Post #4385600
2026-08-04 20:18 UTC
I wrote a thing on linkedin and then read more and should not have. Looks like if you want to walk into academia, just sprinkle "ai" on shit we already knew.
Strong flashbacks to rejecting a paper by asking the authors to explain a fudge factor needed for a deep learning process to do a job better than plain old stats
[it was never published. Which was sad - because it had some great stuff and I said so! It was all framed on the deep learning process being better and it wasn't, so...]
Replies (1)
-
@adamsteer@mapstodon.space 2026-08-04 20:21
...like... imagine where we'd be now if researchers actually published this kind of "well the deep learning /'ai' approach was super resource intensive yet not any better and often worse at [thing]" And coming to the conclusion "hey maybe people looking at this stuff are right when they say we can't do a one size fits all and need to understand the processes" ...instead of insisting that the next universal statistics spewer will be better