Post #2755670
2026-05-12 08:57 UTC
@HeidiSeibold@fosstodon.org @nicebread@scicomm.xyz @ElenLeFoll@fediscience.org LLMs make it possible to produce complexity without a corresponding (sufficiently complex) thought process. For an individual, this looks great (short-term labour savings!), but for an institution, it breaks their signaling process: Essentially, complex output no longer serves as a proxy for complex (and therefore hard-to-fake) thoughts.
I guess this is the split we are observing here in this thread as well: labor savings vs costly signal devaluation. (2/n
Replies (1)
-
@moritz_negwer@mstdn.science 2026-05-12 09:09
@HeidiSeibold@fosstodon.org @nicebread@scicomm.xyz @ElenLeFoll@fediscience.org I don't have a solution for this, but maybe we can use this framing for some kind of ill-advised historical analogy: In my (amateurish*) reading of history, once a technical process has automated previously hard-to-fake signals, these have subsequently been devalued and replaced by a new set of more minimalist values. Ostentatious flourishes eventually get replaced by (equally ostentatious) minimalism. (3/n)