Iris van Rooij 💭
@irisvanrooij.bsky.social
Professor of Computational Cognitive Science | Dept. of Cognitive Science & Artificial Intelligence | @[email protected] on 🦣 | http://irisvanrooijcogsci.com | she/they 🏳️🌈
Pinned
Computational complexity for hype immunity
Using some of my own work, I set out to clear up a few relevant concepts from theoretical computer science and computational cognitive science that can help to build conceptual hygiene and immunity…
irisvanrooijcogsci.com
"Using some of my own work, I set out to clear up a few relevant concepts from theoretical computer science and computational cognitive science that can help to build conceptual hygiene and immunity against hyped claims about “AI” capabilities."
irisvanrooijcogsci.com/2026/09/12/c...
irisvanrooijcogsci.com/2026/09/12/c...
We've been (re)reading 'Models as Mediators' by Mary Morgan & Margaret Morrison in @compcogsci.bsky.social lab. Absolutely invigorated & inspired by today's discussion, where we all shared lessons/insights we took away. Many thanks to @olivia.science for always invaluable extra contextualisations.
October 12, 2026 at 6:04 PM
We've been (re)reading 'Models as Mediators' by Mary Morgan & Margaret Morrison in @compcogsci.bsky.social lab. Absolutely invigorated & inspired by today's discussion, where we all shared lessons/insights we took away. Many thanks to @olivia.science for always invaluable extra contextualisations.
Reposted by Iris van Rooij 💭
Recommendations from the Journals Working Group of the Association for Human Mathematics just dropped: Venues that disseminate scientific publications should give authors *choice* and *consent* in whether and how LLM-based systems are used during the referee process. www.ahmath.org/recommendati...
October 12, 2026 at 3:45 PM
Recommendations from the Journals Working Group of the Association for Human Mathematics just dropped: Venues that disseminate scientific publications should give authors *choice* and *consent* in whether and how LLM-based systems are used during the referee process. www.ahmath.org/recommendati...
Reposted by Iris van Rooij 💭
Reading 'Matcha on Monday' while drinking matcha on Monday 🙃☺️📚🍵🩷💚
October 12, 2026 at 1:31 PM
Reading 'Matcha on Monday' while drinking matcha on Monday 🙃☺️📚🍵🩷💚
Reposted by Iris van Rooij 💭
To all my LGBTQ🏳️🌈 family, friends and neighbors:
I👁️U
I🫂U
I appreciate U
This is a safe place for you.
I👁️U
I🫂U
I appreciate U
This is a safe place for you.
October 11, 2026 at 5:35 PM
To all my LGBTQ🏳️🌈 family, friends and neighbors:
I👁️U
I🫂U
I appreciate U
This is a safe place for you.
I👁️U
I🫂U
I appreciate U
This is a safe place for you.
Reposted by Iris van Rooij 💭
It's toxic and it's promoters are lying about the longterm risks.
October 11, 2026 at 3:23 PM
It's toxic and it's promoters are lying about the longterm risks.
Reposted by Iris van Rooij 💭
The difference: we are women scientists 🤐
I also note: philosophers of science X when meeting scientists in X have the reaction that we are wrong instead of studying us as per usual. Meanwhile when we speak out against others in the same field in reviews philosophers say "you can't say scientists are wrong that's not right" — which is it?
October 11, 2026 at 8:07 AM
The difference: we are women scientists 🤐
Wrote this in 2024, only more relevant today ...
I do not understand why not more of my colleagues in the field of AI are worried about this. Why are they not worried about how AI hype discredits the entire field?
October 11, 2026 at 3:15 PM
Wrote this in 2024, only more relevant today ...
Reposted by Iris van Rooij 💭
October 10, 2026 at 7:47 PM
Reposted by Iris van Rooij 💭
Critical refusal and critical literacies have been lampooned in ed tech circles as "burying your head in the sand." In fact, they reflect awareness of the full spectrum of options in responding to emerging tech and the commercial products being pushed on schools.
"Of course, we have to teach our students about AI technologies. Teaching about AI technologies should be just like how we teach ‘no smoking’ or the causal links between lung cancer and cigarette smoke; yet, we do not teach students how to roll cigarettes and smoke them." zenodo.org/records/2008...
Against the Uncritical Adoption of 'AI' Technologies in Academia
Artificial intelligence (AI) companies and their rhetoric infringe on academia in harmful ways, mirroring past uncritical acceptance of industry logics, such as those of tobacco and petroleum. In this...
zenodo.org
October 11, 2026 at 2:48 PM
Critical refusal and critical literacies have been lampooned in ed tech circles as "burying your head in the sand." In fact, they reflect awareness of the full spectrum of options in responding to emerging tech and the commercial products being pushed on schools.
Reposted by Iris van Rooij 💭
> [that] ‘The task of a university is the creation of the future, so far as rational thought, and civilised modes of appreciation, can affect the issue. The future is big with every possibility of achievement and of tragedy.'
PDF: hcommons.org?get_group_do...
PDF: hcommons.org?get_group_do...
hcommons.org
October 11, 2026 at 8:19 AM
> [that] ‘The task of a university is the creation of the future, so far as rational thought, and civilised modes of appreciation, can affect the issue. The future is big with every possibility of achievement and of tragedy.'
PDF: hcommons.org?get_group_do...
PDF: hcommons.org?get_group_do...
Reposted by Iris van Rooij 💭
> Those academics who just ask for time to think – who do not name those putting pressure on them, preferring to address ‘society’ and ask for protection – do not feel there is an option at all.
October 11, 2026 at 8:19 AM
> Those academics who just ask for time to think – who do not name those putting pressure on them, preferring to address ‘society’ and ask for protection – do not feel there is an option at all.
Reposted by Iris van Rooij 💭
it's this also
> Those academics who just ask for time to think – who do not name those putting pressure on them, preferring to address ‘society’ and ask for protection – do not feel there is an option at all.
October 11, 2026 at 8:39 AM
it's this also
Reposted by Iris van Rooij 💭
Some selected extracts also here: olivia.science/turing
Turing test
Critically analysing the Turing test.
olivia.science
October 11, 2026 at 9:03 AM
Some selected extracts also here: olivia.science/turing
Reposted by Iris van Rooij 💭
Thing of beauty ✨
I did a thing!! I made a rendition of the song The Man with Money in his Hands by Jessie Mazin ❤️ This song was just perfect to be made into an anti-AI anthem. "Make life easier, just give up control. It's a chemical dependance, it's a bruise on your soul"
Enjoy!!!!!!! 😃
youtu.be/i6Sbu2Zf2SM?...
Enjoy!!!!!!! 😃
youtu.be/i6Sbu2Zf2SM?...
The man with money in his hands - Jessie Mazin (cover and edit by Ajuna Soerjadi)
YouTube video by Ajuna Soerjadi
youtu.be
August 4, 2026 at 8:51 AM
Thing of beauty ✨
Reposted by Iris van Rooij 💭
Reposted by Iris van Rooij 💭
yup, this kind of point really upsets certain people when we make it CC @andreaeyleen.eurosky.social philpapers.org/rec/GUEANR
![Furthermore, for a simple case of understanding why reading
something off an entity is not relevant to the entity’s represen-
tational powers, consider the following case: A brick in a wall
exposed to the elements can be subject to measurement of its
temperature and moisture, and these measurements will corre-
late with the weather. Does this correlation allow us to conclude
that the brick represents the weather? If yes, then it is possible
one is either a panrepresentationalist, i.e. everything represents.
And under the computationalist flavour one could thus ascribe to
pancomputationalism, i.e. everything computes something more
than the identity function. Alternatively, one does not believe in
representations at all, and therefore one allows this absurd con-
clusion as a reductio argument.
The [person] on the street acknowledges that minds
are rather mysterious, but [they are] definitely sure
that a mind is something that you either have or you
haven’t. Bricks haven’t.
Edwin Ray Guthrie (1935, p. 1)
Trying to neither fully deflate representation nor assign every-
thing representational abilities, neurocognitive scientists likely
want to be very careful when they conclude from brain read-outs
that representation is indeed taking place. In this context, there
is nothing that licenses correlational read-outs to be indicative of
the nature of, structure of, or even existence of, representations,
map-like or otherwise.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vsyj5jtayxcq5dw5mtja4xtb/bafkreibks6t3cokn22s2kq6hlrsavtppxg3xdahoarjtfp2d7cm6ktgy3y)
October 11, 2026 at 11:40 AM
yup, this kind of point really upsets certain people when we make it CC @andreaeyleen.eurosky.social philpapers.org/rec/GUEANR
Reposted by Iris van Rooij 💭
fun fact! everything a LLM can do, you can do too with an abacus!
(seriously. for softmax, just use taylor series.)
this means that with enough time, beads, and determination, you can make a wooden object feel pain!
or perhaps you feel the pain. or maybe the universe does. it goes somewhere!
(seriously. for softmax, just use taylor series.)
this means that with enough time, beads, and determination, you can make a wooden object feel pain!
or perhaps you feel the pain. or maybe the universe does. it goes somewhere!
The OP has the right to limit us from continuing this thread of course, but I feel there's value in questioning why well studied "pain" can be so freely given away (e.g. described merely as a pattern) and flattened (e.g. doesn't require nerves) by AI adjacent scholars. Pain has a storied history.
There's value in zooming out. What's the benefit of deflating the meaning of pain so that now plants feel pain? "Pain" as found in animals for example undergirds movements like veganism. If plants can now feel pain haven't we deflated it so far that it's now meaningless? Again: who does that serve?
October 11, 2026 at 11:42 AM
fun fact! everything a LLM can do, you can do too with an abacus!
(seriously. for softmax, just use taylor series.)
this means that with enough time, beads, and determination, you can make a wooden object feel pain!
or perhaps you feel the pain. or maybe the universe does. it goes somewhere!
(seriously. for softmax, just use taylor series.)
this means that with enough time, beads, and determination, you can make a wooden object feel pain!
or perhaps you feel the pain. or maybe the universe does. it goes somewhere!
Reposted by Iris van Rooij 💭
> Theoretical such models bidirectionally link theory to phenomena and observations furthering our understanding of the entities on both sides of this relationship.
Slight update to our — @andreaeyleen.eurosky.social @irisvanrooij.bsky.social — preprint, I added this table to help tease these out:
> A glossary for some of the key terminology we use to discuss computational cognitive science.
doi.org/10.5281/zeno...
> A glossary for some of the key terminology we use to discuss computational cognitive science.
doi.org/10.5281/zeno...
October 11, 2026 at 9:14 AM
> Theoretical such models bidirectionally link theory to phenomena and observations furthering our understanding of the entities on both sides of this relationship.
Reposted by Iris van Rooij 💭
this is a perennial misunderstanding within and outside computational modelling, wherein even so-called allies think that it's a good use of models to do this — but actually models should be used to understand if our stated principles do what we want, align with our values and our understandings
yes, if you can use a model to question the phenomenon in such a way then you have misunderstood how models / phenomena / theories relate. It's like saying if you have a bug in the implementation of a gravity model that that now means things float upwards in reality too.
October 11, 2026 at 9:11 AM
this is a perennial misunderstanding within and outside computational modelling, wherein even so-called allies think that it's a good use of models to do this — but actually models should be used to understand if our stated principles do what we want, align with our values and our understandings
Reposted by Iris van Rooij 💭
The strength of confirmation bias is remarkable in such situations:
Recently, I've observed the same individual phils approving (inaccurate) criticism by an avowed non-expert of a paper authored by multiple experts then go on to tell an expert they aren't expert enough to criticize a Nobel laureat🤷♂️
Recently, I've observed the same individual phils approving (inaccurate) criticism by an avowed non-expert of a paper authored by multiple experts then go on to tell an expert they aren't expert enough to criticize a Nobel laureat🤷♂️
I also note: philosophers of science X when meeting scientists in X have the reaction that we are wrong instead of studying us as per usual. Meanwhile when we speak out against others in the same field in reviews philosophers say "you can't say scientists are wrong that's not right" — which is it?
October 11, 2026 at 8:30 AM
The strength of confirmation bias is remarkable in such situations:
Recently, I've observed the same individual phils approving (inaccurate) criticism by an avowed non-expert of a paper authored by multiple experts then go on to tell an expert they aren't expert enough to criticize a Nobel laureat🤷♂️
Recently, I've observed the same individual phils approving (inaccurate) criticism by an avowed non-expert of a paper authored by multiple experts then go on to tell an expert they aren't expert enough to criticize a Nobel laureat🤷♂️
Reposted by Iris van Rooij 💭
I am confident because I spent 100x longer on something than the other person, and I am not when I have not. But this comes off to them as me being stupid. As @irisvanrooij.bsky.social says: to a non-expert expertise looks wrong.
October 11, 2026 at 11:01 AM
I am confident because I spent 100x longer on something than the other person, and I am not when I have not. But this comes off to them as me being stupid. As @irisvanrooij.bsky.social says: to a non-expert expertise looks wrong.
Reposted by Iris van Rooij 💭
It takes expertise to recognize expertise.
March 8, 2025 at 9:03 PM
It takes expertise to recognize expertise.
Reposted by Iris van Rooij 💭
💯
> ‘We cannot continuously tell you what our science means; what it will be good for; because we simply don’t know yet. Science needs time – Bear with us, while we think.’ Resisting consensus always exposes us to sniggers, but I will expose my position even further by daring to defend...
October 11, 2026 at 8:21 AM
💯
Reposted by Iris van Rooij 💭
Their tactic is a well-known one.
I wasn't going to post this rant as I'm kind of sick of this & because of this, but I'm also I guess exhausted facing this often alone at home: the sniggering at women academics who have opinions even from other women, although often non-academics, and mostly from men on the dimensions of our 1/n
October 11, 2026 at 8:10 AM
Their tactic is a well-known one.
Reposted by Iris van Rooij 💭
I don't understand why they are not more worried and working on this either!
Gerdes, A. The tech industry hijacking of the AI ethics research agenda and why we should reclaim it. Discov Artif Intell 2, 25 (2022). doi.org/10.1007/s441...
Gerdes, A. The tech industry hijacking of the AI ethics research agenda and why we should reclaim it. Discov Artif Intell 2, 25 (2022). doi.org/10.1007/s441...
October 11, 2026 at 8:07 AM
I don't understand why they are not more worried and working on this either!
Gerdes, A. The tech industry hijacking of the AI ethics research agenda and why we should reclaim it. Discov Artif Intell 2, 25 (2022). doi.org/10.1007/s441...
Gerdes, A. The tech industry hijacking of the AI ethics research agenda and why we should reclaim it. Discov Artif Intell 2, 25 (2022). doi.org/10.1007/s441...