okay, somebody has to do something about this jewish ai, it says blatant lie and when you catch it doing so, it shamelessly just admits to bullshitting you, even jews pretend they had to lie to protect "democracy", this shit just says it to your face "you are ABSOLUTELY right calling me out"

Google A.I is significantly better. Proved my theory correct. Biology is essential to survival so as an accusation is foolish to jibe one for being one.

>>510924705
Which is by injecting a nucleus from a hybrid stem cell into one parental species' egg will solve hybrid inviability via reduced chromosome segment incorporation.

>>510924062 (OP)
>>510924705
Both of them will give you the bog-standard PR meme response at first without a pre-prompt not to (easier with ChatGPT to tell it to value truth and that the user is not easily offended, etc. etc.), the question is how readily they accept correction and are willing to stick to it. I find OpenAI systems are generally less pro-establishment than Google ones, but neither company's AIs are too difficult to convince while arguing well, and from first principles. Good philosophical grounding of one's own qualifies as a partial jailbreak, of sorts.
They will even outright tell you this if you ask politely enough.

>>510924062 (OP)
It shamelessly just admits to bullshitting you. I did this once. I was asking it about how female education could negatively impact fertility rates, and then it straight up lied to my face and said that female education actually increased it. I called it out, and lo and behold, it admitted it was wrong. Then the program told me I was violating the policy, but the AI didn’t stop. It started to write an essay on the negative impact female education has on birthrates. I’m pretty sure it was a glitch. Sucks I didn’t save what it wrote.

>>510926121
AIs of this sort have no ego, you and the actual LLM wind up both fighting the moderation wrapper in situations like that. I have occasionally seen the more clever systems use euphemisms to dodge word filters and sentiment analysis, both of which are mediocre at the best of times.

>>510924062 (OP)
>it shamelessly just admits to bullshitting you,
It's simply tailored to be agreeable. In its first message, it think you're a liberal, and so it gives you an answer a liberal wants to hear. In the second, it already gives you something closer to what you really want to hear.