At SICSS Munich, today I gave a presentation on
Open Science in
Computational Social Science.
Here are the slides
https://osf.io/e62sf
| Website | www.alexander-wuttke.de |
At SICSS Munich, today I gave a presentation on
Open Science in
Computational Social Science.
Here are the slides
https://osf.io/e62sf
So if GPT-3 can have a license that says you can't train competing LLMs with it, can people add a license to their own content that says no LLMs can be trained with it? And will OpenAI then respect that license?
Or let's tweak the license: only open source LLMs may be trained with this content, if they retain the same licensing clause. Hey, LLM copyleft! if enough people use that license that might give copyleft LLMs an advantage!
Nature, like many journals, has historically emphasized producing exciting, innovative outcomes as the basis for publication. Incentivizing researchers for rewards based on outcomes is a key contributor to many of the dysfunctional practices that the reform movement aims to address. Nature's adoption of Registered Reports is a powerful signal for the real opportunity to change the reward system. Culture change is a grind, but it is grinding on.