Mel Andrews

A thread of my papers. The first was recently published with Cell Press, coauthored with Andy Smart and Abeba Birhane. It tackles the connections between epistemic deficiencies in deployed ML and ethical consequences. www.sciencedirect.com/science/arti...

"In Referees We Trust?" is @mbaldwin.bsky.social's investigation into the origins of the peer review system, a system that is often considered the essential mechanism that protects the trustworthiness of scientific research. Available #OpenAccess: mitpress.mit.edu/978026205474...

www.nytimes.com/2026/10/01/o... So excited to welcome @leifw.bsky.social to @princetoncdh.bsky.social next week! #leifweek

Nihar Shah did a heroic experiment for TMLR: he spent 20-25 hours over two weeks interviewing authors of seemingly low-quality submissions about their own papers. He confirmed what we all suspected: people submitting these papers have *no idea* what is going on in them.

There are no responsible uses of AI in evaluating research funding applications. Just calling it ‘responsible’ doesn’t make it so. #EthicsWashing #GenAI

We’re hiring for two positions! One is TT in Ancient, the other an open-rank TT position in mind, science, or ethics. Please share widely! See more on PhilJobs: TT Ancient: philjobs.org/job/show/32273 Open-rank Mind, Science, or Ethics: philjobs.org/job/show/32265

Hill I'm ready to die on: whether for grants or articles, review by AI is not *peer review*. Review based on a 1st scan by AI without actual reading of the grant or manuscript is not peer review either.

so we just nearly started a war with china (a nuclear state) because of genAI use

First submit of the term, let me know if you want to talk data models (or read a draft!).

“Who wants to see the new Love Island?” Quiet babe my favorite show is on:

One of the few LLM experts worth paying attention to. www.nytimes.com/2026/09/12/o...

Do I have any friends out there in forensic accounting?

The use of major open problems in mathematics as AI benchmarks mistakes the nature and aims of mathematics: “solving problems is only a tool and proxy for achieving the primary goal of conceptual understanding and insight.”

You can now read this invaluable book on methods in philosophy of science open-access. Find me in Chapter 16!

Methods in the Philosophy of Science: A User's Guide, edited by the inestimable @phieveigl.bsky.social (and me!), has 27 chapters from super-cool philosophers of science explaining how they do their thing, and is now open access! Our baby is free! #philsci direct.mit.edu/books/oa-edi...

Very excited that *Methods in the Philosophy of Science: A User's Guide* is now available open access! Lucas and I have a chapter in it on "Doing Public Engagement with Science as a Philosopher of Science" #philsci #scicomm. In our chapter we ...

Expanded these comments into a blog post. These warnings from the A.I. industry are an opportunity, but the opportunity is wasted if we focus only on the solutions they have in mind. The A.I. Industry will never rein itself in davekarpf.beehiiv.com/p/the-a-i-in...

Three notes: (1) this is not *just* marketing. These guys really do mean it. They are convinced they’ve built a vengeful digital god. (2) but also they’re a cult. They are not impartial scientific experts. (3) and one of the byproduct of their endless confidence is marketing hype.

The oldest natural history museum in the nation, the Academy of Natural Sciences in Philadelphia, will cease to exist by the end of the month. The museum’s operating costs are half of 1% of Drexel’s budget. Drexel’s president pockets 2.8 million annually. www.nytimes.com/2026/09/01/u...

This summer I led a series of conversations in my lab about "AI" and our research. Some colleagues have shared "AI policies" for their labs and we discussed whether this would make sense for us. We decided it did not, and came up with a different approach. Please feel free to share & adapt!

Who knew that all along the secret of how to teach the kids lay in merely teaching the kids

The open secret to pedagogy research is that small class sizes/maximizing educator attention per pupil is the simplest, most effective, and unfortunately the least attractive to administration thing you can do to improve learning outcomes.

We know how to teach books in the age of AI: small classes, close reading, research-engaged learning (which is how you ’let students see you struggle’). The issue is nobody will hire and pay the requisite # of faculty to offer this.

In the 1930s, a group of European philosophers fled fascism and reshaped American philosophy. Why did logical positivism have such an impact on US thought? And why did it play a central role in the culture wars of the 1950s and 1960s? Released today! Logical Positivism: An American History (1/3)

can someone rename gas to $2/ gallon

A lesson in how to take education seriously.

what does @tonychemero.bsky.social's new book 'Intertwined Creatures' have to offer to the “AI consciousness” debate? here is my take on it www.nature.com/articles/d41...

At times I feel that the issue is more that the reactionary right often adequately diagnoses the failures of neoliberalism and then proffers a more extremified neoliberalism as antidote.

Spoiler: the reason scientists were hesitant to speak out about issues of scientific soundness and the reason scientific soundness was historically low were not unrelated.

Warum Linkssein und die Nutzung generativer KI unvereinbar ist, habe ich für’s @ndaktuell aufgeschrieben. Danke an @raulzelik.bsky.social für die Anregung zum Text! www.nd-aktuell.de/artikel/1201...