Post #1655928

2026-04-20 06:57 UTC

@tomstafford Agreed, I think a good deal of opposition to AI use comes from the realization that it cheapens previously costly signals. I think we are (as a society) now discovering that a lot of processes relied on long-form text as a costly signal that a human spent some cognitive effort on this. These could have been for example rate-limiting mechanism (as for the scientific proposals you mentioned), or as indicator for learning (all of education).

Replies (2)

@moritz_negwer@mstdn.science 2026-04-20 07:05

@tomstafford maybe our bullshit detectors will adapt in time, but I fear a lot of these processes will either become more selective (as we are seeing in research funding) or more dysfunctional (ask any programmer who has to maintain a be-slopped codebase, or anyone trying to review academic papers right now). I hope that we get to reinvent some of those processes over time - maybe there is a corner of education where there is freedom to experiment (and less of a profit motive to hold them back)?

Open ##1655929
@tomstafford@mastodon.online 2026-04-20 08:05

@moritz_negwer yes, great point From a science funding and costly signalling perspective, one question is how funders could impose application costs which are disproportionately costly for 'dishonest' signallers (and what that would even mean?)

Open ##1655930