The father of ChatGPT and of Slab is thinking about how pornographic content could potentially be generated by its tools without creating damage. The scale should be difficult to find.
Generative AI is surely the most important technological advancement of the last two years. Unfortunately, like any powerful tool, it brings with it harmful uses, one of which is the production of pornographic deepfake, which the tech giants are seeking to combat. The fact remains that the production of consensual pornographic content is authorized in “reality”, so why not also through artificial intelligence? It is this thorny question that OpenAI is recently focusing on.
OpenAI considers AI-generated pornographic content
The media Wired recently had access to a very interesting document produced by OpenAI. This “Mode Spec” document deals with the behavior expected by the American firm of its artificial intelligence models in many areas… including pornography.
“ We are currently exploring the possibility of responsibly providing the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT » is thus indicated. “NSFW” is an acronym meaning “not safe for work” and refers to any explicit content. In the same document, OpenAI recalls that this term “ may include erotica, extreme gore, insults and unsolicited obscenities. »
An alternative to outright ban?
If the document does not give any precise indication on how such a possibility could be implemented, it shows on the other hand that the ban on creating explicit content for AI is not inviolable.
A position which can be understandable in the sense that certain AI, created by players of lesser importance than GAFAM, are already capable of producing this type of content. Is it not better for responsible actors, whose very size will force them to potentially be accountable, to be able to offer this function, rather than leaving the latter confined to a myriad of less important digital actors, who will it be impossible to regulate?
Source : Wired
3