OpenAI, the artificial intelligence research company, recently came under scrutiny when some of the specialists who worked to train ChatGPT for the company revealed the ‘traumatic experience’ they faced. Mathenge, who worked in Narobi, Kenya, claims that the tasks his team were assigned to do were ‘unimaginable’ such as reading explicit materials related to crimes such as child sexual abuse and bestiality. Mathenge says this work had a negative impact on his team and that he noticed it in their attitude toward work.
On the other hand, OpenAI’s CEO Sam Altman has endorsed reinforcement learning from human feedback and stated that it was a good way for humans to train AI on what is good and bad content.
Despite this, the workers endured a difficult ordeal and some of them, such as Mathenge’s team member, Mophat Okinya, are still dealing with medical issues, such as panic attacks, depression, anxiety, and insomnia. He also claims that it caused the end of his marriage.
OpenAI claims that they offered their workers wellness programs and 1:1 counseling, and that they could opt out of any work without any consequence. They also said that exposure was standardized and sensitive information was only handled by trained workers.
However, the workers feel that the support they received was inadequate. With the success of ChatGPT, the workers still feel proud of what they have accomplished despite the hardships they endured.