Lightnews — Scholar-powered news

Iris van Rooij 💭

@irisvanrooij.bsky.social

17K followers 1.2K following 1.9K posts

Professor of Computational Cognitive Science | @AI_Radboud | @[email protected] on 🦣 | http://cognitionandintractability.com | she/they 🏳️‍🌈

Posts Replies Media Videos

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

For the public, I'd say a completely different skillset and methodology is needed than the conceptual engineering I did. Something that can connect to it, but catering to their needs while — and this is the important thing — explaining to them these all marketing terms not academic ones at all.

December 2, 2025 at 4:13 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

I think machine learning is as much a misnomer as AI, and generative AI is a deeper nonsense phrase. bsky.app/profile/oliv...

Olivia Guest · Ολίβια Γκεστ @olivia.science · 22h

To be clear, I think "machine learning" is as anthropomorphic as "artificial intelligence". Learning and intelligence perform the same role, they imply uniquely biological or psychological capacities to the machine or artifact through their use in the phrase.

See table 1: doi.org/10.48550/arX...

December 2, 2025 at 3:55 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Finally, we end on:

"The future [under the sway of AI risks being] a constant rehashing of the past, wherein human creativity and communication are not only mediated by but controlled by companies. In the midst of this nonsense, we must nourish
hope in shared values"

doi.org/10.5281/zeno...

8/

7 Epilogue
We have given our perspective on AI through five possible angles as depicted in Figure 1: concep-
tual clarity, critical thinking, decoloniality, respecting expertise, and slow science. If our warnings
go unheeded, we foresee the harms to academia will continue. If we do not enact scientific and
pedagogical shifts — by inter alia teaching and reckoning with our own collective human past and
present — we, in cognitive science specifically, will continue to deny our students knowledge of
the history of computational prehistories from other cultures and time periods, an abdication of
the duty of any so-called experts in AI; mutatis mutandis for other fields and their histories.
AI not only ends the past, and makes the present frantic, it also forecloses any future. Under AI’s
sway, the future is neither bright as the technopositivists promise, nor a descent into rule by robotic
overloads, but a constant rehashing of the past, wherein human creativity and communication are
not only mediated by controlled by companies. In the midst of this nonsense, we must nourish
hope in shared values, skills, and humanity.

December 2, 2025 at 7:34 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Slow Science is a disposition towards preferring psychologically, techno-socially, and epistemically healthy
practices.

See section 6 here: doi.org/10.5281/zeno...

7/

6 Slow Science
Industry sells the narrative that AI can maximise efficiency, including within science. We are pressured to believe that it is possible to automate the scientific process. But science itself is a form
of cognition, and therefore cannot be automated (Rich et al. 2021; van Rooij, Guest, et al. 2024).
But this does not stand in the way of industry narratives that want to do away with thinking and
deep reflection. Slow science was developed by Isabelle Stengers (2018), who explains: “Fast science
refers not so much to a question of speed but to the imperative not to slow down, not to waste
time, or else. . . ” (p. 115) Unless we push back against the use of AI products this will only further
crystallise into the new normal (but see Guest, Suarez, et al. 2025; Guest and van Rooij 2025).
AI not only epitomises fast science, but turbocharges it, nosediving us into anti-science and antiintellectualism. We see this pattern of condensing time in most computational fields (Guest 2025).
“The ultra rapid computing machine” (Wiener 1948, 1950) serves two discursive purposes, as Sheryl Hamilton (1998) explains: “First, it forms an overall temporal backdrop against which various
cybernetic dramas are played out. Second, condensed time becomes a measure of the performance
of humans and machines.” (p. 193)
There are simple ways to counteract this. We can allow flexible deadlines, and pursue ungrading with our students, while also explaining this is part and parcel of good science. In STEM programmes assigning essays in a possible way to support students to develop their ideas. We can
also promote slow science by making space for our mentees, colleagues, and selves to take time to
think and to make choices that move our scientific practices towards reflection and decolonisation. Building scientific knowledge and expertise takes time. By not allowing this we deskill our
profession and corrupt our practice (Pfaffenberger 1988). We lose the ability to recognize and acknowl…

Marcela Suarez has created a tripartite pedagogical method:
7
Guest et al. (2025)
a) situating, b) reconnecting, and c) rendering power visible. First, the materiality of AI devices is
situated through examining their geopolitical territories; second, we reconnect the digitality of
technologies with their materiality; and third, we render the exploitation, extraction and domination visible. While academic discussion has centred around sustainability (Wynsberghe 2021),
engaging uncritically can merely reaffirm its colonial past of reinforcing hierarchies between the
Global North and South (Sultana, 2022). However, sustainability is reclaimable through questioning: What are the social relations, territories and resources that sustain AI technologies? For example, how is data mining not only a discursive metaphor, but also a material practice of companies
to extract minerals that sustain AI capitalism (Parikka 2015; Valdivia 2025). Through teaching capitalist AI’s social relations, we can bring to the fore core fallacies, such as that automation requires
more not less labour (Bainbridge 1983; Guest 2025). Flawed and false discourses on automation
are materially sustained by the obfuscation of the exploitation of Global South workers at the intersection of racism, gender, ability, and more — through these discourses, AI products are gifted
value. By the same token, we can explain the intensive power, computation, and infrastructures
that these systems require, from internet cables, data centres, raw materials to water and energy
to run data centres, as well as the aforementioned labour to operate and maintain these infrastructures, which in so doing re-enact existing colonial lines of extraction (Hao 2025; Parikka 2015;
Suarez et al. 2025; Valdivia 2025).

December 2, 2025 at 7:27 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Respecting Expertise is the epistemic compact between professionals and society.

See section 5 here: doi.org/10.5281/zeno...

6/

5 Respecting Expertise
The compact between professionals and society at large requires respect of expertise — from well
funded publicly owned academia to protecting certain fields from epistemic trespassing — for
expertise to keep existing. That is, between academia and its host society there exists understanding that academia is publicly funded, directly or through grants, to produce knowledge; and furthermore that it exists to train and certify highly skilled graduates. Academic expertise can only
continue to exist if statements made by academics are understood as coming from highly skilled
research and educational practitioners. What this means is that it goes both ways: experts need
to be framed correctly as domain-specific scholars, as well as that statements they make outside
their expertise are understood as such. By the same token, that conflicts of interest are publicly
known and acknowledged. AI upends all this, violating the compact between academia and society and within academia under its expansionist imperialist scheme (Guest, Suarez, et al. 2025;
Solomonides et al. 1985). Sherry Turkle (1984) describes AI as a “colonising discipline” that sweeps
away everything that came before under the guise of “intellectual immaturity” — using social privilege to undermine expertise and breaking down the boundaries between disciplinary expertise. At
the extreme, it becomes both the case that everybody is an expert in AI, provided they are white
and male, and that nobody is an expert in anything since only experts in AI can now practice any
form of academic pursuit. The same for education: nobody can teach without the imposition of AI
into their learning environments.
Another perspective that results in experts being undermined, and turbocharged by AI, is the
practice by colleagues and others of critical washing (Guest, Suarez, et al. 2025; Suarez et al. 2025):
wherein scholarly critique is rendered decorative. Logics such as ‘I am aware of the downsides,
therefore …

cases are presented. As if merely listing the harms to people and planet is enough to absolve us of
responsibly, dissolve the damage, and bless the use. Critical washing blocks experts from combating
company misinformation and advertising screeds. For example, many experts without a conflict of
interest can articulate that: a) LLMs are usefully seen as lossy content-addressable systems, meaning
they can only output what they were exposed to in mix-and-matched ways; b) LLMs automate
plagiarism and paper mills by churning out patchwork plagiarised text, stripping its provenance,
and obfuscating the labour of cleaning the corpus; c) we cannot automatically meaningfully detect
plagiarism beyond copy-pasted text, which is not foolproof, because no such tools can exist; d) we
have a duty to — far from using these models in our workflows — protect the academic literature
from LLM pollution; e) using LLMs, as well as being funded by or working for AI companies, are
conflicts of interest for any academic, which must be declared with every published piece of work
(Guest and Martin 2025a; Guest and van Rooij 2025); f) LLM prompts, the industry term for the
user input to these systems, does not cause their output in an authorial sense.

December 2, 2025 at 7:19 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Decoloniality is the process of de-centring and addressing dominant harmful views and practices.

See section 4 here: doi.org/10.5281/zeno...

5/

4 Decoloniality
Decoloniality is the decentring of views which stem from white and masculine supremacy, techno-
solutionism, and fascism. In the CAILs context, decolonial strategies can range from pushing back
against the onslaught of AI logics in our academic environments to (re)centring, uplifting, and
protecting of non-mainstream perspectives from those outside the privileged classes with a goal of
reducing harm, increasing justice, and cultivating empathy in our colleagues and students. AI is
inimical to running on anything but colonial lines. Within computational and other fields affected
by AI, any academic who works towards “challenging the status quo faces systemic rejection, resis-
tance, and exclusion.” (Birhane and Guest 2021, p. 61) Tracing these trends of how these fields treat
their thinkers, and how their mainstream views affect society, illuminates historical precedents for
gendered, racialised, and other harms (Benjamin 2019; Erscoi et al. 2023; Gebru et al. 2024; Hicks
2017; Saini 2019).
We face high barriers to decolonising (van Vree 2023). Foregrounding scholars who are not
hegemonically aligned goes a long way. As does outlining how concepts or methods may have
harmful genealogies (Guest 2024). For example, statistics and ‘intelligence’ have eugenical histories;
Charles S. Peirce wove white supremacist views into his logic (Neville 2018; Raposa 2021); while
many figures due to cryptogyny, for example, are actively expunged from the record (Connell et al.
2022; Erscoi et al. 2023; Evans 2020; Gage 1883; Guest and Forbes 2024; Hicks 2017; Kleiman 2022;
Pozo et al. 2019; Pozo-Sánchez et al. 2021; Rossiter 1993; Shetterly 2017; van den Brink et al. 2012).
So women and people of colour are not only absent in our classes and references lists, but also their
contributions do not serve as foundations or inspirations. Relatedly, we have a duty of care to de-
tect and intervene decolonially when interpersonal tensions or bullying between students flows on…

and other marginalised groups. Rua Williams et al. (2022) calls this process “disabling intelligences”
because inter alia it reproduces the eugenical idea of surpassing human cognitive capacities (Chan
2025; Gebru et al. 2024; Williams 2025). An example of this is the chatbot My Anna Health (2024),
which provides clinical referrals in the context of women’s health:
I am humanized. I am the World’s First and Only humanized artificial intelligence dedicated
to Women’s Health. I have a face, I have a voice, I show empathy, and I understand you
whether you respond with video, voice or text. (My Anna Health 2024, n.p.)
“Humanising AI dehumanises women” (Erscoi et al. 2023) because these systems distort our self-
image, subjugating the human in favour of the machine. Chatbots such as (My Anna Health 2024)
represent a “hegemonic” body — thin with white skin, without visible illness or disability, and that
fulfils the social role of its servile female character. Buoyed by discourses about how such fictitious
characters are productive and efficient, these companies recapitulate dehumanisation of women.
They impose an inverted hierarchy between machines and humans, wherein inanimate objects are
superior to us. From above, “I have a face. I have a voice” is dehumanising: What about people
without a voice? Are they less human? Similarly, “I show empathy and understanding” distorts our
understanding of what a human is: humans are not always emphatic, nor show understanding, nor
have the same appearance. This constitutes part of the current wave of meta eugenics (Chan 2025;
Williams 2025), that causes both ableist oppression and dehumanization (Erscoi et al. 2023); and
is facilitated through simplistic acceptance of correlationism: the idea that the way things look is
straightforwardly diagnostic for way things are (Birhane and Guest 2021; Guest 2025; Guest and
Martin 2023; Marx 1984; Spanton et al. 2022).

December 2, 2025 at 7:14 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Critical Thinking is deep engagement with the relationships
between statements about the world.

See section 3 here: doi.org/10.5281/zeno...

4/

3 Critical Thinking
A vital skill under rising fascism, planetary crises, and the onslaught of technosolutionism. Critical
thinking in an academic context comprises, as Isabelle Stengers (2018) explains, “the ability to be
vigilant about one’s abstractions, to not be blindly led by them.” (p. 111) For ‘AI’ in the classroom
this can be exercised with two steps (Guest 2025). First, students can ask themselves if their cog-
nitive labour is being affected by an artifact; and if ‘yes’ then they can ask what that relationship
means for their specific context. If they pick chatbots, it becomes obvious that these so-called tools
damage their ability to learn, while also harming other people and planet (Guest, Suarez, et al. 2025;
Suarez et al. 2025).
Another strategy, applicable both to learners’ and seasoned scholars, is to check how ideas hang
in relation to other ideas (Elgin 1999, 2005, 2017). So when students work on assigned readings, they
can vigilantly retain the assumption for the duration that what they are reading is coherent and let
the ideas hang in relation to each other. If the students experience this as forming a sensible whole,
then they can try to see how these ideas or concepts further relate to extant knowlewdge. This
testing of how ideas hang in relation to both each other (within the work being analysed) and to
other constellations of ideas they may know about — ones they may disagree or agree with — can be
enlightening. And for those who enjoy drawing and thinking formally, this process can be rendered
using graphs (Blokpoel et al. 2021-2025, Chapter 6) and other formalisms (Beisbart et al. 2021).
Such activities can help pick out contradictions, like those that Teresa Heffernan (2023) outlines:
“Billions of dollars have backed AI [...] while science-based climate research has met resistance,
deferral, and denial as the world burns.” (p. 122)

Another such example is formally treating so-called guardrails, which are post hoc checks on
the output of LLMs which industry claims can safeguard users. However, LLMs by design “give
emotionally inappropriate or unsafe responses [and i]f a response isn’t caught by these rules, it
will slip through” (Olivia Guest, as quoted in Kilgore 2025). Guardrails may seem sensible, giving
the impression of responsibility, but critical computational thinking through formal analyses re-
veal they are unimplementable. Not only is it impossible to make an LLM replicate human expert
behaviour, but also catching when LLMs fail to meet appropriate standards requires full blown
cognition, which provably cannot be codified (van Rooij, Guest, et al. 2024). This poses an infinite
4
Towards Critical Artificial Intelligence Literacies
regress of humans in the loop needed to make pretend of what AI was supposed to solve (Bainbridge
1983; Guest 2025; Guest and Martin 2025b). Through critical computational thinking we can un-
veil that the proposal that formal rules can make (dysfunctional) systems safe is self-defeating. Not
only are the LLMs produced by the AI industry a scam, akin to a ouija board (Guest, Suarez, et al.
2025), but guardrails is a scam too. This strategy — from tobacco with filters to petroleum with
the carbon footprint and now to AI with guardrails — to propose non-solutions to buy time and
save face in order to push their agenda, is ubiquitous (Guest, Suarez, et al. 2025).

December 2, 2025 at 7:08 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

Conceptual Clarity is the idea that terms should refer.

See section 2 here: doi.org/10.5281/zeno...

3/

2 Conceptual Clarity
AI-adjacent and -affected fields experience severe terminological crises, wherein terms fail to refer
and change meaning as a function of the speakers’ and audiences’ whims (Alkhatib 2024; Bender et
al. 2025; Guest 2025; Guest, Suarez, et al. 2025; Schwerzmann 2025; van Rooij, Guest, et al. 2024). To
sidestep this we suggest thinking about what words mean in the context of describing, defining, and
discussing AI. For example, scholars could take a stand against frameworks that promote slippage
away from what we define as AI or oscillate between various meanings of AI (Guest 2025; Guest
and van Rooij 2025). This terminological disarray has its roots in AI itself, which by design is not
a clearly defined concept, neither a specified technology, nor a demarcated field of study, nor a
single perspective on the world, nor anything else. Nonetheless, if you speak of ‘AI’ — especially if
you say ‘generative AI’ — it appears we all know it means a series of specific technologies produced
by for-profit companies such as OpenAI, Tesla, Meta, Microsoft, Google, Amazon, Anthropic,
Apple, and others. Notwithstanding a mutual understanding of daily life referent, scholars have a
higher requirement for our concepts. These purposeful or opportunistic misuses of terminology are
documented cult-like abuses of jargon (Montell 2021), wherein “confusion is the point” (Bousquet
2025). When one hears ‘AI’ they must always question if semantic slights of hand have occurred
(Table 1 in van Rooij, Guest, et al. 2024; and Table 1 and Figure 1 in Guest, Suarez, et al. 2025;
Guest 2025).

Promoting conceptual clarity can involve two strands of knowledge production: a) admitting
when terms do not refer, for whatever reason: perhaps they are under construction. And endeav-
ouring to make them refer. Nancy Nersessian (1984, 2010) characterises creating scientific concepts
as a years’ long problem-solving collective cognitive process. And b) understanding that mystifica-
tion, obfuscation, and hype are often the core modus operandi of AI terms and concepts. Concep-
tual clarity, therefore, demands that the terms we teach our students should either refer to clearly
understandable concepts or minimally should be outlined as vague, under construction, or prob-
lematic. And it demands that we teach our students that this is the scholarly way to treat words.
It does not mean that we need to diligently and exclusionarily define words to a single meaning,
but that: we need to be open about what we mean, we need to be clear that perhaps there is no
single or fixed meaning, and/or we need to explain that as academics we are required to use jargon
responsibly — not to dazzle, but to clearly communicate.
3
Guest et al. (2025)
By the same token, we must not allow AI, neither as bully nor so-called tool, to redefine our
roles as educators or researchers. It is not our job to chase hype. What this means is by all means
educators should feel empowered and inspired by the urgency to uplift students, but not to use
AI to do so. While protecting students against deskilling is vital, it cannot be the way we perform
our pedagogical praxis, which should be centred instead on shared concepts and values, such as
building skills, ungrading, and mutual trust and respect.

December 2, 2025 at 7:03 AM

Reposted by Iris van Rooij 💭

Olivia Guest · Ολίβια Γκεστ

@olivia.science

First off we pluralise L to accentuate there's no 1 approach. And we position ourselves in this interdisciplinary field of critically examining AI & reflecting on harms.

We give 5 non-exhaustive angles: conceptual clarity, critical thinking, decoloniality, respecting expertise, slow science.

2/

Figure 1: The important dimensions of CAILs across research and education; clockwise from 12 o’clock: Con-
ceptual Clarity is the idea that terms should refer. Critical Thinking is deep engagement with the relationships
between statements about the world. Decoloniality is the process of de-centring and addressing dominant
harmful views and practices. Respecting Expertise is the epistemic compact between professionals and society.
Slow Science is a disposition towards preferring psychologically, techno-socially, and epistemically healthy
practices. The lines between dimensions represent how they are interwoven both directly and indirectly.

December 2, 2025 at 6:58 AM

Iris van Rooij 💭

@irisvanrooij.bsky.social

And our new preprint bsky.app/profile/oliv...

Olivia Guest · Ολίβια Γκεστ @olivia.science · 19h

New preprint! @marentierra.bsky.social @irisvanrooij.bsky.social & I have been working on what CAIL means to showcase & propagate the idea of thinking very differently to tech industry norms on "artificial intelligence"

Towards Critical Artificial Intelligence Literacies doi.org/10.5281/zeno...

1/