Jay Patel

I wonder what personality psychologists think of this analysis on.ft.com/45G0aEF by @jburnmurdoch.ft.com (which updates this paper from a couple years ago doi.org/10.1371/jour...)

If you had to develop a brief benchmark to assess GPT-5's level of educational science knowledge, what would it include? Any quick gut checks?

Do you have any internal data on accuracy? Would be a quick and easy UX/data analysis project.

Key sections of the US Constitution deleted from government's website | TechCrunch Internet sleuths say the U.S. Constitution's website is now missing key sections from its website, including a key legal provision relating to habeas corpus, which protects citizens from unlawful dete...

The constitution. They sliced out the parts of the constitution that they don’t like from the government website.

Yet another reminder that in an aspiring authoritarian government, you should never rely on online document as definitive sources.

I use:

🟨 : definitions and questions
🟥 : evidence
🟩 : methods
🟪 : claims
🟧 : all else

grey: uncertainty/gaps/limitations

I'd log this example to help teach my undergrad stats students about uncertainty.

Great initiative by Wiley! The publisher has started marking retracted papers in reference lists. When you click on the retraction notice, you can also see the date and reason for the retraction.
All publishers should adopt this practice. But also screen references during submission.

There's also a severe disconnect between the math needed for a given science or CS course and the math taught in the calculus and precalc courses.

I'd happily learn more math if it actually furthered my learning of the sciences.K-10th math should be required with deep understanding of stats.

Read "How People Learn II: Learners, Contexts, and Cultures" at NAP.edu Read chapter Front Matter: There are many reasons to be curious about the way people learn, and the past several decades have seen an explosion of researc...

Why are there so few lab schools?

Idea for edscience grads: Use the latest thought in educational and cognitive science to deliver a uniquely effective experiences for underpriviledged youth and fund it through philanthrophic funding sources.

Create a modern research-practice partnership

Which edtech solutions actually leverage these ideas meaningfully?

After a few tweaks, the open gpt 20b model gave me a correct diagnosis to a complicated health issue I needed years, thousands of dollars, and deep research skills to solve.

The older 3.5 and 4 models used to do the same, but now the pipeline is open for improvement.

This is meaningful progress.

This is why editing with long delays between sessions is a must for great writing. Much initial writing is gold. The rest is repetitive and gives away from vocabulary tics.

Interesting...

If you paid an APC, not even a thrift store...

Can you unpack your critique?

I found some incorrect labels, but I think clicking the thumbs-up and thumbs-down buttons will improve the feature over time.

Users should always click to see whether the labels are correct.

Reason number 9283 to use a reference manager: I discovered I missed a letter and was misnaming an author for the past year. Citation errors can sneak up on you.

Use something like @zotero.org or @paperpile.com

Since I read "Genentech: The beginnings of biotech" I've been craving more non-fiction about slices of scientific history.

Anyone have book recommendations?

The more detailed and specific the topic, the better! Thanks!

Reminded me of TRIPOD-AI which you also authored. Any additional value to these specialized guidelines like CHART? Reading over it, I agree with the items. The full checklist seems even more through.

I worry that few will adopt these guidelines, though.

New journal just dropped. Late night out, don't expect me to return your calls.

Using the same content/learning objectives or more advanced?

What’s the most underrated skill in science?
We’ll go first: writing a good limitations section.

Would love to see metrics like GDP per capita and other economic indicators overlayed on this chart.

@aj-boston.bsky.social is the Lemony Snicket of the scholarly publishing world.

cc @paperstarsorg.bsky.social , @alphaxiv.org The dark matter/energy problem of potential reviews is the grand challenge to fix. But how? Integration with reference managers and notebook software? Voice input via browser extensions?

I've been using @paperstarsorg.bsky.social recently in addition to PubPeer and @alphaxiv.org to leave comments on papers.

I'm finding both to be very smooth and modern. Hopefully, they catch on and scale the practice #openscience.

Would be great to create a community of such tools and collab.

Given the health and economic benefits, the Clinical Trials feature should be provided to Elicit Free users. It would also be excellent for getting the word out about Elicit in the biomedical community.

"This is the first study..." in a paper makes me fume. 😤
On Google Scholar, it returns almost 2 million hits.

"This is the second study..." returns 3,840 hits.
That's a difference of ~520X.

I'm more likely to believe the latter claim.
📖 If you make a novelty claim, then back it up.

📖 I've seen plenty of prototypes and polished programs of LLMs searching for relevant lit, extracting data, and appraising its quality.

But no evaluations of its ability to synthesize the literature into a clear and accurate whole. 🧩

#metascience #AIinScience #LLM #scientificsoftware

📣 We are pleased to announce that recruitment for 2025-26 Crowd Preprint Reviewers & Leads is open!

📝 The goal of the Crowd Preprint Review initiative is to provide an engaging environment for researchers to participate in giving feedback on preprints and support public reviews of preprints.

Meaning? Failure to comply?