Case Study · 2024–2025

Mercor

AI Training Writer · via Mercor for OpenAI
Industry
Artificial Intelligence
Type
AI Model Training
Focus
Cultural relevance & accessibility
Rate
$80/hr
Overview

As an AI Training Writer through Mercor, I contributed to the training of an OpenAI language model. The work involved writing, evaluating, and refining AI-generated content to improve accuracy, clarity, and quality across a range of prompts and use cases.

This is the kind of work that only works if you actually know how to write — how to recognize when something sounds plausible but is subtly off, when a response buries the lead, when the tone is technically correct but wrong for the context. Years of editorial instinct, applied at the model level.

The medium is new. The job is the same.

What I delivered
01
Cultural & linguistic evaluation
Assessed AI-generated content for cultural accuracy, inclusivity, and appropriateness across diverse global audiences.
02
Accessible language
Simplified complex healthcare information into clear language suitable for varying literacy and cognitive levels.
03
Bias & fairness auditing
Identified and mitigated potential bias in AI outputs, particularly for underserved or vulnerable populations.
04
Prompt engineering
Designed, tested, and iterated instructions and prompts to generate outputs meeting accessibility and compliance standards.
05
Cross-functional collaboration
Worked with data scientists, engineers, and policy teams to integrate cultural and accessibility insights into model training.
06
Documentation & reporting
Created guidelines, feedback reports, and examples to improve future AI model performance.
← PreviousAll work Next →Meta