@fourbeerspod on the gap between what an experiment did and the verbal explanation given for it.
This is a pervasive problem in most sciences, and more serious than the replication crisis. Even if your facts are right, what do they actually imply?https://fireside.fm/s/Ah1OZyuo+AilLHx9D …
-
-
(That text block is from my thesis book: https://www.amazon.com/Vision-Instruction-Action-Artificial-Intelligence/dp/0262031817 … )
Prikaži ovu nit -
What's enraging about the current state of AI is that everyone is treating it as "normal science," i.e. mindlessly turning a publication-generating crank, and deliberately choosing not to ask "what does this mean?" So we don't know; so probably the answer is "not much."
Prikaži ovu nit -
In addition to the replication crisis, we have a generalization crisis (per the paper of
@talyarkoni which@fourbeerspod discussed). Probably this is as pervasive across sciences, and as misleading, as "most published findings are false." https://psyarxiv.com/jqw35 pic.twitter.com/ymX5RGgvLe
Prikaži ovu nit -
This can be fixed! It will take serious rethinking of how science is done, and how and why it works. Sciences must continually reflect on whether their current approach/research program is genuinely productive, or meaningless mechanical paper-generation. https://meaningness.com/metablog/upgrade-your-cargo-cult …pic.twitter.com/Q1AERSXSu8
Prikaži ovu nit -
Research communities should take responsibility for revising their epistemic norms. Psychology has developed a scenius—a creative subcultural ferment that is collaboratively rethinking fundamental assumptions. Yay! The
@fourbeerspod is a great window into the scene.pic.twitter.com/ZiEzdzEaif
Prikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.