Post #1802284

2026-04-28 10:52 UTC

@frankreiff @ikuturso @chaucerburnt @sinbad it seems to me that there actually are two possible explanations if you consider chaining of more than one action: - one-way pipeline (consciousness as pure rationalization) - feedback loop (whatever consciousness produces could be used to inform future subconscious actions) maybe I'm not aware of some other experiment that disproves option 2 though

Replies (2)

@archo@mastodon.gamedev.place 2026-04-28 11:00

@frankreiff @ikuturso @chaucerburnt @sinbad on explaining past actions however, I see humans as having some distinct advantages: 1. operating with an extremely broad range of conceptual units (unlike LLM wordpiece tokens) for highly efficient encoding 2. we can reprioritize notability of observations based on past events such that if we end up regretting missing some fact, we can figure out that we need to pay more attention to something in the future, which doesn't seem possible with LLMs

Open ##1802285
@frankreiff@mastodon.social 2026-04-28 11:10

@archo @ikuturso @chaucerburnt @sinbad I’m not a researcher into consciousness, but I find it a fascinating subject. I did cognitive psychology in the 1990s.. and there’s been a lot of advances since then. Connectivist ideas in AI have always interested me, and large LLMs behave in very interesting ways that mirror neuroscience findings.. unsurprisingly since they imitate its low level structure. The big question is how low level structure gives rise to high level intelligence.

Open ##1802288