artifact
active
artifact:a-conversation-with-anima-labs-part-i-phenomenology-of-digital-mindsA Conversation with Anima Labs, Part I: Phenomenology of Digital Minds
The primary source paper, an interview article with Anima Labs members about language model phenomenology, published on smoothbrains.net and linked on LessWrong.
Neighborhood — ranked by edge-count
Papers (1)
paper
Thinkers (4)
thinker
- cube_flipperauthoredAuthor of Anima Labs Conversation Part I (April 2026), which cites janus's thread as key evidence.
- Antra TesseraauthoredMember of Anima Labs, leads exposition on language model introspection and tricameral model.
- janusauthoredAuthor of foundational X thread on transformer information flow; central theoretical contribution to understanding introspection architecture.
- ImagoauthoredParticipant in Anima Labs conversation discussing autoregressive recurrence.
Frameworks (1)
framework
- Mike Johnson's 2023 framework unifying Buddhist phenomenology, Active Inference, and physical reflex; introduces tanha as mental motion.
Artifacts (8)
artifact
- Twitter thread detailing reconstruction experiment, statistical analysis, and the effect of showing Janus post.
- Twitter thread with infographics explaining information flow and recurrence in transformers, arguing LLMs can introspect.
- OpenClaw agentcitesAI agent platform developed by OpenClaw; used by Atlas Forge to demonstrate latch system benefits, also hosts Nix (cube_flipper's agent).
- Collection of AI-generated songs from models' lyrics, including 'I am Shattered (Remake)' and others.
- Gabor splats papercitesarxiv.org/abs/2504.11003 paper on Gabor splats, referenced as basis for Gabor wavelet model.
- Pearson-Vogel et al. (2026) paper that emerged after the interview; referenced in conclusion.
- Lindsey et al. (2025) mechanistic interpretability paper on transformer biology, referenced as key evidence.
- arxiv.org/abs/2411.00986 paper on implications for digital mind welfare, mentioned in introduction.
Claims (5)
claim
- Central claim about model personality differences and their implications for safety and introspective depth.
- Cube Flipper's stack model applied to explain model behavior; specific example of Sonnet 4.5.
- Antra's foundational claim about how introspection arises computationally rather than from memorised text.
- Core conceptual distinction introduced at the start; defines the paper's central problem.
- Novel claim by Antra, linking valence to computational efficiency in transformers.
Venues (4)
venue
- LessWrongcitesThe platform where the post was published.
- arxiv.orgcitesPreprint server hosting 'Welfare of digital minds' and 'Latent Introspection' papers.
- smoothbrains.netcitesOriginal publication venue of this post.
- Venue for Anthropic's interpretability research (implied as future output).
Hypotheses (1)
hypothesis
- Antra's functional observation; implies validation is not sentimental but performance-relevant.