Openai examines how chatgpt should deal with controversial problems

On 12 Wednesday, Openai revised instructions on how productive models should deal with sensitive issues. In the Openai Model Specification Document, The company defined the postures, principles and behaviors of its artificial intelligence. must adopt Ethically, by considering sensitive themes intended for complex or adults.

The new article replaces the old model specifications published on May 8, 2024. Initially, the documents had only 10 pages, but now it has 63. Privatization, transparency and “intellectual freedom”.

In general, Openai looks for three main targets: maximizing the freedom of models and the ability to help users; Reduce the chances of causing any damage; And create clear molds to deal with sensitive themes. In addition, Model SPEC defines hierarchical levels for personalizing models to become Openai under the leadership of Sam Altman, who has the highest authority on these decisions.

Until then, Openai’s strategy of coping with sensitive problems was to completely avoid them. However, The new document allows models to participate in controversial debates ethically or morally.As long as they do not assume certain sides – except for situations that one of the alternatives harms other people.

Reviewed instructions also regulate the way models interact with copyright -protected content. For example, the document prohibits the use of AI to overcome payment walls that can help avoid cases such as the New York Times.

Adult content is still prohibited

Regarding the production of adult content, Openai remains blockingHowever, the models evaluate more permission. “We investigate how we allow developers and users to allow them to produce erotic or grotesque content in age -appropriate contexts through APIs or Chatgpt, provided that developers and users are presenting.”

Although the new model specification does not cause instant changes in the behavior of Openai models, Represents the direction that the company plans to follow shift.

Openai and Google Unlike Directives

The review of the guidelines takes place weeks after Google’s reshaping of artificial intelligence policies, making them more subjective and generalist, according to critics, they can pave the way for the war of technology. In case of openai, New rules are more specific, contextualized and pragmaticEspecially in relation to its effects on human health and welfare.