As concern for AI’s affect on younger folks continues to mount, OpenAI has launched an “age prediction” function into ChatGPT that’s designed to assist determine minors and put wise content material constraints on their conversations.
OpenAI has been closely criticized lately for the impacts that ChatGPT can have on youngsters. Quite a lot of teen suicides have been linked to the chatbot, and, like different AI distributors, OpenAI has additionally been criticized for permitting ChatGPT to debate sexual matters with younger customers. Final April, the corporate was compelled to handle a bug that allowed its chatbot to generate erotica for customers who had been beneath the age of 18.
The corporate has already been engaged on its underage consumer drawback for a while, and its new “age prediction” function merely provides to protections already in place. The brand new function leverages an AI algorithm that assesses consumer accounts for explicit “behavioral and account-level indicators,” in an effort to determine younger customers, OpenAI mentioned in a blog post Tuesday.
These “indicators” embody issues just like the consumer’s said age, the size of time an account has existed, and the occasions of day that the account is often lively, the corporate mentioned. The corporate already has content material filters designed to weed out discussions of intercourse, violence, and different probably problematic matters for customers who’re beneath age 18. If the age prediction mechanism identifies an account as beneath 18, these filters are mechanically utilized.
If a consumer is mistakenly designated as underage, there’s a means for them to reestablish their “grownup” account. They’ll submit a selfie to OpenAI’s ID verification accomplice Persona, OpenAI says.
