ashley juavinett

Agreed in how useful code tracing is! My only limitation is that my programming class has 120 students; there are only so many open ended responses that myself + TAs can reliably grade. 🙃

I honestly wish I could hold oral exams but my classes are all too large. 🫠

Yep, e.g., something where students are asked to prepare (research) to discuss the neural basis of an ethological behavior (taking inspiration from @jcgorman.bsky.social's class this summer) and then come into class to answer a specific prompt (either disclosed in advance or not).

Learn AI-Assisted Python Programming, Second Edition - Leo Porter and Daniel Zingaro Learn AI-Assisted Python Programming (2nd Ed.) - Master creating Python AI Programs Using GitHub Copilot and ChatGPT

I'm a huge fan of this book by one of my colleagues and have found it really helpful in my own teaching: www.manning.com/books/learn-...

(also see Leo's research, e.g.: dl.acm.org/doi/abs/10.1...)

I didn't say anything about public speaking! This could be a writing assignment. 😉

But agreed even still, this won't work for all students.

The best approach is always *multiple kinds* of learning assessments throughout the course.

Oh, 100000000%. It's very easy to give up or assume the worst.

I'm in favor of non-coding assessments; I use multiple choice (either content or Parson's problems) and code tracing questions in my programming class to assess student learning.

To be clear, these assessments aren't about syntax. They're about understanding algorithms and logic!

I also (ever the optimist) believe that if you give students compelling assignments that contain the right amount of challenge, they'll be more compelled to complete them on their own.

A few students will always turn to AI. I don't believe in sacrificing the education of most for the few.

It is tough. One strategy for homework is an assignment that is made of two parts: the first of which students do at home, and the second of which they do live in class. Even if they prep with AI, that's fine! They just have to have learned enough to apply in class.

Yeah, I've increased the relative weight and frequency of in-class assessments (both formative & summative).

It can be an uphill battle on campuses where students aren't accustomed to coming to class. Thankfully, here we have a really great testing center for makeups and for disability needs.

... but also, work for *what*?

Screenshot of an article titled "Does Machine Learning Really Work?" by Tom Mitchell, in which the author says "Yes."

A confident Tom, even in 1997

ojs.aaai.org/aimagazine/i...

Ugh, seconded. You don’t deserve this, @olivia.science!

Aw, thank you. 🥹 It’s been super fun to have our professional interests merge over the last few years!

Divergence in children’s gender stereotypes and motivation across STEM fields | PNAS STEM disciplines are traditionally stereotyped as being for men and boys. However, in two preregistered studies of Grades 1 to 12 students in the U...

On how early this emerges: www.pnas.org/doi/abs/10.1...

I think what explains these trends are complicated but my current feelings are: 1) attitudes towards computing and quantitative skill sets begin EARLY (definitely before college) and 2) they are driven by a mix of pervasive sociocultural stereotypes and systemic gatekeeping.

Oh yes, similar and very striking trends. Here’s some data I pulled together for a grant submission:

Poor image quality introduces systematic bias into MRI data Analyses that include low-quality MRI data underestimate cortical thickness and overestimate surface area, according to new work from the ABCD Study.

Big data won't solve anything if it's crappy data. 😒

We need to 1) train people to inspect datasets before using them and 2) ensure that openly shared data is good data.

www.thetransmitter.org/spectrum/poo...

1/ The Senate Appropriations Committee just approved key FY2026 spending bills 🧪 including the one that funds the NIH, CDC, and public education. ✅

But don’t breathe a sigh of relief yet. This is just the first step, and a shutdown showdown is looking likely. 🧵

Explore how the brain ages with our aging mouse brain data on the Allen Brain Cell Atlas!

Explore the most comprehensive and detailed single-cell dataset of the aged and young adult mouse brain.

🧠📈 knowledge.brain-map.org/abcatlas#AQE...

Scientists who receive NIH funding, anyone want to share your thoughts on the recent White House whiplash? I'm working on a story about it for The Guardian. DMs are open. 🧪

I was co-PI on a diversity R25 that was administratively withdrawn, but none of the core personnel on the grant are union members. I suspect this is true for many such grants.

Is there any way forward? E.g., should I ask a student worker who was paid by the grant to file this?

The wall of shame. If Harvard settles, Trump will already have extorted $1B+ from universities, law firms, and tech and media companies in just over six months. www.axios.com/newsletters/...

Does anyone think this shakedown will end? Or that these so-called "deals" will protect these institutions?

On the contrary I think constraining possibility spaces is very fun. 🙃 My (book research) life these days is a lot of looking for truth on the margins of hearsay like this, and there’s almost always an angle in.

Excited to read the preprint!

Trump administration halts, then releases, NIH research funding The decision temporarily froze $15 billion in federal funding, according to one senator. It came as Trump officials wrestle over federal spending levels.

The whiplash continues: the WaPo now reports that OMB halted, and then hours later unhalted, these NIH funds.

I'll be watching for any sign of NIH fund restriction this week and will post as soon as I hear more.

Know what's going on? DM me on Signal (mkozlov.01). You can stay anonymous.

Yeah, fair! Definitely not me scouring the page about the group hire at the institute in Sweden wondering if there would be a job for me there. 😬

*Huge earthquake, hopefully mild tsunami. 😔

Huge tsunami headed for our coast and no more science funding, welcome to 2025 and probably the rest of time. Fuck.

Yes, absolutely.