Researchers from the University of Zurich discovered that ChatGPT can outperform workers performing text labeling tasks that will be used to train an AI system. They found that ChatGPT can tag text with greater accuracy and consistency.
The researchers fed ChatGPT a sample of 2,382 tweets and asked it to classify the text by relevance, topic, position, problem or solution formulation, and policy formulation. Using ChatGPT, the researchers obtained higher accuracy and agreement between coders. In addition, using AI was twenty times cheaper than paying for human labor.
This research adds to the debate about how the rapid development of AI language models, such as OpenAI’s GPT, will affect jobs.
Technology companies employ tens of thousands of workers to manually label and filter content from datasets of AI models. This is due to the fact that artificial intelligence is often not yet able to recognize the nuances of an image, especially when it is learning. In January, the Times reported that OpenAI paid Kenyan workers less than $2 per hour to make your chatbot safer to use. Workers were regularly exposed to traumatic, indecent content, including graphic text about child sexual abuse, zoophilia, murder, suicide and torture.
Replacing AI-trained workers with artificial intelligence not only does not solve the problems of their terrible working conditions, it also takes away their jobs.
Researchers note that it is too early to say how much ChatGPT can replace human labor:
“Our paper demonstrates ChatGPT’s potential for data annotation tasks, but more research is needed to fully understand ChatGPT’s capacities in this area. For example, our tests used tweets in English and were performed on a relatively limited number of tasks. It will be important to extend the analysis to more kinds of tasks, types of data, and languages,” paper co-author Fabrizio Gilardi said.