π all of this!
@dgbassani
Widower dad of 3. Problem-solver living an unsolvable problem. Fatuity loather. Epidemiologist learning about complexity and networks. https://www.dlsph.utoronto.ca/faculty-profile/bassani-diego-g/ and https://www.isi.it/people/diego-bassani/
π all of this!
We've reached the 'biblical responsibility' levels of madness.
Real change takes a movement that goes into the public square, and as @cdelawalla.bsky.social explains, the public square has expanded & requires moving beyond business-as-usual engagement. More on @standupforscience.bsky.social and its strategic calculus:
www.statnews.com/2026/03/06/s...
In Gaza, Israel and Biden normalized war crimes beyond anything we have seen before.
And now, Israel and Trump are using that model against Iranian civilians.
At least 13 hospitals and health facilities hit during attacks on Iran, WHO says
www.theguardian.com/global-devel...
βIf a U.S. role were to be confirmed, the strike would rank among the worst cases of civilian casualties in decades of U.S. conflicts in the βMiddle East.β www.reuters.com/world/middle...
For a [Professor] Who Used AI to Write a Paper
A DOJ lawyer argued in court today that RFK Jr. has the absolute authority to recommend (hypothetically) that people get a shot *of* measles instead of *against* measles during a hearing for the lawsuit filed against RFK Jr. by the AAP and APHA.
The good side is seeking a preliminary injunctionβ¦
βNo front-page stories in NYT, WSJ, WaPo. No mention on NBC and CBS Sunday Shows. No stand alone segments on Evening News shows. A death count on par with the OKC bombing is relegated to the back page.β www.columnblog.com/p/corporate-...
And yes⦠evergreen.
I just heard Lily Allenβs West End Girl for the first time today and oh boyβ¦ I canβt even explain. Just go listen to it.
π
Save the date!
Join us next Monday, Mar 9th 11 AM ET, for another exciting seminar with Yasaman Asgari (University of Zurich) on global science landscape.
Hoping to see you all there!β¨
"The cruel, uneventful state of apathy releases me
I value them but I won't cry every time one's wiped out
I'll try hard not to give in
Batten down to fare the wind
Rid my head of this pretense
Allow myself no mock defense
As I step into the night"
James Mercer
UVM Prof. of Computer Science, Laurent HΓ©bert Dufresne, has been awarded the 2026 Young Scientist Award for Socio- & Econophysics! A global leader in complex systems, his research connects infectious disease modeling, network science, & ecology to answer one big question: How do things spread?
Yes
A quick web search of βscif container 705 tempestβ shows lots of companies that make them as well as government tenders for bidding.
Arenβt some SCIF inside containers sometimes? The ones used in military bases. Or is that only in movies/fiction?
βItβs difficult to understand the scale of the problem, but OpenAI itself estimates that more than a million people every week show suicidal intent when chatting with ChatGPT.β
UN Security Council, at the request of France, is to hold an emergency meeting on Iran (16:00 EST/21:00 GMT). Secretary General AntΓ³nio Guterres is to address the council. www.lemonde.fr/en/internation...
βThe world is a complicated, messy, and sometimes dangerous place.β
βAnd weβre ensuring it gets much worse.β
There, I fixed it.
The Secret Agent is a masterpiece. And itβs not what you think it is at all. Academics, donβt miss it!
And this was Claude Opus 4.6
The use of these to create human work that would not exist otherwise is revolting. I have to deal with LLM generated feedback almost every week (and I stupidly step on eggshells to not accuse anyone of what they are obviously doing).
But it's also worse: in one session it tried to re-write a verification script (comparing some values with an API reference) and it inserted hard-coded values (after the API fetch) that matched my script estimates (hiding the unmatched values). "Data quality" fabrication.
Exactly. The illusion of proficiency clouds the room. If you do not know what you are doing to be able to assess its work you'll be fooled. Easily. It is an excellent bullshitter.
And no amount of guardrailing can prevent it. It is a baked-in compulsion that creeps in invariantly. Are humans like this? I once had a hire that lied about replicating an analyses and was eventually caught. But that was one person. These tools are NOT reliable in any research setting.
Every now and then I give it another try and come to the same conclusion. Coding agents are DANGEROUS. Even in well structured code and pipelines. Cutting corners and makig stuff up even when you have written a script that does it properly. And don't come at me with 'you do not know how to prompt'.