Tessera
Folgen

Matt Collins

@mattkicks.bsky.social

Product Design & UI/UX. Retired captain of Commonwealth Skateboarding. Writer & illustrator of Matt's Endless Comic endlesscomic.bsky.social I hate driving. Portland, Oregon @matt_kicks most other places. mattkicks.com

643 Posts207 Follower266 Folgt
Luke O'Neil@lukeoneil47.bsky.social · 4w ago

As a US politician you can routinely call for, and use your considerable power to vote for and agitate for the deaths of tens of thousands of innocent people around the world and it doesn't count as killing them for some reason. A person who does that is not a killer for some reason.

Matt Collins@mattkicks.bsky.social · last mo.

Larry Ellison losing even 1/3 of his ass in an AI bubble pop will be so funny. If the Paramount/CNN acquisition collapses...

Jesse Felder@jessefelder.com · last mo.

”Some of our customers [OpenAI] may be highly leveraged and subject to their own operating and regulatory risks and even if our credit review and analysis mechanisms work properly we may experience risks of non-payment in our dealings with such parties.” www.bloomberg.com/news/newslet...

Olivia Guest · Ολίβια Γκεστ@olivia.science · 7mo ago

From the moment we know a technology steals/obfuscates people's labour to perform task X, it matters 0 how effective the people's labour you steal to do X is. They will likely get better at X over time and you will get better at stealing. Nothing of this justifies the theft. bsky.app/profile/oliv...

In this paper, I turn this frame on its head (cf. Andersen et al., 2023; Bannon, 2011; Bishop, 2021; Nick Dyer-Witheford, 2019; Rogers, 2022; Ryan, 2024; Schmager et al., 2025; Shneiderman, 2022; Steinhoff, 2021; Walton & Nayak, 2021). Using a unique (re)definition of AI that releases us from a correlationist1 grip, we then examine three triplets as case studies of techno-social relationships between cognition and artifacts. To presage the coming analyses, herein AI is any techno-social relationship that outsources to machines or algorithms some part, however small, of human cognitive labour (cf. van Rooij et al., 2024). I demonstrate that AI is human-centric, not because it behaves like or is designed to be like humans, but because it requires a ghost in the machine, often literally an obfuscated human-in-the-loop to properly function (also see Guest & Martin, 2025) because AI is humans albeit in fetishised (Braune, 2020; Morris, 2017; Mota & Cosentino Filho, 2024), obfuscated forms (e.g. Erscoi et al., 2023). That is, AI “is in reality produced by relations among people [even though it] appears before us in a fantastic form as relations among things” (Pfaffenberger, 1988, p. 250). AI’s “technological veil” hides human cognitive labour (Mota & Cosentino Filho, 2024). I bring this anthropological, sociotechnical and broadly computational cognitive scientific angle to understanding AI, that is unlike “the skills and techniques of behavioural research, previously applied to humans and other animals[, which claim to help us] better understand the ‘psychology’ of complex black-box ANNs” (Wills, 2025, n.p. cf. Raley and Rhee, 2023). In fact behavioural probing of such systems using such experimental techniques assumes their psychological standing to be equal to, or comparable to, biological organisms: it begs the question (see for relevant analyses: Forbes & Guest, 2025; Guest & Martin, 2023, 2024; Raji et al., 2022).There are no Rube Goldberg-like perpetual motion-like human-in-the-loop-free machines. There never will be. We might crave such a machine, but the only entities we know that are self-sustaining, autopoietic, is everything but machines. The only person who can decide to turn a computer off and then on again is exactly that: a person (Polo, 2024). To effectively avoid correlationism, obfuscation of cognition in AI, we must reject the current mainstream view: no amount of high scores on benchmarks, or any other correlationary evidence, can ever pile up high enough to graduate to a causal claim. To de-fetishise AI, we must accept that AI is any relationship between technology, tools, models, machines and humans where it appears as if some cognitive labour is offloaded onto such artifacts, and furthermore we must accept that such a relationship requires methodical teasing apart to obviate the centrality of the human (Bernardi, 2024; Braune, 2020; Morris, 2017; Mota & Cosentino Filho, 2024; Pfaffenberger, 1988). We cannot rid the machine from its ghost. But we can rid the concept of human from ghostliness — the human “need not be degraded to a machine by being denied to be a ghost in a machine.” (Ryle, 1949, p. 301)
Olivia Guest · Ολίβια Γκεστ@olivia.science · last yr.

I split AI into 3 non-mutually exclusive types (see Table 1 above): displacement (harmful), enhancement (beneficial), and/or replacement (neutral) of human cognitive labour. More later possibly, but see Tables 2 to 4 (attached or here: arxiv.org/pdf/2507.19960) for the worked through examples. 2/n

table 2 from https://arxiv.org/pdf/2507.19960
Hart Cunningham@hart28.bsky.social · 2mo ago

Michigan Governor caught on hot mic with Oracle CEO Clay Magouyrk, discussing citizen opposition to AI data centers, bragging how she ignores their concerns: “We’re used to people saying 'f*ck no,' & doing it anyway.”🇺🇸