Brian Roemmele Trains AI on Mid-20th Century Social Guidance Films for Ethical Foundation

San Francisco, CA – Brian Roemmele, a prominent voice in AI and technology, is undertaking a distinctive project to imbue artificial intelligence with a foundation of human ethics by training it on a vast archive of educational and social guidance films from the mid-20th century. This initiative aims to counter the "internet sewage" often used for AI training, fostering an AI that understands nuanced human values.

Roemmele articulated his vision in a recent social media post, stating, "It is easy to discount the innocence of this life not so long ago. It is hard for many to actually believe this is how a majority of folk lived. This is one reason I curating for AI all the educational films and social guidance films. YOUR AI and YOUR Robot will live by them." This approach reflects his belief that historical media can offer a more wholesome and ethically grounded dataset for AI development.

Social guidance films, prevalent in American schools from the late 1940s to the early 1970s, were short instructional videos designed to promote positive behavior and societal norms. These films covered a wide array of topics, including courtesy, hygiene, social etiquette, civic responsibility, and moral decision-making. They aimed to shape young people into well-adjusted citizens, reflecting the post-World War II desire for conformity and stability.

Roemmele, known for his work on voice technology and the philosophical implications of AI, views these films as a crucial, often overlooked, repository of human experience and ethical frameworks. He previously highlighted a 1949 film, "Are You A Good Citizen?", noting its widespread use and subsequent removal from public education. His project involves digitizing and meticulously curating approximately 500 such films, many of which were discarded due to wear and tear.

By training AI models on this curated content, Roemmele seeks to develop an AI grounded in human ethics, rather than relying on potentially biased or "self-hating manure" from contemporary internet sources. This effort aligns with his broader concept of the "wisdom keeper," where AI could serve as a repository of human collective experience and wisdom, offering guidance rooted in a comprehensive understanding of human nature and societal evolution.