John Sohrawardi

This will have a really significant effect on the academia in the US, given the already cutthroat competition for grants.

Under Trump, AI Scientists Are Told to Remove ‘Ideological Bias’ From Powerful Models A directive from the National Institute of Standards and Technology eliminates mention of “AI safety” and “AI fairness.”

You have to be kidding me ... 🫠

The new agreement removes mention of developing tools “for authenticating content and tracking its provenance” as well as “labeling synthetic content,” signaling less interest in tracking misinformation and deep fakes.
www.wired.com/story/ai-saf...

Even intuitively this makes sense, especially given that we can't version closed models.

However, a useful paper to cite with empirical information for people using LMs for text analysis.

Given this, there are still research avenues for closed source LMs where limited reproducibility is tolerable.