Brief items - news roundup, AI, international issues, good reading

Several items worth reading about as we head into a long weekend in the US.  Starting with news related to funding and other aspects of US government policy:

• US government taking equity stakes in some quantum information sciences companies while investing around $2B (seemingly from the Department of Commerce and the CHIPs Act resources.  (Non-paywall news story here).  This raises a number of thorny issues. 
• Some US funding agencies (NIH, NASA) are enacting restrictions (Science article here, Inside Higher Ed article here) on publishing scientific papers with non-US coauthors.  It's understandable that US funding agencies are concerned about the possibility US funds directly or effectively supporting researchers in foreign countries.  This is not that, though.  Some people making policy seem to be moving toward wanting to ban any co-authorship, but even the agencies seem confused about what they want.
• In a move that will stress out many non-US-citizens in the country, the administration is floating making people leave the US to apply for green cards (PBS article here).  This just was sort of announced yesterday, so I don't know anything about this other than on its face it sounds to me like a terrible idea for multiple reasons.  
• The AAAS is pushing for a Senate hearing on the nominee for NSF director, on the theory that this issue and the nominee at least need to be discussed in a public forum rather than coasting along without a NSB and no end in sight to interim leadership.
• It would seem that some Republican congresspeople are pushing the idea of de-funding the National Academies.  This is directly related to the issues mentioned here.  I think the National Academies should be endowed and thus not so reliant on federal funding; this would be a way to make sure that they always feel secure in delivering reports even if the customer is a part of the government and the conclusions might be something the customer doesn't want to hear.

There was a lot of AI-related news this week:

• 
  There were three papers published in Nature about using AI agents to do science (here, here, and here, with a news and views).  The first two papers are both about drug discovery research, and the third is about using AI to help write scientific software models (also medically related).  It'll be interesting to see how this progresses.  
• One of OpenAI's tools solved an Erdos problem (that's the OpenAI release) by finding a counterexample to a conjecture long thought to be true.  Here is the accompanying paper, which includes commentary by several esteemed mathematicians.  The commentary parts of the paper are very much for non-mathematicians and fascinating to read.  It seems like the AI tools are genuinely good at pulling together complex arguments, and that so far a key advantage they have is an exhaustive familiarity with the full breadth of the literature.
• Unsurprisingly, university graduates are not fans of AI.  This cartoon from this week's New Yorker is topical.  
  

 Additional suggestions that look cool but I haven't had time to actually read:

• Jeremy Levy at Pitt has posted his draft graduate quantum mechanics textbook to the arXiv - looks like an interesting take on the pedagogy of the subject for sure. 
• Dimitrii Makarov at UT Austin has posted a list of critical papers in chemical physics.  
• Samuel Braunstein at York has put up this preprint regarding the Clay Millennium Prize problem about the Navier-Stokes equations.  His main points seem at a quick read to be (1) at finite temperature, fluctuation-dissipation requires some additional noise term be present in the N-S equations; and (2) the N-S equations assume a continuum fluid, so that infinitely-short wavelength excitations are permitted, while for all real fluids wavelengths get cut off at the molecule size.  Basically the issues w/ the N-S equations come from unphysical aspects of the formulation.