banner
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo
@dcm.social.sunet.se.ap.brid.gy
360 followers 21 following 43 posts
Assistant Professor in Philosophy of AI at Umeå University, working on and at the foundations of the sciences of mind and cognition. Searchable through tootfinder. [bridged from https://social.sunet.se/@dcm on the fediverse by https://fed.brid.gy/ ]
social.sunet.se
Posts Media Videos Starter Packs
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · 6d
Cool little adapted tale on AI, by Alison Gopnik:
https://simons.berkeley.edu/news/stone-soup-ai
Stone Soup AI
For some time, I’ve argued that a common conception of AI is misguided. This is the idea that AI systems like large language and vision models are individual intelligent agents, analogous to human agents. Instead, I’ve argued that these models are “cultural technologies” like writing, print, pictures, libraries, internet search engines, and Wikipedia.
simons.berkeley.edu
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · 29d
Interesting and worrying article by Marcus and Hamiel going over recent work exploring the very serious security weaknesses of agentic AI:

https://garymarcus.substack.com/p/llms-coding-agents-security-nightmare

#ai #cybersec
大型语言模型（LLMs）和编码代理是安全噩梦。
LLMs and coding agents are a security nightmare (garymarcus.substack.com) 08-18  ↑ 102 HN Points
garymarcus.substack.com
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Aug 7
A quick note to mention that the preprint above, 'AI Mimicry and Human Dignity: Chatbot Use as a Violation of Self-Respect' has turned into a revised, published paper, just out open access in the Journal of Applied Philosophy.

https://onlinelibrary.wiley.com/doi/epdf/10.1111/japp.70037 […]
Original post on social.sunet.se
social.sunet.se
1
Reposted by Dimitri Coelho Mollo
pbloem.sigmoid.social.ap.brid.gy
Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · Jul 8
Veo3 shows that particular blindness that tech people have to the way culture develops.

It looks impressive now, but that's because it usually takes a lot of work to film shots like this. Once people learn to recognize video AI's "style", that style becomes […]

[Original post on sigmoid.social]
A tweet advertising a Turkish music video made with Veo3.
1 1
Reposted by Dimitri Coelho Mollo
tomstafford.mastodon.online.ap.brid.gy
Tom Stafford @tomstafford.mastodon.online.ap.brid.gy · Jul 7
Yuguang Yang has shared a textbook about LLMs here

https://yangyutu.github.io/llm_book.github.io

"This book aims to provide a comprehensive understanding of Large Language Models, from their foundational concepts to their practical applications, with a special focus on their role in […]
Original post on mastodon.online
mastodon.online
2 2
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jul 3
Interesting reporting and analysis by Gordon Hull about recent decisions in the US courts about AI and copyright:

https://www.newappsblog.com/2025/07/ai-and-copyright-training-data-and-transformative-fair-use.html

#aiethics #ailaw
3
Reposted by Dimitri Coelho Mollo
0xabad1dea.infosec.exchange.ap.brid.gy
abadidea @0xabad1dea.infosec.exchange.ap.brid.gy · Jul 1
Wikipedia has a cheat sheet of well-known tells for identifying generated text. (With an appropriate warning not to over-index on minor ones as absolute proof) https://en.m.wikipedia.org/wiki/Wikipedia:WikiProject_AI_Cleanup/AI_catchphrases
Wikipedia:WikiProject AI Cleanup/AI catchphrases - Wikipedia
en.m.wikipedia.org
5 83 31
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 29
Some good venting by Steve Klabnik about the sorry state of significant chunks of the AI debate today:

"What is breaking my brain a little bit is that all of the discussion online around AI is so incredibly polarized. This isn’t a “the middle is always right” sort of thing either, to be clear […]
Original post on social.sunet.se
social.sunet.se
2
Reposted by Dimitri Coelho Mollo
fintanmallory.com
Fintan Mallory @fintanmallory.com · Jun 27
He’s called Telemachus
Front of a posh clothes shop called ‘nobody’s child’
1 5 64
Reposted by Dimitri Coelho Mollo
facultyofphilosophygroningen.social.edu.nl.ap.brid.gy
UG Faculty of Philosophy @facultyofphilosophygroningen.social.edu.nl.ap.brid.gy · Jun 27
#job alert! Fully funded 4-year #phd position on the #philosophy of #datascience in relation to #psychiatry and #psychopathology at the department of #theoreticalphilosophy in #groningen.

All details here: https://www.rug.nl/about-ug/work-with-us/job-opportunities/?details=00347-02S000BDMP […]
Original post on social.edu.nl
social.edu.nl
2 4
Reposted by Dimitri Coelho Mollo
simon.fedi.simonwillison.net.ap.brid.gy
Simon Willison @simon.fedi.simonwillison.net.ap.brid.gy · Jun 24
There are some interesting details about how Anthropic trained their models tucked away in today's summary judgement: they bought, chopped up and scanned millions of dollars worth of books! https://simonwillison.net/2025/Jun/24/anthropic-training/
Anthropic wins a major fair use victory for AI — but it’s still in trouble for stealing books
Major USA legal news for the AI industry today. Judge William Alsup released a "summary judgement" (a legal decision that results in some parts of a case skipping a trial) …
simonwillison.net
1 13 10
Reposted by Dimitri Coelho Mollo
fintanmallory.com
Fintan Mallory @fintanmallory.com · Jun 15
This looks like a valuable paper for any educational institution interested in developing an evidence-based policy on LLM usage (if such an institution exists)
dryohanjohn.bsky.social
Yohan J John @dryohanjohn.bsky.social · Jun 15
"Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels."

arxiv.org/abs/2506.08872
Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task
This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed thr...
arxiv.org
1 5 21
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 10
Through these pathways, LLMs can establish connections to the world sufficient for intrinsic meaning. One potentially surprising implication of our discussion is that that multimodality and embodiment are neither necessary nor sufficient to overcome the Grounding Problem.

\end
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 10
Second, drawing on philosophical theories of representational content, we provide two arguments for the claim that LLMs and related systems can achieve referential grounding: (1) through preference fine-tuning methods that explicitly establish world-involving functions, and (2) through […]
Original post on social.sunet.se
social.sunet.se
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 10
... can possess intrinsic meaning that is not parasitic on external interpretation. Although modern LLMs compute over vectors rather than symbols, an analogous problem arises for these systems, which we call the Vector Grounding Problem. This paper has two main goals. First, we distinguish five […]
Original post on social.sunet.se
social.sunet.se
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 10
The remarkable performance of large language models (LLMs) on complex linguistic tasks has sparked debate about their capabilities. Unlike humans, these models learn language solely from textual data without directly interacting with the world. Yet they generate seemingly meaningful text on […]
Original post on social.sunet.se
social.sunet.se
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 10
A new, updated, streamlined, and generally improved version of The Vector Grounding Problem paper, joint work by @raphaelmilliere and me on the meaningfulness or else of LLM outputs and internal representations is now available on ArXiv.

https://arxiv.org/abs/2304.01481

New abstract in the […]
Original post on social.sunet.se
social.sunet.se
2 3 2
Reposted by Dimitri Coelho Mollo
edouardmachery.bsky.social
Edouard Machery @edouardmachery.bsky.social · Jun 9
Open access
Underdetermination and Theoretical Virtues
Cambridge Core - Philosophy of Science - Underdetermination and Theoretical Virtues
www.cambridge.org
7 23
Reposted by Dimitri Coelho Mollo
mistakenotmy.mastodon.social.ap.brid.gy
Wolf Ha @mistakenotmy.mastodon.social.ap.brid.gy · Jun 8
"These insights challenge prevailing assumptions about LRM capabilities and suggest that current approaches may be encountering fundamental barriers to generalizable reasoning."

While it's a bit of a "Well, I could have told you that" style of hypothesis […]

[Original post on mastodon.social]
2
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 5
We also offer some suggestions on how to improve things by arguing for the need of sociotechnical, rather than purely technical, work on increasing the safety of LLMs, and large connectionist systems more generally.

Such considerations, we argue, should not only inform our tackling of the […]
Original post on social.sunet.se
social.sunet.se
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 5
We try and shed some light on the origins and meaning (or lack thereof) of the criteria commonly used in RLF, i.e. the 3Hs: helpfulness, harmlessness and honesty, and we examine the assumptions underlying the very notion of alignment.

We show the internal tensions and ethical risks of RLF as […]
Original post on social.sunet.se
social.sunet.se
1
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 5
Another of my forays into AI ethics is just out! This time the focus is on the ethics (or lack thereof) of Reinforcement Learning Feedback (RLF) techniques aimed at increasing the 'alignment' of LLMs.

The paper is fruit of the joint work of a great team of collaborators, among whom @pettter and […]
Original post on social.sunet.se
social.sunet.se
1 6 3
Reposted by Dimitri Coelho Mollo
fintanmallory.com
Fintan Mallory @fintanmallory.com · Jun 5
Bach and Handel never met but they were both blinded by the same quack oculist: en.m.wikipedia.org/wiki/John_Ta...
John Taylor (c. 1703 - 1770 or 1772) was an early British eye surgeon, self-promoter and medical charlatan of 18th-century Europe. He was responsible for the surgical mistreatment of George Frideric Handel, Johann Sebastian Bach, and perhaps hundreds of others. Both Handel and Bach died shortly after the botched surgery performed by Taylor. [1)
2 5 23
Reposted by Dimitri Coelho Mollo
felwert.fedihum.org.ap.brid.gy
Frederik Elwert @felwert.fedihum.org.ap.brid.gy · Jun 5
I can’t help it, I somehow feel gaslighted by the whole #genai debate. People who are critical of GenAI are often told “but when done right, you’re just so much more productive, so obviously you just didn’t to it right.” So I’m trying. Not because I feel I need to, but because I just want to get […]
Original post on fedihum.org
fedihum.org
1 8 6
dcm.social.sunet.se.ap.brid.gy
Dimitri Coelho Mollo @dcm.social.sunet.se.ap.brid.gy · Jun 1
Just read that Mozilla has decided to shut down Pocket (!?!). Its integration with Kobo e-readers is great and something I will miss. It will be difficult to find a similarly practical way to read web articles on my e-reader, other than the somewhat clunky pdf-saving + cloud sync method. Any […]
Original post on social.sunet.se
social.sunet.se
2