To clarify: I just do not know how it would help to strike, as all my tasks would still need to get done if I take a day off. I would just be punishing myself by having an even higher workload the rest of the week, so what is the point? Honestly curious about other people's thoughts.

Moral of the story: don't ever let ChatGPT write your papers or proposals, because it will actually make up complete references, and will (initially) insist that they are real.

When I tell ChatGPT again that it is wrong, it finally admits that it cannot find the right paper and that I am probably looking for one of the actual van der Marel 2021 papers (which it lists correctly).

Like seriously: it is not just garbling of authors. THERE IS NO PAPER WITH THIS TITLE OR TOPIC. The DOI points to a non-existing wegpage.

It got worse. When I told ChatGPT that it was mistaken, it told me that it got confused by the journal name, which should be ApJ, not AJ. And then it produced the 'correct' paper (see below) WHICH DOES NOT EXIST AT ALL!!

Nee, geen geluid, ik had het even snel in elkaar gezet en ik weet uit ervaring dat ik ga hakkelen bij zo'n opname, en dan moet het telkens weer opnieuw. Misschien doe ik dat later nog.

At 0.25" (~35 au) resolution, these small disks would remain 'unresolved': that means that they appear to be as large as the beam size and you cannot know whether they are indeed the beam size or much smaller. Their size would be listed as upper limit of <20 au. Now it's shown to be much smaller!

Before ALMA, there was no possibility to infer the disk size from such faint disks at all! For the brightest and largest ones we could measure them (somewhat) with SMA, but no more. However, these disks were mapped before with earlier ALMA observations at low-resolution, ~0.25" (Ansdell+2016/2018).

And I was working with ALMA Early Science data during grad school and I'm still amazed with what ALMA can do these days 😄

Please note, mas stands for milliarcseconds, not microarcseconds! It's still impressive of course.