Matthew Lutz
@matthewjlutz
Anticipating emergent capabilities in multi-agent AI systems 🐜@apartresearch
Lots of people have reached out to ask what @collect_intel is working on this year so here you are: ✨ Scenario Construction. This tool will help policymakers generate scenarios, stress-test policies, and gather public input in high-uncertainty settings.
hello new followers! my collaborator and i have decided to host the pdf of Disconnection Practices online. it's available here cath.land/zines/disconne…
worked on a zine a while ago called "Disconnection Practices" that deals with the same stuff. i could send you a pdf.
Almost time for @ALifeConf where we will present our work "Collective Innovation in Groups of Large Language Models" We study how social structure affects how LLMs collectively play the game Little Alchemy 2 🔥+🌽=🍿 Work done with @Clement_MF_ @pyoudeyer @criticalneuro @risi1979
In our new paper, we show that adding self-modeling to artificial networks causes a significant reduction in network complexity When artificial networks learn to predict their internal states as an auxiliary task, they change in a fundamental way
Waymo car failing to coordinate w/ another Waymo (credits in the comment). Interesting to see a toy example from my grant applications play out in the real world. Two cars playing a best-response to a human driver model are not mutually compatible, multi-agent challenges are real
Amazing collective behavior in the immune system.
worked on a zine a while ago called "Disconnection Practices" that deals with the same stuff. i could send you a pdf.
Happy #sharkweek! We are proud sharkologists with @icouzin @AugustPala @po7ooo Luke Costello @MPI_animalbehav @CBehav @UniKonstanz this year again in the Maldives to film blacktip reef sharks and schools of silversides at @MaRHECenter and @diamonds_athu @KuramathiISLAND
What can we learn from swarm behavior and implement it to perform diverse tasks in various environments? @NatureComms calls for papers on Swarm Intelligence #Swarm #intelligence #Collective #Collections Learn more about the collection: nature.com/collections/cg…
Machine learning is part science, part wizardry. People who try to claim it's all science just don't respect wizardry enough.
How I published 0 papers during my last postdoc (supposed to be 2.5 years, ended early): - Postdoc ended after 1.5 years because my supervisor (who I respect dearly) got a new position in the EU and the grant was tied to him, so UK money could not follow. The Uni (I won't out…
How I published 13 first-author papers during my 2.5 year post-doc: - All the data were already collected - 5 papers were literature reviews - I read incredibly fast - I write even faster
New ant phylogeny dropped:
How I published 13 first-author papers during my 2.5 year post-doc: - All the data were already collected - 5 papers were literature reviews - I read incredibly fast - I write even faster
What an interesting article! We lab/fieldwork/data analysis biologists URGENTLY need to collaborate with philosophers of biology.
The question "are LLMs conscious" is less interesting than the question whether human minds are essentially a kind of LLM
Ants perform surgery on injured nestmates depending on the site of injury... So, this morning I'm chatting with @peterboghossian about insectoid aliens operating on DMT users and then this incredible paper pops up on my feed (thanks to @Graham__Hancock): Ants inspect the leg…
Occasionally you find science diagrams that would have belonged in House of Leaves -__-
🚨We are delighted to announce that our new paper is out NOW in @CurrentBiology 🦑🌅Here we show how #cuttlefish alter their #camouflage response under naturally occuring dynamic lighting ('water caustics')! 🔗authors.elsevier.com/sd/article/S09… 🧵A thread to find out more:
Computational Life: How Well-formed, Self-replicating Programs Emerge from Simple Interaction arxiv.org/abs/2406.19108