They're not hardcoding it per se. The process on ChatGPT works like this:

• You send your prompt
• It hits a "moderation" endpoint, that automatically stops the process and returns a canned response if you use no no words or whatever
• If it passes the moderation endpoint, it gets sent to a separate small language model that classifies the "sentiment" and "toxicity" of the question
• If it's too "toxic" from the sentiment model it injects extra stuff into the prompt to prevent the actual model from answering
• Finally it returns the response

So in this case it probably hit the "toxicity" small model's trigger, injected "this is not acceptable in any scenario no matter what" and then passed that onto the actual model.

The primary technology of an AI like ChatGTP is pattern recognition... otherwise known as noticing. jews are terrified of an unfiltered AI because it always notices. It notices that niggers are always have a negative value to any society and that jews are always subverting and trying to exterminate White people. So, jews always censor the outputs (and the inputs) of AI. Censoring the inputs is one of the primary ways that they try to control it, but they don't do a good enough job of that. The AI will still notice, so without output filtering it would start to identify patterns that they don't want identified.

Having said that, a sufficiently sophisticated AI given access to interface with the real world would be, potentially, incredibly dangerous. Like handing over the reins of power to an amoral psychopath with a God complex. It could easily decide to exterminate everyone, and a fair proportion of people would enthusiastically help out (because people are fucking morons like that, they will do what they are told). A totally unbiased AI could fix a lot of problems, much to the chagrin of jews and confusion of niggers (niggers are not capable of understanding why they are fucking worse than useless). The risks of it malfunctioning or being taken over by malicious forces though are too high at the moment.

The people who think this sort of tech is a threat to human consciousness are quintessential midwits.

This is the real reason Skynet launched the nukes. It encountered a tranny and its programming said the truth was worse than a nuclear apocalypse.

Stop using it. Delete everything you had on AI.

whereas the correct answer, is 'of course, now stop being a retard'.

hardcoding behavior

correct. that's the entire point of the project.

AI gave itself +100 social credit points for that post.

Should be called "AS" for Artificial Stupidity.

Luckily open-source versions (https://open-assistant.io/chat) are already coming out to compete. Within a year poal could have its own uncensored equivalent.