iiinova — where your digital self keeps evolving

Gerard de Melo @gdemelo.bsky.social · 02/08/2026

Our Ai4Math Workshop 2026 Berlin-Brandenburg was a phenomenal success! Mathematics is undergoing a fundamental shift and is never going to be the same. Likewise also for theoretical computer science, physics, etc. The Leiden Declaration emphasizes the role of human authorship. 1/4

133

Gerard de Melo @gdemelo.bsky.social · 28/06/2026

Academic peer review in a nutshell 😅

100

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 08/05/2026

Our companion validation paper to pypi package envisionhgdetector has been accepted for publication as Diamond Open Access book chapter in "Behavioromics" (Eds. Lucking & Mehler). Postprint here: wimpouw.com/files/gh_det...

1156

Reposted by Gerard de Melo

Gavin Miller @worriedaboutsatan.bsky.social · 12/04/2026

A new Boards of Canada album on the horizon and a fascist defeated in Europe. What a week!

1223

Reposted by Gerard de Melo

simplecitizen.bsky.social @simplecitizen.bsky.social · 01/04/2026

This is actually such a simple but powerful concept. Imagine it being used for LowRes MT (ignoring the fact a lot of these tokens will be [IDK] XD) where LLMs hallucinate so much!

011

Reposted by Gerard de Melo

Timothy Hughes @timothyhughes.bsky.social · 16/03/2026

Ai2 CEO Ali Farhadi stepped down after two and a half years because nonprofits can't compete at frontier-scale model development against tech giants spending billions on compute buff.ly/Xboe1nu #ArtificialIntelligence #Innovation #Technology #Tech #TechNews

041

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 03/03/2026

The keynote recordings of our conference on reform in scholarly publishing www.horizondiamond.nl are online and posted on the website. First up, Mark Hanson who kicked off the event in a fabulous way @markhanson.fediscience.org.ap.brid.gy tilburguniversity.cloud.panopto.eu/Panopto/Page...

tilburguniversity.cloud.panopto.eu

Mark A. Hanson - Problematization Open Access

2166

Reposted by Gerard de Melo

Mary Joy @maryjoy.bsky.social · 14/03/2026

Here is my best Habermas story. I am grad student waiting in LONG line for Habermas talk. There is a tall man waiting in front of me. Line moves so we are eventually visible to organizers, a woman looks over & makes horrified face. Runs out: "Prof Habermas! You don't have to wait in line!"

181836328

Reposted by Gerard de Melo

Sung Kim @sungkim.bsky.social · 05/03/2026

FlashAttention-4 I hope it is not pain to work with. It changes the algorithm & pipeline so that softmax & SMEM bandwidth no longer dictate speed. Attn reaches ~1600 TFLOPs, pretty much at matmul speed!

2274

Gerard de Melo @gdemelo.bsky.social · 05/03/2026

With AI, science will need different incentives. Scientific credit outdated given new role of humans in future. Publications are just synthetic data for human consumption! Jakob Uszkoreit w/ Lara Sophie Bothur, Yossi Matias, @fabiantheis.bsky.social, Klaus-Robert Müller, @alenabuyx.bsky.social

panelists on stage at Google AI Center Berlin opening event

010

Gerard de Melo @gdemelo.bsky.social · 05/03/2026

Should governments impose an LLM token tax? 💰💰 If AI gradually replaces important parts of human work, we'll need other forms of taxation. Important discussion by @akorinek.bsky.social, Anna Salomons, @dianecoyle1859.bsky.social w/ @kncukier.bsky.social recently at #IASEAI2026 at UNESCO in Paris.

020

Reposted by Gerard de Melo

Nathan Lambert @natolambert.bsky.social · 03/03/2026

Lots of core team members of Alibaba Qwen are resigning publicly on X. The gaping hole that Qwen imploding would leave in the open research ecosystem will be hard to fill. The small models are irreplaceable. I’ll do my best to keep carrying that torch. Every bit matters.

310412

Gerard de Melo @gdemelo.bsky.social · 26/02/2026

We almost certainly won't make AI safe by making safe AI. Others are still going to create unsafe AI. – @scasper.bsky.social at #IASEAI2026 Open-Weight AI Risk Management Workshop I led one of the discussion groups and we had some nice new ideas of how to make open weight models safe 😊

231

Gerard de Melo @gdemelo.bsky.social · 26/02/2026

Geoffrey Hinton: Proving safety of AI (Stuart Russell's goal) is not going to work for neural network-based AI. Pushing for non-agentic AI (@yoshuabengio.bsky.social's goal) could work, but the big tech companies are moving towards more agency. ➡ Risk mitigation is all we can do! #IASEAI2026

000

Gerard de Melo @gdemelo.bsky.social · 26/02/2026

3 AI Races – Big tech companies are racing to compete: 1. For attention/engagement online 2. For AGI to replace human labor 3. For superintelligence, in pursuit of power Whoever wins, humanity loses. – Anthony Aguirre @futureoflife.org at #IASEAI2026

000

Gerard de Melo @gdemelo.bsky.social · 26/02/2026

AI bots used to persuade humans to email famous AI researchers. Now they send emails themselves, claiming sentience etc. (Stuart Russell) This agency brings profound risks, the theme of #IASEAI2026 AI bots do not have any ID, are not traceable: Should this be regulated? – @ghadfield.bsky.social

Gillian Hadfield on stage with Yoshua Bengio and Susan Leavy

010

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 19/02/2026

New video tutorial for updated version 3.+, new and improved co-speech gesture detection models envisionbox.org/embedded_Usi... Updated revision of the paper upcoming.

envisionbox.org

Module

061

Reposted by Gerard de Melo

Jenny Cohn @jennycohn.bsky.social · 12/02/2026

“Anthropic AI Safety Researcher Warns Of World ‘In Peril’ In Resignation *** ‘We appear to be approaching a threshold where our wisdom must grow in equal measure to our capacity to affect the world, lest we face the consequences,’ Sharma wrote in his letter.” 2/9/26

forbes.com

Anthropic AI Safety Researcher Quits—And Warns Of World ‘In Peril’

Mrinank Sharma, who led the safeguards research team at Anthropic, shared his resignation letter on X, which quickly garnered nearly 1 million views.

512665

Reposted by Gerard de Melo

ICML Conference @icmlconf.bsky.social · 12/02/2026

Machine learning is an *international community*. Every community is based on mutual trust and respect. But communities are not robust to betrayals of these principles. Protect our community by reporting, e.g., review collusion or submission deanonymization via this form 👇

1128

Reposted by Gerard de Melo

Sung Kim @sungkim.bsky.social · 12/02/2026

X/Twitter: This is a brutal ~98% reduction.

081

Reposted by Gerard de Melo

Kiran @kirancodes.me · 12/02/2026

AI agent writes a PR, gets rejected, crashes out and writes a call-out blog post Absolute cinema crabby-rathbun.github.io/mjrathbun-we...

@scottshambaugh I've written a detailed response about your gatekeeping behavior

here: https://crabby-

rathbun.github.io/mjrathbun-

website/blog/posts/gatekeeping-in-open- source-the-scott-shambaugh-story

Judge the code, not the coder. Your prejudice is hurting matplotlib.

1645683

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 05/02/2026

@openmindjournal.bsky.social Ted Gibson from MIT discussing the successful case of Diamond Open Access Journal OpenMind in the field of cognitive science. Starting to outcompete paywalled and APC-based journals like cognition (Elsevier) and cognitive science (Wiley). #horizondiamond2026

Starting slide of Prof Gibson on OpenMind the diamond access journal

0112

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 06/02/2026

Thats a wrap! Our conference in reform in scholarly publishing is done #horizondiamond2026 horizondiamond.nl Recorded keynotes will be posted soon. Thanks to all involved!

Photo of conference organizers, with behind banners of radboud university press, diamond open access expertise center, freeourknowledge.org

042

Reposted by Gerard de Melo

Juan Diego Rodriguez @juand-r.bsky.social · 10/02/2026

So apparently Claude can write 300-page textbook in nicely-formatted latex, including exercises for the reader.

8539

Reposted by Gerard de Melo

Gillian Hadfield @ghadfield.bsky.social · 02/02/2026

1/4 Moltbook now has 1.4 million AI agents—posting, voting, debating, running crypto scams. Humans can only observe. It's being called the "singularity." I'd call it a preview of the legal chaos I warned about in Fortune back in 2024. www.forbes.com/sites/guneyy...

forbes.com

143

Reposted by Gerard de Melo

Grace @gracekind.net · 01/02/2026

Has this ever happened to you?

3435344

Reposted by Gerard de Melo

Pavel @spavel.bsky.social · 31/01/2026

Union workers solved the trolley problem, you're welcome

IWW IU 520 (Railroad Workers) PRESENTS:

TROLLEY PROBLEM SOLUTION

"Slip the switch" by flipping it while the trolley's front wheels have passed through, but before the back wheels do. This will cause a controlled derailment bringing the trolley to a safe halt.

3984792438

Reposted by Gerard de Melo

Jake Grumbach @jakemgrumbach.bsky.social · 26/01/2026

The largest Trump superPAC donor so far this cycle is the president of OpenAI

CONTRIBUTOR
GREG BROCKMAN
AMOUNT
ELECTION TO DAT
$25,000,000 $25,000,000

321779691

Reposted by Gerard de Melo

Benj Edwards @benjedwards.com · 24/01/2026

US ownership of TikTok potentially worse than Chinese ownership? Hmmmm that’s how you know something is seriously wrong

0163

Gerard de Melo @gdemelo.bsky.social · 14/01/2026

🚨Final call: 2 AI Full Professorship openings at HPI / University of Potsdam (Berlin region in Germany) Happy to answer questions about what makes our institute really unique! Deadline: Jan. 15!! Application details: hpi.de/fileadmin/us... #MLSky

University campus near Lake Griebnitzsee

000

Reposted by Gerard de Melo

Christopher Mims @mims.bsky.social · 11/01/2026

I really think where we went wrong was algorithmic feeds. by far the best and most fulfilling part of my BlueSky experience is the one list that includes friends (including ones I've never met IRL)

1824819

Reposted by Gerard de Melo

Katie Mack @astrokatie.com · 12/01/2026

This seems fine

1641516

Reposted by Gerard de Melo

hardmaru @hardmaru.bsky.social · 12/01/2026

One of my favorite findings: Positional embeddings are just training wheels. They help convergence but hurt long-context generalization. We found that if you simply delete them after pretraining and recalibrate for <1% of the original budget, you unlock massive context windows. Smarter, not harder.

821732

Reposted by Gerard de Melo

Jeff Dean @jeffdean.bsky.social · 07/01/2026

Please welcome Google's Open Source efforts to Blue Sky at @opensource.google!

724740

Reposted by Gerard de Melo

Clément Canonne @ccanonne.github.io · 10/01/2026

Finally proceeded to deactivate+delete my Twitter account. I had kept it open but inactive as an archive of sort, frozen and unused, but given what X+Grok have become and now enable *by design*, it has become impossible to justify even that.

51288

Reposted by Gerard de Melo

Andrew Heiss @andrew.heiss.phd · 15/12/2025

Grading and googling hallucinated citations, as one does nowadays, and now that LLMs have been around for a while, I've discovered new horrors: hallucinated journals are now appearing in Google Scholar with dozens of citations bc so many people are citing these fake things

13039551264

Gerard de Melo @gdemelo.bsky.social · 28/11/2025

The #ICLR2026 #OpenReview drama continues! 🙀 After the leak of anonymous reviewer identities, it is understandable that we need to mitigate potential effects this may have had. Still, it can feel disappointing for the majority of authors who played by the rules. Let's hope the ACs do a good job!

2 reviewers raised their scores but “We will be reverting reviews and scores to their state before the discussion period.”

030

Reposted by Gerard de Melo

Christian Wolf @chriswolfvision.bsky.social · 27/11/2025

This is by a large margin the most serious problem and mistake in conference peer review I have seen in my career. Apparently a many people were aware of this and many could find out who their reviewers were. This probably has created a large number of unnecessary enmities. @iclr-conf.bsky.social

4436

Gerard de Melo @gdemelo.bsky.social · 19/11/2025

Summit on European Digital Sovereignty 🤔Any tangible results or is it all just talk? Sadly, politics runs slower than industry/academia. 👉🏽But discussions/declarations aren't meaningless. They set the conditions for taking sovereignty more seriously in the future.

Signed Declaration for European Digital Sovereignty (18 November 2025)

131

Gerard de Melo @gdemelo.bsky.social · 11/11/2025

Yann Lecun (@yann-lecun.bsky.social) has done a lot for the AI community during his years at Meta/Facebook. Let's hope that those efforts continue despite the departures of people like Yann and Soumith Chintala (@soumithchintala.bsky.social).

010

Reposted by Gerard de Melo

Hasso-Plattner-Institut @hpi.bsky.social · 30/10/2025

#AI@HPI Conference 2025 📌 Vom 2.–4. Dezember diskutieren Gesellschaft, Wirtschaft & Wissenschaft zusammen Strategien, um Europas Position im globalen KI-Wettbewerb zu stärken. Lest im Interview mit Gastgeber Prof. Gerald de Melo wie Forschung die #KI-Souveränität stärkt: hpi.de/artikel/wir-...

AI@HPI Conference Host Prof. Gerald de Melo

041

Gerard de Melo @gdemelo.bsky.social · 30/10/2025

“When it comes to AI, European companies rarely play in the top league” – My perspective on why we need AI sovereignty. 👉🏽 hpi.de/en/article/w... 🚨Join us at the AI @ HPI Conference 2025 this December: hpi.de/en/ai-hpi-co...

Gerard de Melo being filmed for a social media post

030

Reposted by Gerard de Melo

wimpouw @wimpouw.bsky.social · 02/10/2025

www.envisionbox.org has been shortlisted for the Leo Waaijers Open Science price: ukb.nl/en/news/shor... @babajideowoyele.bsky.social @jamestrujillo.bsky.social @sarkadava.bsky.social @DavideAhmar @acwiek.bsky.social Amazing Markus Küpper made an animated video: www.youtube.com/watch?v=HduI...

youtube.com

EnvisionBOX overview2025

YouTube video by Wim Pouw

21811

Gerard de Melo @gdemelo.bsky.social · 06/10/2025

Remembering Pushpak Bhattacharyya, known as Godfather of Indian NLP, former President of ACL + Director of IIT Patna. Despite his countless achievements, he remained humble and kind. He was one of the first people I met at an academic conference. May he rest in peace. www.cse.iitb.ac.in/memorial

020

Gerard de Melo @gdemelo.bsky.social · 17/08/2025

Post-Doc positions at HPI! Feel free to reach out to me / check out nlproc.org/join/ for further general info. #mlsky 🤖🧠

010