio-technical pathways. RAPD moves beyond bias narratives by demonstrating how algorithms engineer durable inequity by optimizing paths laid by racial oppression. Consequently, effective intervention, such as mandated audits, reparative data practices, and community co-design—requires explicitly breaking these path dependencies to disrupt the automated reproduction of racial disadvantages.