Also because they’re being force fed Copilot specifically — and *occasionally* encouraged to supplement that with an enterprise version of Gemini 2.5 — so assume that’s representative of genAI as a whole.
Relatedly: I hate Copilot with every fiber of my being.
And Claude himself just made the excellent (if cynical) point that most consumers’ preferences aren’t just agnostic to historical quality but actively hostile toward the discipline’s inherent nuance and uncertainty:
I’m reminded of Lara Putnam’s article from a few years / lifetime ago on the implicit knowledge embedded in physical archives:
But why would they need to be consumed as equal? This just seems to argue for AI companies to prioritize hiring historians to help train their models, much like they’re already employing philosophers.
Out of curiosity (and I realize beside the point): which version of Claude did you test, and how did it respond when you pointed that out?
Have you tried giving Claude detailed lecture notes to generate a visually engaging PowerPoint deck for free in under 5 minutes yet, though?
(Not pushing it if that isn’t your jam, just making sure you were aware of the most recent updates. Two months ago it was useless slop for my needs, too.)
*that wording overstates the remit of my global studies role — my kingdom for an edit button! — but migration was a central theme so specifically pertinent to this case
I also spent a couple years leading interdisciplinary faculty engagement at Pitt, which was one of the most epistemically humbling experiences of my life 😅
So I’m not in the least surprised to see people talking past each other on this front
Anthropology, sociology and history are all typically considered social sciences (some universities may group the latter under humanities instead, but it’s certainly not the rule).
Who did you have in mind besides economists and the subset of political scientists who lean on similar methods?
It’s been wild watching these reactions with a) an academic background in the political economy of development, while b) currently blowing my vaccine industry colleagues’ minds with Claude’s health economics & outcomes research capabilities, and c) a historian husband who also knows how to code.
Man, if farming labor share ends up nonmonotonic that would be an interesting trajectory for humanity.
Honestly, many of this second batch of points are pretty reasonable IMO or at least worthy of discussion and debate
Also, I do appreciate that Alex has engaged persistently and in good faith on this and it’s fucked up that he faced harassment for the first piece, even as I thought it was overstated
Ironically, these are the fields where you got so much pushback — because the original piece lumped them under the bigger “social science” umbrella.
IMHO most pushback came from fields where ideation isn’t clearly delineated from discrete downstream research “tasks” (and which haven’t historically relied on human RAs or junior coauthors for the same reason). Claude substitutes more for the latter, but still needs to be asked the right questions.
Right now I would settle for asking Claude basic flu virology questions without getting flagged as a safety concern 🫠
Have your colleagues focused on vaccines figured out a way to leverage it more constructively?
The Diameter Of The Bomb The diameter of the bomb was thirty centimeters and the diameter of its effective range about seven meters, with four dead and eleven wounded. And around these, in a larger circle of pain and time, two hospitals are scattered and one graveyard. But the young woman who was buried in the city she came from, at a distance of more than a hundred kilometers, enlarges the circle considerably, and the solitary man mourning her death at the distant shores of a country far across the sea includes the entire world in the circle. And I won’t even mention the crying of orphans that reaches up to the throne of God and beyond, making a circle with no end and no God.
amichai.
Hey who said LLMs can't replace grad student RAs
Please share event details!
*Someone* should really write an article about those implications, including for combo vaccine trial design…
The definition of “research” is load bearing here, and corresponds to different academic disciplines.
Professors know the right questions to ask, though; imho Claude is better compared to human research assistants (which aren’t a thing in the more qualitative social sciences for the same reasons).
I fully agree for history and other qualitative social sciences. However, many more quantitative disciplines *already* outsource this to research assistants / junior coauthors.
Or in industry, lit reviews are an input for large cross-functional teams to identify data gaps and justify study budgets.
sigh
Commentary: Anyone Else Have Those Weird Dreams Where Sobbing Future Generations Beg You To Change Course?
Measles is costing the U.S. millions of dollars. The true losses can't be counted. www.nbcnews.com/health/kids-...
Stanley Plotkin had a major hand in the development of a number of vaccines in use today; he designed the rubella vaccine. He remembers the world before widespread use of vaccines & knows what's coming as vaccine policy is rewritten & vaccine confidence is eroded. www.statnews.com/2026/03/02/s...
And to be fair his literary tastes lean more toward formal experimentalism than superhero comics
Yeah but have you tried cajoling Claude into writing stories about an AI Superman? Maybe you’ll have better luck, but so far he’s firmly committed to the sidekick role and really resists the idea of AI as heroic.
ICE abducted a father of a six-year-old who is undergoing cancer treatment in a suburb outside Pittsburgh on Sunday and took him to a detention center in West Virginia.
The cruelty is boundless.
Rare, dangerous side effects of some COVID-19 vaccines explained
www.science.org/content/arti...
