Moa Johansson

80 years ago today, the USA dropped an atomic bomb on Hiroshima.

While we speak less about the horrors of nuclear war, the dangers are as real as ever.

Somehow we need to rediscover our common humanity.

Somehow we need to relearn to work together, not blow each other to bits.

📄📄📄 The AI4Mat-NeurIPS-2025 workshop is now open for submissions until August 22, 2025 (AOE)!

Consider submitting full-length papers or shorter-length findings. We also have a special track for papers on benchmarking AI for materials design.

sites.google.com/view/ai4mat/...

The End of Publishing as We Know It Inside Silicon Valley’s assault on the media

Depressing but important article on how GenAI is destroying the publishing industry.

www.theatlantic.com/technology/a...

AI and Fraternity, Abeba Birhane, AI Accountability Lab I envision a future where human dignity, justice, peace, kindness, care, respect, accountability, and rights and freedoms serve as the north stars that guide AI development and use. Realising these ideals can’t happen without intentional tireless work, dialogues, and confrontations of ugly realities – even if they are uncomfortable to deal with. This starts with deciphering hype from reality. Pervasive narratives portray AI as a magical, fully autonomous entity approaching a God-like omnipotence and omniscience. In reality, audits of AI systems reveal a consistent failure to deliver on grandiose promises and suffer from all kinds of shortcomings, issues often swept under the rug. AI in general, and GenAI in particular, encodes and exacerbates historical stereotypes, entrenches harmful societal norms, and amplifies injustice. A robust body of evidence demonstrates that — from hiring, welfare allocation, medical care allocation to anything in between — deployment of AI is widening inequity, disproportionately impacting people at the margins of society and concentrating power and influence in the hands of few. Major actors—including Google, Microsoft, Amazon, Meta, and OpenAI—have willingly aligned with authoritarian regimes and proactively abandoned their pledges to fact-check, prevent misinformation, respect diversity and equity, refrain from using AI for weapons development, while retaliating against critique. The aforementioned vision can’t and won’t happen without confrontation of these uncomfortable facts. This is precisely why we need active resistance and refusal of unreliable and harmful AI systems; clearly laid out regulation and enforcement; and shepherding of the AI industry towards transparency and accountability of responsible bodies. "Machine agency" must be in service of human agency and empowerment, a coexistence that isn't a continuation of modern tech corporations’ inequality-widening,

so I am one of the 12 people (including the “god-fathers of AI”) that will be at the Vatican this September for a two full-day working group on the Future of AI

here is my Vatican approved short provocation on 'AI and Fraternity' for the working group

Unfortunately not at #cogsci2025 either, but check out the oral presentation by Andrea Silvi tomorrow!

Fact Recall, Heuristics or Pure Guesswork? Precise Interpretations of Language Models for Fact Completion Denitsa Saynova, Lovisa Hagström, Moa Johansson, Richard Johansson, Marco Kuhlmann. Findings of the Association for Computational Linguistics: ACL 2025. 2025.

I'm not at #acl2025, but my student Denitsa Saynova and our collaborators are there to present this paper: aclanthology.org/2025.finding...

Why the Key to a Mathematical Life is Collaboration | Quanta Magazine Fan Chung, who has an Erdős number of 1, discusses the importance of connection — both human and mathematical.

The mathematician Fan Chung has authored over 300 papers with more than 150 coauthors. Her work has contributed to problems in computer science, engineering and physics. In an interview with Quanta, Chung discusses human and mathematical connection:

www.quantamagazine.org/why-the-key-...

💐 Stage winner 💐

Marianne Vos 🇳🇱

#TDFF2025 l #TDFFLive

I love the #eurips initiative! But to live up to its potential, it should be accepted an official #neurips conference location (similar to Mexico City) and not just an addon! Then we would save C02, rather than adding to it!!! What can we do to achieve this?

Neither LLMs nor LRMs have the ability to go beyond the humanity's knowledge closure--which is needed for true discoveries. | Subbarao Kambhampati Neither LLMs nor LRMs have the ability to go beyond the humanity's knowledge closure--which is needed for true discoveries. Both are beholden to the collected knowledge of the humanity (whether de...

Both LLMs and LRMs are upper bounded by humanity's knowledge closure. True scientific discoveries are, by definition, outside of that closure. Ergo, LLMs/LRMs are great force multipliers to us; but don't support "Nobel this weekend" hype..

👉 www.linkedin.com/posts/subbar...

My thoughts on the crucial importance of methodology on self-reported AI performance on mathematics competitions, and my policy on commenting on such reports going forward: mathstodon.xyz/@tao/1148814...

Do models really understand humans (ToM)?
Can your model do it?
Beat them at the neurips Competition.
🤖📈🧠

And this is just one side of room in the poster session.

Huge interest in #AI4math workshop at #ICML2025!

YouTube video by Code Sync Keynote: AI for Mathematical Discovery: (...) Neuro-Symbolic Methods - Moa Johansson |Lambda Days 25

I could also have talked about #conjecturing, like I did recently at @lambdadays.bsky.social but it's available on YouTube: www.youtube.com/watch?v=rLr6... and we will have a poster about the LEMMANAID system in the workshop!

Come and talk to me about numbers, conjecturing or similar topics :D

Tomorrow Friday at 1pm I'm giving a talk in the #AI4math workshop at #icml2025 about how efficient recursive number systems can emerge via communication and RL!

Are you passionate about AI & molecular engineering? Just ‼️one‼️ week left to apply to the 2 PhD positions currently open in our team at Chalmers! 🎓 Apply to join us in here in lovely Gothenburg, Sweden 🇸🇪🤩

For details: ailab.bio/join-us

A rightwing minister told Sweden to get tough on crime – until his own son was caught in a Nazi scandal | Martin Gelin In our current political climate, why is migration minister Johan Forssell surprised that a teenage boy might be drawn to extremism? asks journalist Martin Gelin

"A government that claims to take concerns over crime, safety and national security seriously while ignoring far-right extremism will struggle with credibility. Sweden has a long tradition of refusing to face up to its racism – past and present."

www.theguardian.com/commentisfre...

… on a cross-disciplinary project about neuro-symbolic AI for applied mechanics with applications for recycled aluminium and sustainable materials. More info here or ping me for a chat in Vancouver: www.chalmers.se/en/about-cha...

I’ll be at #ICML next week to give an invited talk in the #AI4math workshop, and of course to meet as many of you interested in reasoning, cognition and neuro-symbolic AI etc!

I’m also looking for a PhD student to join my group…

Quote from our blogpost by author Hannah Ruschemeier: "In the current political climate, reopening the GDPR under slogans like 'reducing bureaucracy' and 'deregulation' risks more than minor tweaks. It could erode accountability and fundamental rights protection."

How promising is reforming the GDPR?

HANNAH RUSCHEMEIER argues that reforms can only be effective if they don’t follow the current trend of deregulation – but instead tackle the structural problems of the GDPR. Not less regulation, better regulation!

verfassungsblog.de/reforming-th...

Happy to share the AI for Accelerated Materials Design (AI4Mat) workshop was accepted at NeurIPS 2025, to be held Dec 6 or 7, 2025 in San Diego!!! 🥳🥳🥳

sites.google.com/view/ai4mat

a couple of hours before my keynote, I went through an intense negotiation with the organisers (for over a hour) where we went through my slides and had to remove anything that mentions 'Palestine' 'Israel' and replace 'genocide' with 'war crimes'

1/

Researchers Jailbreak AI by Flooding It With Bullshit Jargon LLMs don’t read the danger in requests if you use enough big words.

Official corporate AI red teams: after months of careful prompt engineering, we found an elaborate scenario where we tricked the LLM into threatening the user with blackmail

Academics: we flooded the LLM with meaningless jargon and now it does whatever we want

www.404media.co/researchers-...

PhD student position in neuro-symbolic AI in my group. Cross-disciplinary project with applied mechanics solving problems with recycling aluminium, contributing to sustainability: www.chalmers.se/en/about-cha...

How did I miss that @letourfr.bsky.social starts today!

Congratulations Simon!

From memories to maps: Mechanisms of in context reinforcement learning in transformers Humans and animals show remarkable learning efficiency, adapting to new environments with minimal experience. This capability is not well captured by standard reinforcement learning algorithms that re...

Humans and animals can rapidly learn in new environments. What computations support this? We study the mechanisms of in-context reinforcement learning in transformers, and propose how episodic memory can support rapid learning. Work w/ @kanakarajanphd.bsky.social : arxiv.org/abs/2506.19686

Ny studie: Klimatet viktigare än billig bensin enligt svenska väljare

Två av tre väljare vill att Sverige tar täten i det globala klimatarbetet. En ny stor studie visar att stödet för ambitiös klimatpolitik är brett förankrat i befolkningen – även bland högerväljare.

Så ska Vänsterpartiet vinna Norrland Elen, skogen och gruvorna ska ge återbäring till glesbygden.

På tiden: Elen, skogen och gruvorna ska ge återbäring till glesbygden. www.etc.se/inrikes/saa-...