Lord Forgive Me, It is Time to Go Again to the Previous ChatGPT

Earlier this yr, OpenAI scaled back some of ChatGPT’s “personality” as a part of a broader effort to enhance person security following the loss of life of a teen who took his personal life after discussing it with the chatbot. However apparently, that’s all up to now. Sam Altman announced on Twitter that the corporate goes again to the outdated ChatGPT, now with porn mode.

“We made ChatGPT fairly restrictive to verify we have been being cautious with psychological well being points,” Altman mentioned, referring to the corporate’s age-gating that pushed customers right into a extra age-appropriate expertise. Across the identical time, customers began complaining about ChatGPT getting “lobotomized,” offering worse outputs and fewer persona.  “We understand this made it much less helpful/pleasing to many customers who had no psychological well being issues, however given the seriousness of the problem we wished to get this proper.” That change adopted the submitting of a wrongful death lawsuit from the mother and father of a 16-year-old who requested ChatGPT, amongst different issues, for recommendation on the best way to tie a noose earlier than taking his personal life.

However don’t fear, that’s all mounted now! Regardless of admitting earlier this yr that safeguards can “degrade” over the course of longer conversations, Altman confidently claimed, “We now have been in a position to mitigate the intense psychological well being points.” Due to that, the corporate believes it might “safely loosen up the restrictions normally.” Within the coming weeks, based on Altman, ChatGPT might be allowed to have extra of a persona, like the corporate’s earlier 4o mannequin. When the corporate upgraded its mannequin to GPT-5 earlier this yr, customers started grieving the loss of their AI companion and lamenting the chatbot’s more sterile responses. You recognize, simply common wholesome behaviors.

“If you would like your ChatGPT to reply in a really human-like method, or use a ton of emoji, or act like a buddy, ChatGPT ought to do it (however solely if you would like it, not as a result of we’re usage-maxxing),” Altman mentioned, apparently ignoring the corporate’s personal earlier reporting that warned individuals might develop an “emotional reliance” when interacting with its 4o mannequin. MIT researchers have warned that customers who “understand or want an AI to have caring motives will use language that elicits exactly this conduct. This creates an echo chamber of affection that threatens to be extraordinarily addictive.” Now that’s apparently a characteristic and never a bug. Very cool.

Taking it a step additional, Altman mentioned the corporate would additional embrace its “deal with grownup customers like adults” precept by introducing “erotica for verified adults.” Earlier this yr, Altman mocked Elon Musk’s xAI for releasing an AI girlfriend mode. Seems he’s come round on the waifu method.