Rest of World · Technologie

I refused to train the AI that could replace me

An individual with a Ph.D. in agricultural AI in South Africa was offered a job to train an AI system to perform tasks like designing assessments, teaching, and grading essays, skills acquired over a decade of study and experience.

The author, after completing a Ph.D. on AI's impact on agricultural work in South Africa, sought an academic career but was instead offered a lucrative position to train an AI in teaching and assessment.

The decision to interview for the role was influenced by South Africa's socioeconomic conditions, including high youth unemployment and low minimum wage, presenting a conflict between livelihood and ethical concerns about AI replacing human jobs.

This trend of highly skilled individuals training AI systems that may replace them is global, with writers paid to humanize AI text and workers in India paid to record their activities for robot training.

The author notes that transferring human knowledge and judgment to machines is a common practice in the AI economy, potentially exposing Africa, which has low AI adoption rates but a young, educated population and high unemployment, to exploitation by AI companies seeking lower-cost expertise.

Unlike basic data labeling, the AI training offered involved transferring nuanced judgment, such as understanding core concepts, evaluating essays, and distinguishing memorization from genuine understanding, skills typically developed over years of experience.

Professions relying on discretion and judgment, like law and medicine, are now targets for AI recruitment, with AI platforms actively seeking professionals to transfer their decision-making capabilities.

The author's own recruitment process was conducted by an AI, which also provided feedback and suggested further assessment, a departure from traditional human-led hiring.

Ultimately, the author declined the AI training job, though the exact reasons remain unclear, prompting reflection on the implications of entrusting human judgment and professional identity to machines and the ethical considerations of the AI systems being developed.

AI-samenvatting op basis van de bron.