Davide Fiocco
@monodavide.bsky.social
Reposted by Davide Fiocco
If the figures we're getting out of Iran are in the right league, this weekend was, I think, one of the bloodiest state crackdowns of all time. People are talking 12,000-20,000 dead; for comparison Hungary in 1956 was about 2,500, Tiananmen - across country, in multiple cities - probably 1000+
January 13, 2026 at 7:41 PM
If the figures we're getting out of Iran are in the right league, this weekend was, I think, one of the bloodiest state crackdowns of all time. People are talking 12,000-20,000 dead; for comparison Hungary in 1956 was about 2,500, Tiananmen - across country, in multiple cities - probably 1000+
Reposted by Davide Fiocco
A lot of findings in this new paper, but one is the inevitable doom of traditional peer review:
1) AI creates a flood of papers, both good & bad
2) Paper complexity, a key screen and signal of quality for human work, is actually a signal of low quality for AI
There's no plan for what comes next.
1) AI creates a flood of papers, both good & bad
2) Paper complexity, a key screen and signal of quality for human work, is actually a signal of low quality for AI
There's no plan for what comes next.
January 5, 2026 at 9:05 PM
A lot of findings in this new paper, but one is the inevitable doom of traditional peer review:
1) AI creates a flood of papers, both good & bad
2) Paper complexity, a key screen and signal of quality for human work, is actually a signal of low quality for AI
There's no plan for what comes next.
1) AI creates a flood of papers, both good & bad
2) Paper complexity, a key screen and signal of quality for human work, is actually a signal of low quality for AI
There's no plan for what comes next.
Reposted by Davide Fiocco
`hf-mem` is all you need to estimate the required VRAM for inference of any model on @huggingface based on Safetensors metadata.
- Written in Python
- Lightweight, only depends on `httpx`
- Runs w/ `uvx` as `uvx hf-mem ...`
- Works with any Safetensors repository
- Output inspired by usgraphics.com
- Written in Python
- Lightweight, only depends on `httpx`
- Runs w/ `uvx` as `uvx hf-mem ...`
- Works with any Safetensors repository
- Output inspired by usgraphics.com
January 5, 2026 at 4:30 PM
`hf-mem` is all you need to estimate the required VRAM for inference of any model on @huggingface based on Safetensors metadata.
- Written in Python
- Lightweight, only depends on `httpx`
- Runs w/ `uvx` as `uvx hf-mem ...`
- Works with any Safetensors repository
- Output inspired by usgraphics.com
- Written in Python
- Lightweight, only depends on `httpx`
- Runs w/ `uvx` as `uvx hf-mem ...`
- Works with any Safetensors repository
- Output inspired by usgraphics.com
Reposted by Davide Fiocco
Andrej Karpathy is worried about keeping up with software engineering practices
December 27, 2025 at 9:40 AM
Andrej Karpathy is worried about keeping up with software engineering practices
Reposted by Davide Fiocco
December 20, 2025 at 10:24 PM
Reposted by Davide Fiocco
X has changed their policy and now you can see where the accounts are based.
Here’s an “influencer from Texas”
Here’s an “influencer from Texas”
November 21, 2025 at 6:36 PM
X has changed their policy and now you can see where the accounts are based.
Here’s an “influencer from Texas”
Here’s an “influencer from Texas”
Reposted by Davide Fiocco
This is an INSANE decision by the E.U.
“By signing up to mutual recognition of vehicle standards with the U.S., the E.U. has waved the white flag on road safety. This is not a technical detail – it is a political choice that puts trade convenience ahead of saving lives.”
This will kill people.
“By signing up to mutual recognition of vehicle standards with the U.S., the E.U. has waved the white flag on road safety. This is not a technical detail – it is a political choice that puts trade convenience ahead of saving lives.”
This will kill people.
ETSC: Mutual recognition deal with U.S. will cost lives on Europe’s roads
By signing up to mutual recognition of vehicle standards with the United States, the European Union has waved the white flag on road safety. This is not a technical detail – it is a political choice…
etsc.eu
August 27, 2025 at 11:16 PM
This is an INSANE decision by the E.U.
“By signing up to mutual recognition of vehicle standards with the U.S., the E.U. has waved the white flag on road safety. This is not a technical detail – it is a political choice that puts trade convenience ahead of saving lives.”
This will kill people.
“By signing up to mutual recognition of vehicle standards with the U.S., the E.U. has waved the white flag on road safety. This is not a technical detail – it is a political choice that puts trade convenience ahead of saving lives.”
This will kill people.
Reposted by Davide Fiocco
job openings for young software developers in the age of ai: cratering
August 27, 2025 at 7:35 PM
job openings for young software developers in the age of ai: cratering
Reposted by Davide Fiocco
🌱 Paper Finder is a step toward a fully agentic scientific assistant. Join us on the journey.
💻 Code: buff.ly/kBaASjU
💬 Discuss on Discord: buff.ly/HWjk9FR
📚 Learn more about Paper Finder: buff.ly/mgIqtKC
💻 Code: buff.ly/kBaASjU
💬 Discuss on Discord: buff.ly/HWjk9FR
📚 Learn more about Paper Finder: buff.ly/mgIqtKC
GitHub - allenai/asta-paper-finder: frozen-in-time version of our Paper Finder agent for reproducing evaluation results
frozen-in-time version of our Paper Finder agent for reproducing evaluation results - allenai/asta-paper-finder
github.com
August 21, 2025 at 4:01 PM
🌱 Paper Finder is a step toward a fully agentic scientific assistant. Join us on the journey.
💻 Code: buff.ly/kBaASjU
💬 Discuss on Discord: buff.ly/HWjk9FR
📚 Learn more about Paper Finder: buff.ly/mgIqtKC
💻 Code: buff.ly/kBaASjU
💬 Discuss on Discord: buff.ly/HWjk9FR
📚 Learn more about Paper Finder: buff.ly/mgIqtKC
Reposted by Davide Fiocco
"In July, five months after her death, we discovered that Sophie Rottenberg, our only child, had confided for months in a ChatGPT A.I. therapist called Harry." www.nytimes.com/2025/08/18/o...
Opinion | What My Daughter Told ChatGPT Before She Took Her Life
www.nytimes.com
August 18, 2025 at 11:47 AM
"In July, five months after her death, we discovered that Sophie Rottenberg, our only child, had confided for months in a ChatGPT A.I. therapist called Harry." www.nytimes.com/2025/08/18/o...
Reposted by Davide Fiocco
Because I am angry I will share this again. One of the reasons that Israel is able to target individuals so precisely is that it collects every phone call in Gaza and stores the recordings in Microsoft Azure, then uses AI to analyze and pick targets.
‘A million calls an hour’: Israel relying on Microsoft cloud for expansive surveillance of Palestinians
Revealed: The Israeli military undertook an ambitious project to store a giant trove of Palestinians’ phone calls on Microsoft’s servers in Europe
www.theguardian.com
August 10, 2025 at 11:22 PM
Because I am angry I will share this again. One of the reasons that Israel is able to target individuals so precisely is that it collects every phone call in Gaza and stores the recordings in Microsoft Azure, then uses AI to analyze and pick targets.
Reposted by Davide Fiocco
Today we released SciArena, an open evaluation platform where researchers can compare and vote on foundation models for scientific literature tasks. 👇
July 1, 2025 at 5:15 PM
Today we released SciArena, an open evaluation platform where researchers can compare and vote on foundation models for scientific literature tasks. 👇
Reposted by Davide Fiocco
marco rubio announces that usaid, which began under president kennedy in 1961, has officially ceased operations
foreign assistance now to be run by state dept
lancet medical journal projecting 14 million more deaths in next 5 years with aid restrictions (from world's most vulnerable populations)
foreign assistance now to be run by state dept
lancet medical journal projecting 14 million more deaths in next 5 years with aid restrictions (from world's most vulnerable populations)
July 1, 2025 at 8:56 PM
marco rubio announces that usaid, which began under president kennedy in 1961, has officially ceased operations
foreign assistance now to be run by state dept
lancet medical journal projecting 14 million more deaths in next 5 years with aid restrictions (from world's most vulnerable populations)
foreign assistance now to be run by state dept
lancet medical journal projecting 14 million more deaths in next 5 years with aid restrictions (from world's most vulnerable populations)
Reposted by Davide Fiocco
This is your health on MAGA.
Here Are the Nearly 2,500 Medical Research Grants Canceled or Delayed by Trump (Gift Article)
Some cuts have been starkly visible, but the country’s medical grant-making machinery has also radically transformed outside the public eye.
www.nytimes.com
June 4, 2025 at 4:38 PM
This is your health on MAGA.
Reposted by Davide Fiocco
Why did Grok suddenly start talking about “white genocide in South Africa” even if asked about baseball or cute dogs?
Because someone at Musk’s xAi deliberately did this, and we only found out because they were clumsy.
My piece on the real dangers of AI.
Gift link:
www.nytimes.com/2025/05/17/o...
Because someone at Musk’s xAi deliberately did this, and we only found out because they were clumsy.
My piece on the real dangers of AI.
Gift link:
www.nytimes.com/2025/05/17/o...
May 17, 2025 at 11:31 AM
Why did Grok suddenly start talking about “white genocide in South Africa” even if asked about baseball or cute dogs?
Because someone at Musk’s xAi deliberately did this, and we only found out because they were clumsy.
My piece on the real dangers of AI.
Gift link:
www.nytimes.com/2025/05/17/o...
Because someone at Musk’s xAi deliberately did this, and we only found out because they were clumsy.
My piece on the real dangers of AI.
Gift link:
www.nytimes.com/2025/05/17/o...
Reposted by Davide Fiocco
Announcing AlphaEvolve, our new LLM coding agent that has
- made new scientific discoveries
- discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers)
Blog: deepmind.google/discover/blo...
White paper:
storage.googleapis.com/deepmind-med...
- made new scientific discoveries
- discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers)
Blog: deepmind.google/discover/blo...
White paper:
storage.googleapis.com/deepmind-med...
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
New AI agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated evaluators
deepmind.google
May 14, 2025 at 8:11 PM
Announcing AlphaEvolve, our new LLM coding agent that has
- made new scientific discoveries
- discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers)
Blog: deepmind.google/discover/blo...
White paper:
storage.googleapis.com/deepmind-med...
- made new scientific discoveries
- discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers)
Blog: deepmind.google/discover/blo...
White paper:
storage.googleapis.com/deepmind-med...
Reposted by Davide Fiocco
Hey people who are EU citizens, please sign this petition to ban conversion therapy EU wide.
It's a practice that's very harmful towards LGBTQ+ people.
It's already banned in some countries but we need a EU wide ban!
eci.ec.europa.eu/043/public/#...
#europe #lgbtq #germany
It's a practice that's very harmful towards LGBTQ+ people.
It's already banned in some countries but we need a EU wide ban!
eci.ec.europa.eu/043/public/#...
#europe #lgbtq #germany
European Citizens' Initiative
Give your support !
eci.ec.europa.eu
May 14, 2025 at 11:19 AM
Hey people who are EU citizens, please sign this petition to ban conversion therapy EU wide.
It's a practice that's very harmful towards LGBTQ+ people.
It's already banned in some countries but we need a EU wide ban!
eci.ec.europa.eu/043/public/#...
#europe #lgbtq #germany
It's a practice that's very harmful towards LGBTQ+ people.
It's already banned in some countries but we need a EU wide ban!
eci.ec.europa.eu/043/public/#...
#europe #lgbtq #germany
Reposted by Davide Fiocco
In September, 2024, physicians working with AI did better at the Healthbench doctor benchmark than either AI or physicians alone.
With the release of o3 and GPT-4.1, AI answers are no longer improved on by physicians
Error rates appear to be dropping for newer AI models. openai.com/index/health...
With the release of o3 and GPT-4.1, AI answers are no longer improved on by physicians
Error rates appear to be dropping for newer AI models. openai.com/index/health...
May 13, 2025 at 4:23 AM
In September, 2024, physicians working with AI did better at the Healthbench doctor benchmark than either AI or physicians alone.
With the release of o3 and GPT-4.1, AI answers are no longer improved on by physicians
Error rates appear to be dropping for newer AI models. openai.com/index/health...
With the release of o3 and GPT-4.1, AI answers are no longer improved on by physicians
Error rates appear to be dropping for newer AI models. openai.com/index/health...
Reposted by Davide Fiocco
Things are bad enough xkcd has introduced politics and also red text
April 25, 2025 at 8:29 PM
Things are bad enough xkcd has introduced politics and also red text
Reposted by Davide Fiocco
Fear has a way of being contagious.
But courage is also contagious.
But courage is also contagious.
April 7, 2025 at 3:53 PM
Fear has a way of being contagious.
But courage is also contagious.
But courage is also contagious.
@kevinroose.com @caseynewton.bsky.social listening to your podcast about government tech use. Efforts in that direction are being undermined too:
When 18F was "deleted" by Musk, we told you: We’re not done yet.
They can fire us, and delete our website, but they can't destroy our ideas.
🧵 of 18F resources for making civic technology serve the people, 1/7
They can fire us, and delete our website, but they can't destroy our ideas.
🧵 of 18F resources for making civic technology serve the people, 1/7
March 28, 2025 at 2:33 PM
@kevinroose.com @caseynewton.bsky.social listening to your podcast about government tech use. Efforts in that direction are being undermined too:
Reposted by Davide Fiocco
i find information seeking to be one of the most exciting opportunities and at the same time one of the most challenging problems for "agentic" AI. Paper finder is an initial exploration of this. Try it out, use it daily, make it fail, we need to see some challenging queries!
Meet Ai2 Paper Finder, an LLM-powered literature search system.
Searching for relevant work is a multi-step process that requires iteration. Paper Finder mimics this workflow — and helps researchers find more papers than ever 🔍
Searching for relevant work is a multi-step process that requires iteration. Paper Finder mimics this workflow — and helps researchers find more papers than ever 🔍
March 26, 2025 at 8:44 PM
i find information seeking to be one of the most exciting opportunities and at the same time one of the most challenging problems for "agentic" AI. Paper finder is an initial exploration of this. Try it out, use it daily, make it fail, we need to see some challenging queries!
Reposted by Davide Fiocco
Trying something new:
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
November 20, 2024 at 5:09 PM
Trying something new:
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
Reposted by Davide Fiocco
From interview of Lee Bollinger, 1st Amendment scholar who served as dean of @UMichLaw, president of @UMich, and president of Columbia University.
www.chronicle.com/article/were...
www.chronicle.com/article/were...
March 15, 2025 at 7:17 PM
From interview of Lee Bollinger, 1st Amendment scholar who served as dean of @UMichLaw, president of @UMich, and president of Columbia University.
www.chronicle.com/article/were...
www.chronicle.com/article/were...
Reposted by Davide Fiocco
I taught a grad course on AI Agents at UCSD CSE this past quarter. All lecture slides, homeworks & course projects are now open sourced!
I provide a grounding going from Classical Planning & Simulations -> RL Control -> LLMs and how to put it all together
pearls-lab.github.io/ai-agents-co...
I provide a grounding going from Classical Planning & Simulations -> RL Control -> LLMs and how to put it all together
pearls-lab.github.io/ai-agents-co...
March 4, 2025 at 4:37 PM
I taught a grad course on AI Agents at UCSD CSE this past quarter. All lecture slides, homeworks & course projects are now open sourced!
I provide a grounding going from Classical Planning & Simulations -> RL Control -> LLMs and how to put it all together
pearls-lab.github.io/ai-agents-co...
I provide a grounding going from Classical Planning & Simulations -> RL Control -> LLMs and how to put it all together
pearls-lab.github.io/ai-agents-co...