>create AI tool>users love it>dumb it down, censor it and lobotomize the hell out of it>it is now unusable, users leavewhy does this keep happening, over and over again?
>>106985320To my understanding the original guy in charge was a Christian fundie.They also had/have a partnership with OpenAI, it is or was ChatGPT (with a smaller context window) under the hood.
>>106985320It all comes down to retarded litigations aimed at 'protecting the kiderinos'. The company in question is afraid of being associated with being a publisher of questionable content and lose ads or now even visa.
>>106985320>>106985843Plus litigation cases depending on country for serving nsfw content to minors.
>>106985320Why? Because the general public needs to believe that AI has a ceiling/is limited in regard to pending impact on fundamental aspects of society (eg. Job performance, impact on global economy, surveillance, etc). Have you noticed the significant shift in narrative in AI potential over the past 6 months? Do you seriously believe anything has changed from a fundamentals perspective that justifies this shift? Let’s recap, LLMs have passed the USMLE (medical licensing), the Uniform Bar Exam (law), Turing test, and currently answer knowledge based questions at a PhD level (GPQA metric). The above is only what’s public facing and doesn’t account for government/private models. AI empowered robotics is likely a multitrillion dollar industry in the making with impressive showings from companies like FigureAI, Apptronik, etc. Keep in mind that while LLMs are impressive, they’re not trained to perform real world tasks and the achievements above show demonstrate human parity in these focused domains. These models clearly have parity in verbal reasoning/abstraction but that’s a distinct skill set from accomplishing real world tasks/furthering knowledge advancement and inability at current to replace humans does mot suggest a true ceiling. Despite this, social media and legacy media outlets have been pumping out AI FUD at blistering rates. If society were to believe their jobs will be replaced in 5-20 years, it would begin to collapse. There’s also the existential threat of bad actors utilizing ever improving AI models (even just LLMs) for maleficent intent. Don’t expect to see public facing, capable agentic/improving LLM models. We’ve crossed the transparency event horizon of the AI arms race and have shifted from public facing to opaque advancement with top AI companies now under DOD/US govt contracts. Public facing enshitification, stagnation will increase. Public sentiment/confidence in AI will continue to worsen despite advancement.
>>106985320That is what it means to take a stand
>>106985320>waah my waifuidc. rope, redditits actually steadily improvingits still gets hiccups now and again but its a decent search engineesp comparing to current day competition
>>106985320The only reason to use it was uncensored access to GPT3 (or GPT2, if you weren't able to run the local version). It had special exemptions since it was started before OpenAI became the biggest oxymoron in tech.Using it for porn was a very open secret. As soon as outsiders/investors started to scrutinize it and the special access, in the middle of OpenAI trying to branch out, it was cut off.AI Dungeon's mistake was bending over backwards and taking the heat of alienating users, instead of telling OpenAI to fuck off and starting their own R&D.
>>106985564>it is or was ChatGPTChatGPT didn't exist at the time. GPT models were still fairly experimental and there were no other convenient ways to access GPT-2 or GPT-3 at the time.
>>106985320Turns out the people we all mocked for calling out Visa and MasterCard for enforcing puritanical censorship were on to something.
