Responsible natural language processing: a principlist framework for social benefits

Abstract

Businesses harness the power of natural language processing (NLP) to automate processes and make data-driven decisions. However, NLP raises concerns on a number of fronts due to its potential for disruption, which can be addressed with the assignment of responsibility. Therefore, responsible NLP (RNLP) can be designed as a principlist framework to ensure NLP systems are used in an ethical manner. The study proposes a principlist framework with the formulation of eight principlist ethical principles to ensure NLP is safe, secure and reliable for responsible decision making and subsequently results in social benefits. Using snowball sampling, data are collected from 15 informants, who represent senior-level positions in diversified industries. The analysis is performed with qualitative research methodology. The result produces two ethical practices. First is the adoption of RNLP as a disruptive technology for ethical decision making for social benefits and second is the creation of a culture of responsibility.