Artificial intelligence (AI) is a reflection of a part of humanity. When you work with AI, you’re not talking to a machine, but to a representation of a selection of human culture. And this is true whether you’re generating text, images or music. The selections for what constitutes ‘human culture’ has been decided long before the arrival of neural networks, a process which has gradually diluted everything that is niche, unique or distant from the dominant culture. In this way, the societal problem in reflecting a diverse representation of human culture and creative thought is funneled through the bias of dominant cultural selections which are then amplified in these new artistic technological tools. If these tools rely on our ability to generate representations of human culture datasets, then together with these communities, we want to build an AI with an artistic vocation that reflects different angles of society we live and make creative work within. Chimère is the starting point for broadening our collective actions in diversifying the representation of human culture. The absence of these cultures can only be resolved by creating new, more inclusive data sets with the people concerned.