Let OpenAI screen article text for non-inclusive language and list the reasons in a field before anything gets published.
Marcus Johansson
October 12, 2023 · 2 min read
About this video
Implementing the AI Interpolator OpenAI module for assessing inclusive language is a straightforward task. The user is required to input their text into a designated HTML field. This feature, one of the many I coded, is specifically designed to evaluate the content for elements of inclusivity. It quickly sifts through the language used, using algorithms good at identifying terms or expressions that may not be inclusive.
The outcome is a comprehensive analysis highlighting instances where the language could be improved for broader inclusivity. Specific phrases or words are pinpointed, and reasons are provided to explain why they might exclude people. This is all part of the design: a tool that is not just about identifying issues but also about educating users on more inclusive language practices. The user can expect actionable insights and suggestions for changes, making the text more accommodating to a diverse audience.
This video is from before the AI Interpolator module became AI Automators, part of the AI module. Module and menu names have changed since, and some steps look different today.
Watch the video
This video is hosted on YouTube, which sets cookies once it loads.
0:00
Transcript
YouTube's automatic captions, lightly corrected. Click a line to jump to that moment in the video.
Chapters
·Updated ·2 min read
Other ways to consume this
The post in three sentences
Two test articles, one inclusive and one full of words like mankind and boomers, are checked by an OpenAI automator on an unlimited long text field. The prompt asks GPT-4 to list reasons why the text is not inclusive towards genders, the LGBTQ community, disabled people, age or minorities, with OpenAI moderation turned on. The inclusive article gets no reasons and the other gets a list.
