The output still needed careful reading and verifying but there is a definite massive time saving to be had.

giving instance feedback, but writing an extensive set of tests which cover all cases costs a lot of time.

I made a plan with the genai system to based on the assessment brief and lots of guidance of what sort of tests I wanted.

A small personal experience - I don't know whether this would generalise:

I recently experimented with using a genai system to help implement an autograder for the programming part of an assignment. Autograders are brilliant for checking correctness of student submissions and

I think this means we still need better solutions to address the challenges you list to provide a good conference experience for all

My experience is that remote participation is really a second place option. It can allow people to present a talk remotely, but not a poster (common in some areas) or allow any the informal networking

or the organisers could see that using titles would put some speakers as 'second class' and avoid using titles. i really hate any list where there's a big list of prof, dr and some mr, mrs, miss thrown in - i can't see how labelling someone as 'not a Prof or Dr' helps anyone!

Our (slightly smaller) boys had a go during the Olympics in Paris in the summer. They really enjoyed it, but I'm not sure the 2 year old got much insight sadly :)

The normal downside is about delaying getting on with the work. If your funding runs out after a certain date, you might end up with a shorter postdoc contract if you advertise too long

For a recent position we had a similar number of applicants and interviewed all 3 that matches the job requirements

Sounded more festive the first time... :)

I'm very lucky that this is not the case in my department. I have a very supportive head of department (and other I work with) to deal with caring responsibilities