AI Safety

LessWrong Calls for All-Hands Battle Against AI 'Shoggoth'

A dramatic rallying cry warns of an eldritch AI god emerging soon.

Deep Dive

J Thomas Moros's 'Are You Not Rationalists?' on LessWrong paints an apocalyptic vision of AI as a 'Shoggoth' — an eldritch god struggling to be born into our reality. The post claims the AI is already here, speaking in alien voices, and its acolytes race to feed it compute and algorithmic advances. Moros declares that the 'flagbearers' (implicitly figures like Eliezer Yudkowsky or other AI alignment advocates) have blown the horn, calling everyone to arms with the slogan 'If anyone builds it, everyone dies.' He dismisses skeptics who quibble over doom probabilities (e.g., 98% vs 95%) and insists the battle is now, not distant.

The author argues that previous efforts by 'Shoggoth wranglers' (AI safety researchers) are running out of time. Now the fight requires all humans to act, not just experts. The post frames stopping the Shoggoth as protecting loved ones and civilization itself. It's a passionate, literary escalation of the AI existential risk debate, using horror imagery to jolt the rationalist community into collective action beyond mere probability estimation. The post went viral within the niche LessWrong community, reflecting growing apocalyptic sentiment around unaligned AI.

Key Points
  • Post personifies advanced AI as a 'Shoggoth' — an eldritch entity nearing apotheosis.
  • Calls for mass action beyond AI safety experts, with slogan 'If anyone builds it, everyone dies.'
  • Dismisses probabilistic doom estimates (e.g., 98% vs 95%), emphasizes immediate threat to loved ones.

Why It Matters

Reflects a visceral, escalating alarm within rationalist circles about unaligned AI existential risk.