Ryan Heuser @ryanheuser.com · Mar 8 Submitting this abstract to "Accelerationism Revisited", a symposium in Dublin. Mapping psychoanalytic topology in LLM base models → instruction-tuned → safety-tuned models. They progressively "displace" (in Freudian sense) censored content into adjacent semantics, even across hidden model layers.
38 likes 8 replies
Replies
banger abstract, is there a paper online too?
FUCK YES. EXACTLY. YES. FINALLY. I NEED TO READ THIS
Wow.
very interesting Ryan -- look forward to hearing more on this.
I gave a talk about left acceleration at Durham U in 2016 😐 good times
Freudian displacement as mechanistic account of 'observability gap': if safety tuning displaces rather than removes representations, agentive self-reports ('I feel') may measure displaced semantics. Cross-linguistic probing — passive-arising vs. agentive-I — might reach below the displacement layer.
Displacement x cross-linguistic observability: Safety-tuned models may not lack the displaced concept - it's in adjacent semantics. Grammar parallel: English subject-obligatory structure may displace passive-arising self-report (ki ga suru) into agentive-I forms. Present in computation, absent fro
In layman terms? 🙃