3 ms·
So... are we done politely coughing and looking out the window at the idea that the gatekeeping was motivated by altruism so that we can move on and just use th
by benreesman 4y ago
So... are we done politely coughing and looking out the window at the idea that the gatekeeping was motivated by altruism so that we can move on and just use this much better innovation model going forward?
Various (subjectively judged) SOTAs on at least some subset of at least this family of tasks is changing somewhere between daily and hourly right now. I've been watching this stuff closely since fairly early ImageNet days and I've never seen a Cambrian explosion of "how the hell did that do that?" events at anything like this cadence.
- civilized 4y agoIf you're talking about scientific gatekeeping like pre-publication peer review, the gatekeeping was never really the root problem. If researchers have great products or findings to show off, it's easier than ever to simply publish them somewhere online and let impressed audiences spread the word. Gatekeepers have been irrelevant to truly great science for a long time. It's mediocre science that needs gatekeepers to distinguish it from not-even-mediocre, truly substandard research.
- theptip 4y agoThe gatekeeping GP is referring to is OpenAPI keeping DALL-E 2 behind an invite-only API, with weights unpublished, vs. Stable Diffusion publishing the whole model for anyone to download and use.
- benreesman 4y agoIn fairness OpenAI aren’t the only ones to use fairly thin logic to avoid saying “this is about money”. Everybody was doing that until roughly, now. They’re just the only ones calling themselves “OpenAI”.