Fabio Celli
@facells.bsky.social
data scientist, humanist, pioneer in personality computing. producing data for computational history
the Zig-zag hypothesis tested on data. #academicsky #autocracy #democracy
fabiocelli.substack.com/p/la-storia-...
fabiocelli.substack.com/p/la-storia-...
November 17, 2025 at 2:26 PM
the Zig-zag hypothesis tested on data. #academicsky #autocracy #democracy
fabiocelli.substack.com/p/la-storia-...
fabiocelli.substack.com/p/la-storia-...
web search #data show that queries about ideologies had a regular pattern util 2019. Now, AI search causes web traffic to human-written sites to collapse. This is how LLMs will have disruptive impact on #collectivememory
November 4, 2025 at 3:09 PM
web search #data show that queries about ideologies had a regular pattern util 2019. Now, AI search causes web traffic to human-written sites to collapse. This is how LLMs will have disruptive impact on #collectivememory
At #ecai: LLMs reflect their culture (e.g., Chinese LLMs see more autocracy than US ones). #academicsky: while unbiased models are possible, adoption is the issue—people prefer "like-minded" AI. The question: Should the EU mandate models like Mistral to protect our culture?
October 25, 2025 at 3:57 PM
At #ecai: LLMs reflect their culture (e.g., Chinese LLMs see more autocracy than US ones). #academicsky: while unbiased models are possible, adoption is the issue—people prefer "like-minded" AI. The question: Should the EU mandate models like Mistral to protect our culture?
#LLMs larger than 400 B parameters agree better than humans do in interpreting #history. Llama3.1-large is also very consistent when run repeatedly. PAPER (eng) -> github.com/facells/fabi... #academicsky #digitalhumanities Blog (ita) open.substack.com/pub/fabiocel...
La storiografia ai tempi dell'IA
Eseguendo uno stesso prompt, i Large Language Models sono in grado di annotare la storia in modo replicabile e valutabile in modo quantitativo. Questo cambia tutto.
open.substack.com
October 20, 2025 at 1:14 PM
#LLMs larger than 400 B parameters agree better than humans do in interpreting #history. Llama3.1-large is also very consistent when run repeatedly. PAPER (eng) -> github.com/facells/fabi... #academicsky #digitalhumanities Blog (ita) open.substack.com/pub/fabiocel...
October 19, 2025 at 1:02 PM
we invented the wearable poster - with @mrovera.bsky.social at CLIC-it @ailc-nlp.bsky.social
September 24, 2025 at 3:24 PM
we invented the wearable poster - with @mrovera.bsky.social at CLIC-it @ailc-nlp.bsky.social
tools like AI Overview are transforming web search, increasing reliance on generated answers. This is impacting historical memory, as models like GPT4 can generate distortions. We created a benchmark to study how #LLMs interpret history. #academicsky #clic-it2025 #NLP
September 15, 2025 at 7:27 AM
tools like AI Overview are transforming web search, increasing reliance on generated answers. This is impacting historical memory, as models like GPT4 can generate distortions. We created a benchmark to study how #LLMs interpret history. #academicsky #clic-it2025 #NLP
August 12, 2025 at 8:09 AM
#History interpretation is subjective, tainted by #biases, leading to frequent disagreements among humans. We showed that the largest #LLMs can reach an agreement level similar to #humans and can be used as annotators, opening new avenues for studying history. #academicsky #research
July 30, 2025 at 8:17 AM
Reposted by Fabio Celli
🚔 #Crime has a rhythm — can we decode its beat? In a new study co-authored by CSH's @rafaelprietocuriel.bsky.social, 12 years of detailed crime data reveal relatively stable temporal “heartbeats” for different crime types over time. journals.sagepub.com/doi/10.1177/...
July 22, 2025 at 2:04 PM
🚔 #Crime has a rhythm — can we decode its beat? In a new study co-authored by CSH's @rafaelprietocuriel.bsky.social, 12 years of detailed crime data reveal relatively stable temporal “heartbeats” for different crime types over time. journals.sagepub.com/doi/10.1177/...
data suggest that polytheisms supported the social #complexity boost #academicsky open.substack.com/pub/fabiocel...
La storia con i dati: la Grande Narrazione dei politeismi
Con i politeismi le società diventano più grandi e con più istituzioni (ad esempio la legge). Questa complessità si diffonde tra le società in modo esponenziale.
open.substack.com
June 16, 2025 at 1:40 PM
data suggest that polytheisms supported the social #complexity boost #academicsky open.substack.com/pub/fabiocel...
Reposted by Fabio Celli
✍️ Out now!
"Distant Writing" by Luciano Floridi explores literary production in the era of generative AI.
Available now on SSRN:
#AI #Creativity #Writing
"Distant Writing" by Luciano Floridi explores literary production in the era of generative AI.
Available now on SSRN:
#AI #Creativity #Writing
Distant Writing: Literary Production in the Age of Artificial Intelligence
<div>
Revised version 5
</div>
This article introduces the concept of "distant writing", a novel literary practice in which authors act as designers, employing Artificial Intelligen
ssrn.com
May 23, 2025 at 8:08 PM
✍️ Out now!
"Distant Writing" by Luciano Floridi explores literary production in the era of generative AI.
Available now on SSRN:
#AI #Creativity #Writing
"Distant Writing" by Luciano Floridi explores literary production in the era of generative AI.
Available now on SSRN:
#AI #Creativity #Writing
May 5, 2025 at 1:02 PM
I worked a lot with Process Mining this year, so today I release Learnipy v0.10, enriched with some algorithms for extracting petri net, heuristic net and BPMN from data. Enjoy. #datascience #academicsky #opensource
github.com/facells/lear...
github.com/facells/lear...
GitHub - facells/learnipy: making data science easier
making data science easier. Contribute to facells/learnipy development by creating an account on GitHub.
github.com
April 30, 2025 at 1:44 PM
I worked a lot with Process Mining this year, so today I release Learnipy v0.10, enriched with some algorithms for extracting petri net, heuristic net and BPMN from data. Enjoy. #datascience #academicsky #opensource
github.com/facells/lear...
github.com/facells/lear...
April 7, 2025 at 8:17 PM
April 3, 2025 at 6:43 PM
what the world will be in 100 years. #data #demography
April 1, 2025 at 9:24 AM
what the world will be in 100 years. #data #demography
March 24, 2025 at 2:13 PM
#LLMs are excellent for historical analysis, but their single-source nature leaves them vulnerable to #manipulation or censorship. That's why comparing historical data annotated by LLMs and humans is essential for fostering critical awareness of bias. 👉 www.mdpi.com/2079-9292/13...
March 18, 2025 at 5:03 PM
#LLMs are excellent for historical analysis, but their single-source nature leaves them vulnerable to #manipulation or censorship. That's why comparing historical data annotated by LLMs and humans is essential for fostering critical awareness of bias. 👉 www.mdpi.com/2079-9292/13...
March 10, 2025 at 1:40 PM
A couple of weeks ago I presented this paper on the extraction of historical structured knowledge from Large Language Models at the conference on digital libraries. The paper is now available! ceur-ws.org/Vol-3937/pap...
It was a very nice opportunity, Thaks to you all! #llm #academicSky
It was a very nice opportunity, Thaks to you all! #llm #academicSky
March 9, 2025 at 6:47 PM
A couple of weeks ago I presented this paper on the extraction of historical structured knowledge from Large Language Models at the conference on digital libraries. The paper is now available! ceur-ws.org/Vol-3937/pap...
It was a very nice opportunity, Thaks to you all! #llm #academicSky
It was a very nice opportunity, Thaks to you all! #llm #academicSky