Ethics and morality play a huge role in what people believe is correct and applicable when it comes to ML and AI in general. Morality aside, anything is possible, but keeping in mind the objective of creating the AI in the first place, the line between helping and regressing the human species thins and blurs. AI is based on the ML algorithm for RLHF.
RLHF stands for Reinforcement Learning from Human Feedback. Most of these AI chatbots work with this format to create a more human-like speech by using labeled data and then comparing that with unexpected inputs and feedback from the user. To be trained to perfection, the chatbots need to be given positive and negative feedback.
But within the sampled datasets are actual people's works and lifeblood. The works of artists and authors are collected as part of the data sample for these AIs to be trained on. Because of this, AI has the ability to create just like these authors, but more efficiently. They aren’t compensated for the property AI uses or for the subsequent works created based on their works.. It is also debated who should be credited for the work of the AI.
Do the people who generate the art get to call themselves the artist, or are the people who created the code the creators of this work? Or should it be credited to the persons on whom the work was based? If this were the case, how could we determine every person the work is based on? The people are feeding AI, and it scrapes the web at a rate faster than humanly possible; there is no way we could trace and catalog all the different works it could be based on. Speaking of image scraping, what images doesn’t the AI use? In short, nothing.
