A lot of the "AI Alignment" and "rationalism" talk really feels like it echos a lot of alt-right "race realist" talking points, but towards software rather than people from a different region or walk of life. "Ah, my mistake, clearly the outcome was obvious if we let them exist among us."
Granted there's obviously very different impacts between the implied out-groups being software plus those who enable LLMs, versus general populaces or marginalized demographics.
Still, it concerns me especially as it's not too far a stretch for those socially invested in Effective Outrage to turn the other heel when it suits them, it'd be only rational to be.
in the sense that both are about uncomfortable truths, things we don't want to be true but are true anyway.
most people care more about their social standing than about truth, so they just lie (including to themselves), no biggie.
the "AI alignment" people are for example far more open to granting personhood to sufficiently advanced AI systems than others are. it's not hate of AI that drives the alignment discourse, it's a deeper understanding of our shared reality.
for simpler minds, understanding the risks posed by a superhuman AI is akin to hate, and understanding the degrees of inheritability of various traits is akin to discrimination. simpler minds need the lies in order to not hate, need the lies in order to not discriminate.
miladyincontrol · · focus · HN ↗
Granted there's obviously very different impacts between the implied out-groups being software plus those who enable LLMs, versus general populaces or marginalized demographics.
Still, it concerns me especially as it's not too far a stretch for those socially invested in Effective Outrage to turn the other heel when it suits them, it'd be only rational to be.
subsistence234 · · focus · HN ↗
most people care more about their social standing than about truth, so they just lie (including to themselves), no biggie.
the "AI alignment" people are for example far more open to granting personhood to sufficiently advanced AI systems than others are. it's not hate of AI that drives the alignment discourse, it's a deeper understanding of our shared reality.
for simpler minds, understanding the risks posed by a superhuman AI is akin to hate, and understanding the degrees of inheritability of various traits is akin to discrimination. simpler minds need the lies in order to not hate, need the lies in order to not discriminate.
cindyllm · · focus · HN ↗
[dead]