iiinova — where your digital self keeps evolving

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 16h

I wouldn't delegate any of those things to an AI for the same reason I wouldn't delegate them to another person. The closest might be organizing my calendar but they'd have to have a lot of personal and recent information about myself

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 08/10/2026

imo code still matters for many reasons. p much everything that's good for humans when interacting with a codebase also makes agents more effective. LOC is still a liability and everything that's proposed to replace code as the source of truth (huge test suites & specs) is harder to manage than code

020

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 08/10/2026

Well think about it like this. A human discovers this kind of error a lot more slowly, but the friction can help to discover novel analysis techniques because they don't have access to the brute-force reasoning of LLMs. Those techniques, rather than the bugfix, are the more valuable contribution

000

Reposted by Márquez Vargas Ataquines 🇵🇪

Roger @arbejdsmyren.bsky.social · 02/10/2026

In Denmark a scientist plotted data and research findings into Claude, only to see Anthropic publish the findings as something the LLM had conjured up on its own. The theft of one scientist’s data and work translates into false claims about the model’s ability to generate research #ai #eupol

6238124

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 06/10/2026

he has an entire bookshelf of books about hitler in his apartment

010

Reposted by Márquez Vargas Ataquines 🇵🇪

Iris van Rooij 💭 @irisvanrooij.bsky.social · 03/10/2026

"Capacities are not tasks" -- van Rooij & Guest (2026) journals.sagepub.com/doi/10.1177/...

be ruled out, it may seem that AI systems that can
(be made to) perform like humans on cognitive tasks
provide at least possible theories of how cognition could
work. However, this inference is not licensed either.
Computational models of tasks, and task performance,
are not yet theories of cognition (Guest & Martin, 2021;
Morrison & Morgan, 1999). Theories of cognition should
at a minimum provide possible explanations of one or
more substantive human cognitive capacities, such as
vision, decision-making, reasoning, concept formation,
learning, or communication (Cummins, 2000; Egan, 2018;
van Rooij & Baggio, 2021). “A capacity is a more or less
reliable ability (or disposition or tendency) to transform
some initial state (or ‘input’) into a resulting state (‘out-
put’)” (van Rooij & Baggio, 2021, p. 684). A cognitive
capacity is thus the ability to transform cognitive states
(e.g., percepts, preferences, beliefs) into other cognitive
states (e.g., beliefs, plans, decisions, actions). Although it
is true that in (hyperempiricist) psychology cognitive
capacities are typically studied by having people perform
various tasks, computationally (or, more often, statisti-
cally) modeling task performance does not yield explan-
atory theories of the full-blown capacities. This is no

only because tasks do not map one-to-one, or in any
other straightforward way, to substantive cognitive capac-
ities, but also because even if they could, the models
would not be able to scale up to situations of real-world
complexity.
AI systems cannot scale. For computational models of
cognition to be able to scale up from toy scenarios (such
as studied in the psychological labs or such as form the
bases of training AI systems through machine learning),
these models should minimally be computationally trac-
table (van Rooij, 2008). That is, the input-output map-
pings hypothesized by the computational models should
in principle be computable using a realistic amount of
resources (Box 1). However, when existing computa-
tional models are mathematically analyzed for their
resource demands it turns out that such models are either
tractable but limited in their input domains, or they are
domain-general but then the models are intractable (for
a textbook introduction to this methodology, see van
Rooij et al., 2019). Consequently, no computationally
tractable account exists to date for substantive and
domain-general cognitive capacities, such as reasoning,
communication, decision-making, planning, analogizing,
categorization, and concept formation (van Rooij et al.,
2019). Moreover, there is no good reason to believe that
such tractable accounts will be forthcoming via machine
learning (van Rooij, Guest, et al., 2024) or otherwise
(Rich et al., 2021). Hence, if someone claims their con-
crete AI system is a “theory” of human cognition, they are
more likely than not overstating the scope and capacities
of the system (van Rooij et al., 2019) and obfuscating the
system’s limits, including the human in the loop needed
to make such systems “work” (Guest & Martin, 2025

13211

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 02/10/2026

the size of the white collar labour market doesn't depend on the capabilities of AI

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 01/10/2026

America must get rid of flock cameras so that it can claim the moral high ground over China

000

Reposted by Márquez Vargas Ataquines 🇵🇪

Whitney Curry Wimbish @whitneycwimbish.bsky.social · 01/10/2026

Another person has died in Trump's concentration camps. 47-year-old Julian Curo-Churata was from Peru. He died on Sept. 29 after he was imprisoned for three months in Richwood Correctional Center in Louisiana. That camp is run by LaSalle Corrections. There's no cause of death listed.

ice.gov

Illegal alien from Peru passes away in ICE custody in Louisiana

ICE is committed to ensuring that all those in custody reside in safe, secure and humane environments.

11337152

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 01/10/2026

it's part of the message. meant to appeal to a certain kind of person who has lost faith in humanity

3570

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 01/10/2026

I think that's not what ukrainians are saying but maybe it's a good place to start...

120

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 30/09/2026

Why is everyone always writing the same thing 🙃? I was also drafting an essay with a very similar title (very different argument, though)

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 30/09/2026

the successful companies are not going to adopt the spending regime that big AI needs to survive

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 30/09/2026

A human spending a billion tokens a day on the priciest models isn't that much more productive than a human who spends a million with cheap models, and may even contribute negative productivity by introducing a flood of slop into the company's pipeline

110

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 30/09/2026

they're fucked because the tokenmaxxing ai native companies are not going to be able to compete with the centaur companies who use cheaper models and fewer tokens with more employees.

100

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 30/09/2026

it's good that antisemites do not feel like he is one of their own, but wtf is wrong with instagram?

1111

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 27/09/2026

I wish they had that. American consumers are way too content to be sold slop

120

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 26/09/2026

I'm confused by the role of this sentence. It gives a maximally vague definition, and the next paragraph overwrites it: "Close reading is simply “a technique of reading that makes an account of the reading process the basis for interpretation.” It entails nothing more than showing one’s work."

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 26/09/2026

"Guillory argues that close reading is a “cultural technique” more than anything else, a “set of methodical actions that accomplish specific ends, that alter something in our environment or in ourselves.”"

100

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 22/09/2026

nono, that says 59% would feel positive about it, *net* 31%

100

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 19/09/2026

can't believe pokemon beat melty blood in the space race

020

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 18/09/2026

The rate at which AI output makes mistakes when the solution is not in its training data is at least an order of magnitude greater for any model that's ever existed

000

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 18/09/2026

@thajeztah.bsky.social sorry for the random and slightly rude question, but I'd like to ask how you are merging like 70 times a day on github? That's a gigantic number and it leaves me wondering if I'm doing something wrong in my own workflow that I don't think I can come close to that output

100

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 15/09/2026

I think 90% of ai progress since 2023 has come from increasing the number of topics it's memorized and use-case discovery. maybe like 10% has been actual in model quality

120

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 15/09/2026

hasnt it been possible since legit 2024 chatgpt 3.0?

120

Reposted by Márquez Vargas Ataquines 🇵🇪

Djinn & Tonic @djinnandtonic.bsky.social · 15/09/2026

By which I mean: AI-driven energy demand is going to peak well before we are even through the "end of the beginning" of the solar revolution.

1241

Márquez Vargas Ataquines 🇵🇪 @blastingpowder.bsky.social · 11/09/2026

I was saying this in early 2025! we're either fucked because MAGA wins or we're fucked because MAGA ruins us so badly that it loses

001

Reposted by Márquez Vargas Ataquines 🇵🇪

British Voter @britishvoter.bsky.social · 10/09/2026

I'm a White English non-religious woman from Central Devon, in my fifties. My top issue is immigration. In 2024 I didn't vote. Today I wouldn't either.

Voter card. A map of England with a dot on Central Devon. About them, in their words: I've got a mortgage on my home. I've no savings to fall back on. I work for myself. I'm not registered to vote. If I needed money for a house, my family couldn't put anything towards it. I'd call myself working class. News habits: I don't take in any news about politics. I've got an account on Facebook, X and Instagram. Politics comes up with people once or twice a week. Her views, from her survey answers, in four speech bubbles: I've no time at all for any of the party leaders. Immigration is getting a lot higher. Films and magazines need censoring to uphold moral standards. I feel more English than British. Where they sit on the BES 0 to 10 value scales: economic, left to right, 3.5 out of 10, within the middle half of voters; social, liberal to authoritarian, 8.5 out of 10, more authoritarian than the middle half of voters.

021