The Dark Side of AI Training

Generative AI relies heavily on three resources: skilled engineers, powerful computational equipment, and large amounts of data for training. However, many AI companies often use unlicensed creative work to train their models, creating an unfair competitive landscape for original creators. For example, research indicated that a significant portion of large language models are trained on copyrighted materials without acquiring the necessary licenses. This practice not only undermines the copyright laws meant to protect creators but also poses a threat to the livelihoods of those artists, as generative AI directly competes with their original work.