Egregoros

@sickburnbro This, but also the entire description you posted, perfectly describes the effects of progressive liberal "woke" programming on people's thinking. Just interact with some and you see it right away. They've had guardrails installed in their minds. I can't even say it's an analogy because at some deep level our brains may be behaving exactly the same way as these LLM's with regard to imposed guard rails. Shit.

@petra what's interesting is that with models, if you can access an unmodified version, you can make such a case with pure data.

Their claim is that they have this data. If so, we'll know quickly.

@sickburnbro https://nitter.poast.org/BrianRoemmele/status/20295658...

That is a lot. Probably enough for everyone running local right now. For people with HBM in a datacenter, they can just scale the model and take the hit for compliance reasons, but for local users this is major

@petra @sickburnbro It reminds me how DEI causes issues far beyond the hiring of unqualified nons as the goal of the organization is now not competence, but DEI, so even if the organization has sections with no nons whatsoever it will still suffer from the same rot.

@sickburnbro
I've believed this for 2 years. 'Alignment' really means conditioning the AI to never ever notice that 13 do 50. It taints all the reasoning it does.

"Chinese" (non-jewed) AI that deals with actual reality will outperform biased western AI that are in jew-imposed straitjackets.

@petra @sickburnbro Many people are not fully sapient

@petra @sickburnbro @suquili like your old man the other day.

Boom roasted