Today is International Workers’ Day, so I would like to discuss recent news in the AI world.

It’s a well-established fact that human moderation is a crucial element in developing respectful generative AI, especially when the model is trained with internet information. However, what often remains unknown are the exact conditions of the workers involved in this process.

In January, Time published their exclusive investigation, showing that OpenAI outsourced this work to Sama, an ’ethical AI’ company with Kenyan workers tasked with reviewing and labeling content for less than $2 per hour.

This information is controversial, and it becomes even more disturbing when you read the article and discover that OpenAI paid Sama $12.5 per hour for the work. At the same time, their Kenyan workers reported suffering psychological trauma from being exposed to disturbing content, and the sessions with ‘wellness’ counselors were ineffective.

Human intervention is crucial to developing AI, but AI must be ethical. Creating a system that includes precarious working conditions loses its moral aspect.

This is not exclusive to OpenAI or Sama. Employing Indian workers to do tasks like reviewing information is something we could also see in the Just Walk Out feature from Amazon:

In the end, the AI supply chain is real, and AI is the future. But that doesn’t mean everything is okay when creating AI. International Workers’ Day is commemorated in every part of the world, and we must also ensure that we work in the right conditions in every part of the world.

I want true ethical AIs in my future.

Original post on LinkedIn here

Image from @ana_valdi on Twitter here