Angelo Varlotta
banner
ang42.bsky.social
Angelo Varlotta
@ang42.bsky.social
Background: biology, data science. I'm interested in applied ml, responsible ai and sustainability. Volunteer @climatematch & @neuromatch. "È tutto sbagliato, è tutto da rifare". (Bartali)
Reposted by Angelo Varlotta
Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.
June 19, 2025 at 11:21 AM
Reposted by Angelo Varlotta
December 4, 2025 at 5:30 PM
Reposted by Angelo Varlotta
La nostra stella oggi. Da Napoli. Di Enrico Cascone.
December 3, 2025 at 10:25 PM
Reposted by Angelo Varlotta
Le megacostellazioni di satelliti presto potrebbero avere un impatto considerevole anche sulle osservazioni condotte con i telescopi spaziali
www.nature.com/articles/d41...
Satellite swarms set to photobomb more than 95% of some telescopes’ images
Planned megaconstellations would contaminate the view of the cosmos of four orbiting telescopes
www.nature.com
December 3, 2025 at 6:45 PM
December 2, 2025 at 9:24 PM
Piccolo spazio pubblicità.
"Ciao sono l'AI generativa, dicono che sono qui per salvare il mondo ma io intanto -- nell'attesa di infilarti ads ovunque per sopravvivere (economicamente) -- ti spammerei un po' di robaccia da comprare per Natale. Che ne dici?"
December 2, 2025 at 6:26 PM
Reposted by Angelo Varlotta
🎨Andrea Mantegna, Il Parnaso, 1497. Tempera su tela, 160x192 cm. Parigi, Museo del Louvre #arte
November 29, 2025 at 11:07 AM
Reposted by Angelo Varlotta
Avanti? #Altan
November 29, 2025 at 8:18 AM
With a 56k modem, it will take a while to upload all the instructions.
November 28, 2025 at 11:07 PM
Reposted by Angelo Varlotta
The length of ads on stuff like YouTube and Amazon Prime is ridiculous. They’re just reinventing cable.
November 28, 2025 at 10:53 PM
Reposted by Angelo Varlotta
Lanciati dalla California otto satelliti della costellazione italiana IRIDE per l'osservazione della Terra, finanziata con fondi PNRR, e il satellite HydroGNSS dell' @esa.int
November 28, 2025 at 6:50 PM
It's a loop.
November 28, 2025 at 10:51 PM
Reposted by Angelo Varlotta
23:23 😴
November 28, 2025 at 10:22 PM
Reposted by Angelo Varlotta
What's the polite and effective way to tell analytic philosophers to stop being so modest about what they do and start making bigger claims about the place of philosophy in the university?

If philosophy gets relegated, it's partly because it doesn't attract a lot of majors, but also because...
November 28, 2025 at 8:49 PM
If there are em dashes in your text, it may be an indicator that it was written by AI.
Now that this concept is everywhere in the Italian news, it brings me back to the “delve” controversy.

#ai #chatgpt #ethics
November 19, 2025 at 8:08 AM
Reposted by Angelo Varlotta
Our paper on a neuromorphic algorithm for solving finite element models is out, which is exciting for a number of reasons. But really I want to talk a bit about how this relates to understanding the brain and #NeuroAI. 🤖🧪🧠🧵 1/

www.nature.com/articles/s42...
Solving sparse finite element problems on neuromorphic hardware - Nature Machine Intelligence
Theilman and Aimone introduce a natively spiking algorithm for solving partial differential equations on large-scale neuromorphic computers and demonstrate the algorithm on Intel’s Loihi 2 neuromorphi...
www.nature.com
November 16, 2025 at 4:16 PM
Reposted by Angelo Varlotta
Reservoir computing is good for #neuroAI because
1. it avoids the biological implausibility of backpropagation
2. it naturally models temporal dynamics
3. Random recurrent reservoirs naturally sit near the edge of chaos (complexity is good)
I can continue...

#AI #neuromorphic
November 16, 2025 at 10:17 PM
xkcd: Car Size
xkcd.com
November 12, 2025 at 9:37 PM
🎧: In a world of unpredictable leaders, are nuclear weapons making us more vulnerable?
@prospectmagazine.co.uk on Bluesky
🎧: In a world of unpredictable leaders, are nuclear weapons making us more vulnerable? Serhii Plokhy joins @ellenhalliday.bsky.social and @aloner.bsky.social on the Prospect Podcast to discuss. https://www.prospectmagazine.co.uk/podcasts/prospect-podcast/71563/serhii-plokhy-how-close-are-we-to-nuclear-war
bsky.app
November 12, 2025 at 1:05 PM
About AI in education.
AI has probably only highlighted the limits of the current scholar system, based on tests, grades, and knowledge.
If we think that teaching and studying are aimed at passing a test or getting a high grade, then AI has probably already won.
1/3
#generativeAI #ethics #school
November 10, 2025 at 4:23 PM
🧪
This is a great input I received. To respond that:
1/ Social influence doesn’t stop at politics or media—it’s present in science too.
Scientists are not immune to bandwagon effects—repeated claims in literature can start to feel more “true” (illusory truth effect), even when replication is weak.
September 17, 2025 at 7:41 AM
September 11, 2025 at 12:11 PM
Silence, please
In 1994, Mr. Rodgers made a prescient point: our information-overloaded world leaves too little time for quiet reflection.
September 11, 2025 at 10:25 AM
Si possono salvare i post.
Abbiamo svoltato.
September 10, 2025 at 6:24 PM